NonStop TS/MP Pathsend

and Server Programming

Manual

Abstract
This manual describes how to write two types of programs as part of a Pathway
application: requester programs that use the Pathsend application program interface
(API) and server programs that service requests from all types of Pathway requesters.
Product Version
NonStop TS/MP D44
Supported Releases
This manual supports D44.00 and all subsequent D4x releases and G03.00 and all
subsequent G-series releases until otherwise indicated in a new edition.

Part Number Published Release ID

132500 July 1997 D44.00
Document History
Part Number Product Version Published
110074 NonStop TS/MP D31 July 1995
123813 NonStop TS/MP D31 December 1995
132500 NonStop TS/MP D44 July 1997
New editions incorporate any updates issued since the previous edition.
A plus sign (+) after a release ID indicates that this manual describes function added to the base release, either by an
interim product modification (IPM) or by a new product version on a .99 site update tape (SUT).

Ordering Information
For manual ordering information: domestic U.S. customers, call 1-800-243-6886; international customers, contact
your local sales representative.
Document Disclaimer
Information contained in a manual is subject to change without notice. Please check with your authorized Tandem
representative to make sure you have the most recent information.
Export Statement
Export of the information contained in this manual may require authorization from the U.S. Department of
Commerce.
Examples
Examples and sample programs are for illustration only and may not be suited for your particular purpose. Tandem
does not warrant, guarantee, or make any representations regarding the use or the results of the use of any examples
or sample programs in any documentation. You should verify the applicability of any example or sample program
before placing the software into productive use.
U.S. Government Customers
FOR U.S. GOVERNMENT CUSTOMERS REGARDING THIS DOCUMENTATION AND THE ASSOCIATED
SOFTWARE:
These notices shall be marked on any reproduction of this data, in whole or in part.
NOTICE: Notwithstanding any other lease or license that may pertain to, or accompany the delivery of, this
computer software, the rights of the Government regarding its use, reproduction and disclosure are as set forth in
Section 52.227-19 of the FARS Computer Software—Restricted Rights clause.
RESTRICTED RIGHTS NOTICE: Use, duplication, or disclosure by the Government is subject to the
restrictions as set forth in subparagraph (c)(1)(ii) of the Rights in Technical Data and Computer Software clause at
DFARS 52.227-7013.
RESTRICTED RIGHTS LEGEND: Use, duplication or disclosure by the Government is subject to restrictions
as set forth in paragraphþ(b)(3)(B) of the rights in Technical Data and Computer Software clause in
DAR 7-104.9(a). This computer software is submitted with “restricted rights.” Use, duplication or disclosure is
subject to the restrictions as set forth in NASA FARþSUP 18-52þ227-79 (Aprilþ1985) “Commercial Computer
Software—Restricted Rights (Aprilþ1985).” If the contract contains the Clause at 18-52þ227-74 “Rights in Data
General” then the “Alternate III” clause applies.
U.S. Government Users Restricted Rights — Use, duplication or disclosure restricted by GSA ADP Schedule
Contract.
Unpublished — All rights reserved under the Copyright Laws of the United States.
 New and Changed Information
This is the third edition of the NonStop TS/MP Pathsend and Server Programming
Manual.
Writers of Pathsend requesters and all Pathway servers should read this manual. Writers
of SCREEN COBOL requesters should read the Pathway/TS TCP and Terminal
Programming Guide.
This third edition includes changes to reflect product changes, and also additional
enhancements. Substantive changes (changes that are not simply editorial) are marked
by change bars in the right-hand margin of the page.

Product Changes
This manual documents the following changes to NonStop TS/MP and related products:
• New Pathsend procedure calls have been added to support context-sensitive
communication (dialogs) between requesters and servers.
Section 3, Writing Pathsend Requesters, now describes how to use these procedure
calls in a Pathsend requester. Section 5, Pathsend Procedure Call Reference,
provides detailed syntax for the procedure calls. Section 6, Pathsend Errors,
describes new error messages related to the calls.
• A new TUXEDO to Pathway translation server is now available as part of Release 2
of the NonStop TUXEDO system. This translation server allows TUXEDO
requesters (that is, TUXEDO clients and TUXEDO servers acting as clients) to
access Pathway servers.
Other Transaction Processing Environments on page 1-13 and Writing Pathway
Servers That Interoperate With TUXEDO Requesters on page 4-17 now mention the
availability of the TUXEDO to Pathway translation server. Details on the use of this
translation server are provided in the NonStop TUXEDO System Pathway
Translation Servers Manual.
• Changes are being made to the Remote Server Call (RSC) product, and some of the
protocols formerly listed in Section 2 of this manual may no longer be supported.
Therefore, all mention of support for specific platforms and protocols under Clients
Using RSC and POET on page 2-15 has been removed and replaced by a more
general statement. For detailed information about the platforms and protocols
supported by the RSC product, refer to the Remote Server Call (RSC) Programming
Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

iii
New and Changed Information Enhancements to the Manual

Enhancements to the Manual

The following enhancements have been made to the material in this manual:
• A new subsection, Consideration for Servers Used With Remote Server Call (RSC)
Clients on page 4-4, has been added to mention the optional server reply code for
servers used with Remote Server Call (RSC) clients. For details, refer to the Remote
Server Call (RSC) Programming Manual.
• Corrections have been made to the discussion of Server-Class Send Timeout on
page 5-27.
• Additional cause information has been added to the description of Pathsend error
902 on page 6-3.
• Several minor technical corrections and editorial changes have been made.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

iv
 Contents
New and Changed Information iii
About This Manual xi
Notation Conventions xvii
1. Introduction to Pathway Application Programming
Which Sections Do You Need? 1-1
Advantages of the Pathway Environment 1-3
Ease of Development 1-3
Manageability 1-4
Data Integrity 1-4
Fault Tolerance 1-5
Other Tandem Fundamentals 1-6
Pathway Applications 1-7
Servers and Server Classes 1-8
Requesters 1-9
The Pathsend Environment 1-10
Pathsend Processes 1-10
LINKMON Processes 1-11
Client/Server Capabilities 1-12
Other Transaction Processing Environments 1-13
Development Tools and Utilities 1-14
Programming Languages and Related Tools 1-14
The Inspect Symbolic Debugger 1-14
The Pathmaker Application Generator 1-14
Client/Server Development Tools 1-15
Transaction Processing Scenario 1-15
2. Designing Your Application
Designing Transactions 2-1
Analyzing Data Flow 2-2
Identifying Transaction Components 2-4
Protecting Transactions 2-6
Designing the Database 2-9
Logical Design 2-9
Physical Design 2-10
Database Managers 2-10
Remote Duplicate Database Facility (RDF) 2-11

NonStop TS/MP Pathsend and Server Programming Manual– 132500

v
Contents 2. Designing Your Application (continued)

2. Designing Your Application (continued)

Designing Requester Programs 2-11
SCREEN COBOL Requesters 2-12
IDS Requesters 2-12
Pathsend Requesters 2-13
Clients Using RSC and POET 2-15
Requesters Using GDSX 2-16
Dividing Function Between Requester and Server 2-19
Designing Server Programs 2-19
Design Considerations 2-20
Server Program Structure 2-25
Designing Applications for Batch Processing 2-27
3. Writing Pathsend Requesters
The Pathsend Procedure Calls 3-1
Interprocess Communication in the Pathsend Environment 3-2
Basic Pathsend Programming 3-3
Programming for Failure Recovery 3-3
Security Issues 3-6
Avoiding Coded PATHMON Names 3-7
Context-Sensitive Pathsend Programming 3-8
Using Context-Sensitive Requesters With Context-Free Servers 3-8
Resource Utilization 3-8
Programming for Failure Recovery 3-9
Cancellation of Server-Class Send Operations 3-10
Writing Requesters That Interoperate With NonStop TUXEDO Servers 3-11
4. Writing Pathway Servers
Basic Pathway Server Programming 4-1
Servers Shared by Different Types of Requesters 4-1
Guardian Servers and Pathway Servers 4-2
Server Stop Protocol 4-2
Handling of Messages from $RECEIVE 4-2
Pathsend Requester Failures 4-2
LINKMON Process Failures 4-3
Linkage Space Considerations 4-3
Considerations for Servers Used With SCREEN COBOL Requesters 4-3
Consideration for Servers Used With Remote Server Call (RSC) Clients 4-4
Nested Servers 4-4
Using Context-Free Servers With Context-Sensitive Requesters 4-4

NonStop TS/MP Pathsend and Server Programming Manual– 132500

vi
Contents 4. Writing Pathway Servers (continued)

4. Writing Pathway Servers (continued)

Considerations for Servers That Use the TMF Subsystem 4-5
Recommended Structure for Applications 4-5
Writing a Server to Use the TMF Subsystem 4-6
Using Audited and Nonaudited Files 4-7
Locking Records 4-8
Grouping Transaction Operations 4-8
Servers as Process Pairs 4-10
Transaction Deadlocks 4-10
Considerations for Debugging Pathway Servers 4-11
LINKMON Process and TCP Timeouts 4-11
PATHMON Process Timeouts 4-12
Server Timeouts 4-12
Avoiding Timeout Errors 4-12
Writing Context-Sensitive Servers 4-13
Functions of a Context-Sensitive Server 4-13
Detecting a Newly Established Dialog 4-14
Receiving, Servicing, and Replying to Messages in a Dialog 4-14
Correlating Messages With a Dialog 4-16
Continuing a Dialog 4-16
Aborting a Dialog 4-16
Terminating a Dialog 4-16
Detecting an Aborted Dialog 4-16
Writing Pathway Servers That Interoperate With TUXEDO Requesters 4-17
5. Pathsend Procedure Call Reference
Calls From C or C++ 5-2
Calls From COBOL85 5-3
Calls From Pascal 5-4
Calls From TAL or pTAL 5-5
SERVERCLASS_DIALOG_ABORT_ Procedure 5-6
Syntax 5-6
Considerations 5-6
SERVERCLASS_DIALOG_BEGIN_ Procedure 5-7
Syntax 5-7
Considerations 5-11
SERVERCLASS_DIALOG_END_ Procedure 5-12
Syntax 5-12
Considerations 5-12

NonStop TS/MP Pathsend and Server Programming Manual– 132500

vii
Contents 5. Pathsend Procedure Call Reference (continued)

5. Pathsend Procedure Call Reference (continued)

SERVERCLASS_DIALOG_SEND_ Procedure 5-13
Syntax 5-13
Considerations 5-16
SERVERCLASS_SEND_ Procedure 5-17
Syntax 5-17
Considerations 5-20
SERVERCLASS_SEND_INFO_ Procedure 5-21
Syntax 5-21
Considerations 5-22
Usage Considerations for Pathsend Procedures 5-23
Condition Code 5-23
Waited I/O 5-23
Nowait I/O 5-23
Calls Within a TMF Transaction 5-24
Server-Class Send Operation Number 5-24
Timeout Considerations for Pathsend Programming 5-27
6. Pathsend Errors
Types of Errors Returned by the Pathsend Procedures 6-1
Descriptions of Pathsend Errors 6-1
A. NonStop TS/MP Limits for Pathsend Requesters
B. Examples
Pathsend Requester Example B-1
Nested Server Example B-53
Glossary
Index

NonStop TS/MP Pathsend and Server Programming Manual– 132500

viii
Contents Examples

Examples
Example 2-1. Sample Pathsend Requester Program Structure 2-15
Example 2-2. COBOL85 Server Program Example 2-25
Example B-1. Context-Free Pathsend Requester Program B-2
Example B-2. Context-Free Server Program B-53
Figures
Figure i. Related Documentation xiii
Figure 1-1. Pathsend Interprocess Communication 1-11
Figure 1-2. Example Application Using a Pathsend Requester 1-16
Figure 2-1. Data Flow for a Business Task 2-3
Figure 2-2. Relationships Between Transaction Functions 2-5
Figure 2-3. Pathway Application Programming for the TMF Subsystem 2-7
Figure 2-4. GDSX as a Front-End Process 2-18
Figure 2-5. GDSX as a Back-End Process 2-24
Tables
Table 1-1. Task and Manual Correspondences 1-2
Table 2-1. Considerations for Requester Programs 2-11
Table 4-1. Meaning of Error Codes Returned by Context-Sensitive Server in
Reply 4-15
Table 5-1. Summary of Pathsend Procedure Calls 5-1
Table A-1. Limits for Pathsend Requesters A-1

NonStop TS/MP Pathsend and Server Programming Manual– 132500

ix
Contents

NonStop TS/MP Pathsend and Server Programming Manual– 132500

x
 About This Manual
This section describes the purpose and the contents of this manual and of other manuals
closely related to this manual.
This manual is one of a set of manuals that describe the NonStop Transaction
Services/MP (NonStop TS/MP) and Pathway/Transaction Services (Pathway/TS)
products. The contents of these products are as follows:
• NonStop TS/MP: This product consists of the PATHMON process, the LINKMON
process, the PATHCOM process and interface, and the Pathsend procedure calls.
• Pathway/TS: This product consists of the terminal control process (TCP), the
SCREEN COBOL compiler and run-time environment, and the SCREEN COBOL
Utility Program (SCUP).
Together with the NonStop Transaction Manager/MP (NonStop TM/MP) product, the
NonStop TS/MP product forms the foundation for Tandem’s open transaction processing
and client/server products on NonStop Himalaya systems. Such products include the
NonStop TUXEDO system, the Remote Server Call (RSC) product, and the Pathway
Open Environment Toolkit (POET).
The Pathway/TS product supports requester programs that run in the Guardian
environment and communicate with terminals and intelligent devices. It requires the
services of the NonStop TS/MP product.

Purpose of This Manual

This manual is a combined reference manual and programming guide. It contains
reference information about the Pathsend procedure calls, and it also contains
information about how to design and code applications using these calls.
In addition, this manual contains information about how to design and code Pathway
servers to be used with all types of requesters and clients.

Who Should Read This Manual

This manual is intended for programmers writing Pathway applications that include
Pathsend requesters, Pathway servers, or both. Readers should be experienced
programmers familiar with the Guardian environment and the Tandem programming
languages they are using.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xi
About This Manual What Is in This Manual

What Is in This Manual

This manual contains application design and programming information for writers of
Pathsend requesters and Pathway servers. This information includes syntax, usage, and
error-handling information for the Pathsend procedure calls. The manual has the
following structure:
• Section 1, Introduction to Pathway Application Programming, contains an overview
of Pathway application programming, with emphasis on Pathsend requesters and
Pathway servers.
• Section 2, Designing Your Application, provides information about Pathway
application design, with emphasis on Pathsend requesters and Pathway servers.
• Section 3, Writing Pathsend Requesters, describes how to write requesters that use
the Pathsend procedure calls.
• Section 4, Writing Pathway Servers, describes how to write Pathway servers.
• Section 5, Pathsend Procedure Call Reference, provides the syntax, parameter
descriptions, and coding considerations for the Pathsend procedure calls.
• Section 6, Pathsend Errors, provides cause, effect, and recovery information for all
Pathsend error codes.
• Appendix A, NonStop TS/MP Limits for Pathsend Requesters, summarizes the
limits that apply to the Pathsend programming environment.
• Appendix B, Examples, provides source code for an example Pathsend requester and
an example Pathway server.
• A standard Glossary for all NonStop TS/MP and Pathway/TS manuals defines all
terms relevant to these products.

Related Documentation
This manual is one in a set of Tandem manuals for the NonStop TS/MP and Pathway/TS
products. Figure i, Related Documentation, shows the manuals that are most closely
related to this manual.
The following paragraphs describe each of the supporting manuals shown in Figure 1.
• Pathway/TS TCP and Terminal Programming Guide. This guide describes how to
write requester programs using SCREEN COBOL.
• Guardian Programmer’s Guide. This guide provides information about
programming in the Guardian environment, including use of the Guardian procedure
calls. It is useful to programmers who are writing Pathway servers, especially if
they are writing them in C, C++, the Transaction Application Language (TAL), or
the Portable Transaction Application Language (pTAL).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xii
About This Manual Related Documentation

Figure i. Related Documentation

Further information
Pathway/TS Operator
about Pathway
TCP and Messages
application
Terminal Manual
programming
Programming Information about
Guide problem
management Guardian
Guardian Procedure
Programmer's Errors and
Guide Messages
NonStop Manual
TS/MP
Pathsend
and Server
Programming
Manual

Introduction
Information Information to NonStop
about about related Transaction
NonStop
managing products Processing
TS/MP
a PATHMON
System
environment
Management NonStop
Manual TM/MP
Application
Programmer's
Guide
NonStop
TUXEDO Sys
App Dev
Guide

NonStop
TUXEDO Sys
Pathway
Transl Servers
Manual

CDT022

For information about informational, warning, and error messages, refer to the following
manuals:
• Guardian Procedure Errors and Messages Manual. This manual describes the
Guardian messages for Tandem systems that use the Tandem NonStop Kernel. The
manual covers various types of error codes and error lists associated with Guardian
procedure calls and also the interprocess messages sent to application programs by
the operating system and the command interpreter.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xiii
About This Manual Related Documentation

• Operator Messages Manual. This manual describes system messages. For each
message the manual provides an explanation of the cause, a discussion of the effect
on the system, and suggestions for corrective action. The “PATHWAY Messages”
section describes the operator messages generated by the PATHMON environment.
For information about managing the PATHMON environment in which your Pathway
applications run, you might want to read the following manual:
• NonStop TS/MP System Management Manual. This manual describes how to start,
configure, and manage a PATHMON environment. This manual also includes
information about monitoring and adjusting your PATHMON environment to
optimize performance, suggestions for diagnosing and fixing problems, and
manageability guidelines such as how to start objects in parallel to improve
performance. It also describes the commands for configuring, starting, and
managing a PATHMON environment.
For information about related Tandem products, see the following publications:
• Introduction to NonStop Transaction Processing. This manual describes the
architecture, components, and benefits of Tandem transaction processing products,
including NonStop TS/MP, Pathway/TS, and related products such as the NonStop
TUXEDO system.
• NonStop TM/MP Application Programmer’s Guide. This guide provides
information about programming for the Transaction Management Facility (TMF)
subsystem, including use of the TMF procedure calls. It is useful to programmers
who are writing Pathsend requesters and Pathway servers, especially if they are
writing them in C, C++, the Transaction Application Language (TAL), or the
Portable Transaction Application Language (pTAL).
• NonStop TUXEDO System Application Development Guide. This guide provides
information about writing NonStop TUXEDO applications. Pathsend requesters can
interoperate with NonStop TUXEDO applications either directly by using the
NonStop TUXEDO Application Transaction Monitor Interface (ATMI) functions, or
indirectly by using the Pathway translation server for the NonStop TUXEDO
system.
• NonStop TUXEDO System Pathway Translation Servers Manual. This manual
provides information about writing Pathsend and SCREEN COBOL requesters that
interoperate with NonStop TUXEDO servers by using the Pathway to TUXEDO
translation server, and information about writing Pathway servers that interoperate
with TUXEDO requesters (clients or servers acting as clients) by using the
TUXEDO to Pathway translation server. It also provides configuration information
for the two translation servers.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xiv
About This Manual Other Manuals in the Manual Set

Other Manuals in the Manual Set

In addition to the NonStop TS/MP and Pathway/TS manuals shown in Figure i, the
manual set for these products includes the following manuals:
• Pathway/TS System Management Manual. This manual describes how to start,
configure, and manage Pathway/TS objects (TCPs, terminal objects, SCREEN
COBOL programs, and tell messages) in a PATHMON environment. It also
describes the commands for configuring, starting, and managing Pathway/TS
objects.
• NonStop TS/MP Management Programming Manual. This manual describes the
management programming interface to the Pathway subsystem, including
programmatic commands and related objects, event messages, and error lists.
• Pathway/TS Management Programming Manual. This manual describes the error
lists and event messages issued by the management programming interface to the
Pathway subsystem.

Other Manuals of Interest

Other manuals that might be of interest to readers of this manual include the following:
• Availability Guide for Application Design. This manual describes the features of
Tandem NonStop systems that support the availability of applications. This manual
includes a section about application availability in the Pathway transaction
processing environment.
• Guardian Programmer’s Guide. This guide provides information about
programming in the Guardian environment, including use of the Guardian procedure
calls. It is useful to programmers who are writing Pathway servers, especially if
they are writing them in C, C++, the Transaction Application Language (TAL), or
the Portable Transaction Application Language (pTAL).

Your Comments Invited

After using this manual, please take a moment to send us your comments. You can do
this by returning a Reader Comment Card or by sending an Internet mail message.
A Reader Comment Card is located at the back of printed manuals and as a separate file
on the Tandem CD Read disc. You can either FAX or mail the card to us. The FAX
number and mailing address are provided on the card.
Also provided on the Reader Comment Card is an Internet mail address. When you
send an Internet mail message to us, we immediately acknowledge receipt of your
message. A detailed response to your message is sent as soon as possible. Be sure to
include your name, company name, address, and phone number in your message. If
your comments are specific to a particular manual, also include the part number and title
of the manual.
Many of the improvements you see in Tandem manuals are a result of suggestions from
our customers. Please take this opportunity to help us improve future manuals.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xv
About This Manual Your Comments Invited

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xvi
 Notation Conventions
General Syntax Notation
The following list summarizes the notation conventions for syntax presentation in this
manual.

UPPERCASE LETTERS. Uppercase letters indicate keywords and reserved words; enter
these items exactly as shown. Items not enclosed in brackets are required. For example:
MAXATTACH

lowercase italic letters. Lowercase italic letters indicate variable items that you supply.
Items not enclosed in brackets are required. For example:
file-name

[ ] Brackets. Brackets enclose optional syntax items. For example:

TERM [\system-name.]$terminal-name
INT[ERRUPTS]
A group of items enclosed in brackets is a list from which you can choose one item or
none. The items in the list may be arranged either vertically, with aligned brackets on
each side of the list, or horizontally, enclosed in a pair of brackets and separated by
vertical lines. For example:
LIGHTS [ ON ]
[ OFF ]
[ SMOOTH [ num ] ]
K [ X | D ] address-1

{ } Braces. A group of items enclosed in braces is a list from which you are required to
choose one item. The items in the list may be arranged either vertically, with aligned
braces on each side of the list, or horizontally, enclosed in a pair of braces and separated
by vertical lines. For example:
LISTOPENS PROCESS { $appl-mgr-name }
{ $process-name }
ALLOWSU { ON | OFF }

| Vertical Line. A vertical line separates alternatives in a horizontal list that is enclosed in
brackets or braces. For example:
INSPECT { OFF | ON | SAVEABEND }

… Ellipsis. An ellipsis immediately following a pair of brackets or braces indicates that you
can repeat the enclosed sequence of syntax items any number of times. For example:
M address-1 [ , new-value ]...
[ - ] {0|1|2|3|4|5|6|7|8|9}...

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xvii
Notation Conventions General Syntax Notation

An ellipsis immediately following a single syntax item indicates that you can repeat that
syntax item any number of times. For example:
"s-char..."

Punctuation. Parentheses, commas, semicolons, and other symbols not previously described
must be entered as shown. For example:
error := NEXTFILENAME ( file-name ) ;
LISTOPENS SU $process-name.#su-name
Quotation marks around a symbol such as a bracket or brace indicate the symbol is a
required character that you must enter as shown. For example:
"[" repetition-constant-list "]"

Item Spacing. Spaces shown between items are required unless one of the items is a
punctuation symbol such as a parenthesis or a comma. For example:
CALL STEPMOM ( process-id ) ;
If there is no space between two items, spaces are not permitted. In the following
example, there are no spaces permitted between the period and any other items:
$process-name.#su-name

Line Spacing. If the syntax of a command is too long to fit on a single line, each continuation
line is indented three spaces and is separated from the preceding line by a blank line.
This spacing distinguishes items in a continuation line from items in a vertical list of
selections. For example:
ALTER [ / OUT file-spec / ] CONTROLLER

[ , attribute-spec ]...

!i and !o. In procedure calls, the !i notation follows an input parameter (one that passes data
to the called procedure); the !o notation follows an output parameter (one that returns
data to the calling program). For example:
CALL CHECKRESIZESEGMENT ( segment-id !i
, error ) ; !o

!i,o. In procedure calls, the !i,o notation follows an input/output parameter (one that both
passes data to the called procedure and returns data to the calling program). For
example:
error := COMPRESSEDIT ( filenum ) ; !i,o

!i:i. In procedure calls, the !i:i notation follows an input string parameter that has a
corresponding parameter specifying the length of the string in bytes. For example:
error := FILENAME_COMPARE_ ( filename1:length !i:i
, filename2:length ) ; !i:i

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xviii
Notation Conventions Notation for Messages

!o:i. In procedure calls, the !o:i notation follows an output buffer parameter that has a
corresponding input parameter specifying the maximum length of the output buffer in
bytes. For example:
error := FILE_GETINFO_ ( filenum !i
, [ filename:maxlen ] ) ; !o:i

Notation for Messages

The following list summarizes the notation conventions for the presentation of displayed
messages in this manual.

Nonitalic text. Nonitalic letters, numbers, and punctuation indicate text that is displayed or
returned exactly as shown. For example:
Backup Up.

lowercase italic letters. Lowercase italic letters indicate variable items whose values are
displayed or returned. For example:
p-register
process-name

[ ] Brackets. Brackets enclose items that are sometimes, but not always, displayed. For
example:
Event number = number [ Subject = first-subject-value ]
A group of items enclosed in brackets is a list of all possible items that can be displayed,
of which one or none might actually be displayed. The items in the list might be
arranged either vertically, with aligned brackets on each side of the list, or horizontally,
enclosed in a pair of brackets and separated by vertical lines. For example:
LDEV ldev [ CU %ccu | CU %... ] UP [ (cpu,chan,%ctlr,%unit) ]

{ } Braces. A group of items enclosed in braces is a list of all possible items that can be
displayed, of which one is actually displayed. The items in the list might be arranged
either vertically, with aligned braces on each side of the list, or horizontally, enclosed in
a pair of braces and separated by vertical lines. For example:
LBU { X | Y } POWER FAIL
process-name State changed from old-objstate to objstate
{ Operator Request. }
{ Unknown. }

| Vertical Line. A vertical line separates alternatives in a horizontal list that is enclosed in
brackets or braces. For example:
Transfer status: { OK | Failed }

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xix
Notation Conventions Change Bar Notation

% Percent Sign. A percent sign precedes a number that is not in decimal notation. The
%þnotation precedes an octal number. The %Bþnotation precedes a binary number.
The %Hþnotation precedes a hexadecimal number. For example:
%005400
P=%p-register E=%e-register

Change Bar Notation

Change bars are used to indicate substantive differences between this edition of the
manual and the preceding edition. Change bars are vertical rules placed in the right
margin of changed portions of text, figures, tables, examples, and so on. Change bars
highlight new or revised information. For example:
The message types specified in the REPORT clause are different in the COBOL85
environment and the Common Run-Time Environment (CRE).
The CRE has many new message types and some new message type codes for old
message types. In the CRE, the message type SYSTEM includes all messages
except LOGICAL-CLOSE and LOGICAL-OPEN.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

xx
 1 Introduction to Pathway
Application Programming
This section introduces Pathway transaction processing applications, which you write
and run with the assistance of the NonStop Transaction Services/MP (NonStop TS/MP)
and Pathway/Transaction Services (Pathway/TS) software.
This section discusses the following topics:
• Advantages of the Pathway environment
• Pathway applications, including requesters and servers
• The Pathsend environment
• Client/server capabilities
• Other supported transaction processing environments
• Development tools and utilities
• A sample transaction processing scenario
This section does not describe the components of the NonStop TS/MP software in detail.
You can find more detailed information about these components in the introductory
sections of the NonStop TS/MP System Management Manual. The Introduction to
NonStop Transaction Processing manual summarizes the main features and capabilities
of the NonStop TS/MP software.

Which Sections Do You Need?

The remaining sections of this manual describe how to write two types of programs as
part of a Pathway application: requester programs that use the Pathsend application
program interface (API) and server programs that service requests from all types of
Pathway requesters. The sections are organized into logical groups of information for
easy reference. Depending on the types of requesters and servers in your Pathway
application and which parts of the application you are working on, you might not need

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -1
Introduction to Pathway Application Programming Which Sections Do You Need?

to read all sections of this manual. Table 1-1 is a descriptive map listing which sections
are relevant to particular programming tasks.

Table 1-1. Task and Manual Correspondences

If Your Application
Includes… You Need… To Perform the Following…
Pathsend requesters Section 2, Designing Your Design an application including
Application Pathsend requesters
Section 3, Writing Pathsend Write a Pathsend requester
Requesters program
Section 5, Pathsend Procedure Look up the syntax of Pathsend
Call Reference procedures
Section 6, Pathsend Errors Look up cause, effect, and
recovery for errors returned to a
Pathsend requester program
Appendix A, NonStop TS/MP Look up limits pertaining to
Limits for Pathsend Requesters Pathsend requesters
Appendix B, Examples See examples of Pathsend
requester programs
Pathway servers Section 2, Designing Your Design an application including
Application Pathway servers
Section 4, Writing Pathway Write a Pathway server program
Servers
Appendix B, Examples See examples of Pathway server
programs

If you are writing SCREEN COBOL requesters, you need the Pathway/TS TCP and
Terminal Programming Guide and the Pathway/TS SCREEN COBOL Reference Manual
for programming information.
If you are writing Pathsend requesters that communicate with NonStop TUXEDO
servers, or if you are writing Pathway servers that handle requests from NonStop
TUXEDO requesters (clients or servers acting as clients), you also need the manuals for
the NonStop TUXEDO system, particularly the NonStop TUXEDO System Application
Development Guide, for additional information. If you are using the Pathway to
TUXEDO translation server or the TUXEDO to Pathway translation server, you also
need the NonStop TUXEDO System Pathway Translation Servers Manual. This manual
provides configuration, startup, and programming information.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -2
Introduction to Pathway Application Programming Advantages of the Pathway Environment

Advantages of the Pathway Environment

NonStop TS/MP provides ease of development, manageability, and the fundamental
strengths and benefits of Tandem NonStop systems. The strengths and benefits of
Tandem systems include data integrity, fault tolerance, high performance and low cost,
system security, scalability, and distributed processing. The following paragraphs
describe how NonStop TS/MP and related products—known together as the Pathway
environment—benefit the application designer and programmer. The Introduction to
NonStop Transaction Processing provides a fuller description of how all the Tandem
fundamentals apply to transaction processing.

Ease of Development
Development costs are one of the highest expenses associated with online transaction
processing (OLTP) systems. The more sophisticated the features and safeguards that are
built into your OLTP application—for example, multiprocessing, fault tolerance, and
data integrity—the greater the costs. When you use NonStop TS/MP and related Tandem
transaction processing products to create your OLTP applications, development time and
efforts, and therefore costs, can be measurably reduced.
This cost reduction occurs because:
• NonStop TS/MP and related products provide the most complex components of an
OLTP application. NonStop TS/MP includes the transaction monitor (PATHMON),
the command interpreter for management (PATHCOM), and the means for
interprocess communication.
In addition, the NonStop Transaction Manager/MP (NonStop TM/MP) product
provides the transaction manager, and the Pathway/TS product provides a
multithreaded terminal control process (TCP) for communication with terminals,
including fault tolerance and transaction protection. (On Tandem NonStop system
models earlier than the Himalaya systems, Pathway/TS is packaged as part of the
Pathway transaction processing system.)
Used with or without NonStop TM/MP and Pathway/TS, NonStop TS/MP provides
a run-time Pathway environment to simplify your development efforts for scalable
OLTP applications on a massively parallel processor architecture.
• Tandem makes valuable application development tools and utilities available for the
Pathway environment. These development tools and utilities can significantly reduce
the amount of programming time and effort required to generate a working Pathway
application.
The Remote Server Call (RSC) product facilitates client/server computing, allowing
workstation applications to access Pathway servers. A large number of packaged
tools and utilities are commercially available for use with RSC, including Tandem’s
Pathway Open Environment Toolkit (POET).
• The Pathway environment helps you standardize program code. You can repeat and
reuse code; you do not have to write the same requester and server programs over
and over again. This ability to reuse code saves development time.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -3
Introduction to Pathway Application Programming Manageability

• The Pathway environment allows you to isolate and test your requester and server
programs before adding them to a running application. This capability is important
because coding errors are difficult, time-consuming, and expensive to find after an
application is put into production.
• OLTP products that are compatible with the Pathway environment are available from
third-party vendors through the Tandem Alliance program.
In addition to making initial development faster and easier, the structured Pathway
environment allows you to implement enhancements and develop new applications by
simply adding new requesters, sharing existing servers, or adding new servers to the
existing application. You can use code modules in the existing application as templates
for new modules in the modified or new application.

Manageability
Online transaction processing operations present a dynamic environment in which
hundreds of different transactions—from disparate locations and many different I/O
devices—can be entered concurrently and processed within seconds. To process
hundreds of transactions, thousands to millions more application program instructions
must be executed. It is critical that you be able to control and monitor such a complex
processing environment.
To control and monitor your Pathway environment—as well as simplify the task of
system management—NonStop TS/MP provides the following:
• A PATHMON process, which provides a single point of control over your OLTP
applications and operations
• A choice of two different system management interfaces: the interactive PATHCOM
interface and the Subsystem Programmatic Interface (SPI)
• Status and error reporting capabilities, provided through a log file and through the
Event Management Service (EMS)
Because NonStop TS/MP provides these processes and capabilities, you do not have to
spend the time and money to develop, test, and implement comparable mechanisms.
For more information about the PATHMON process, the management interfaces, and
status and error reporting capabilities in the Pathway environment, refer to the
NonStop TS/MP System Management Manual, the Pathway/TS System Management
Manual, the NonStop TS/MP Management Programming Manual, and the Pathway/TS
Management Programming Manual.

Data Integrity
If your database is corrupted by a hardware or software failure, you might need weeks to
isolate and then correct the problem. Because an inaccessible or inconsistent database
can have a dramatic, adverse effect on business operations, Tandem developed the
Transaction Management Facility (TMF) subsystem, provided in the NonStop TM/MP
product, as a way of ensuring database consistency. The TMF subsystem, which works
with NonStop TS/MP, protects the entire database from catastrophic system failures by
maintaining an audit trail of database changes (that is, transactions); an audit trail is also

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -4
Introduction to Pathway Application Programming Fault Tolerance

commonly known as a transaction log. You can use the audit trail to rebuild the database
in the event of a hardware or software failure.
The design of Pathway servers supports the integrity of individual transactions and
therefore transaction processing protection as a whole. Because the requester/server
model allows a clear division of processing functions, application programmers can code
each server program to handle a specific set of transaction types: for example, checking
an account balance, entering a new customer, or updating the parts inventory. The server
processes service their transactions by performing the same set of tasks over and over
again. In this way, a valid transaction is defined as a specific set of tasks both by the
requester program and within the server logic.
If for any reason a server is unable to complete all tasks involved in processing a
transaction, it can abort the transaction and thereby maintain the transaction’s integrity.
The server does not have to wait for the requester to abort the transaction.

Fault Tolerance
Because OLTP systems automate core business operations and deliver key business
services, companies depend on OLTP applications to stay up and running—even if a
hardware or software component fails.
Tandem NonStop systems, which are specifically intended for online transaction
processing, are designed to remain continuously available during the hours when
transactions are being entered and business is being conducted. Typically, a Tandem
NonStop system can continue processing despite the failure of any single software or
hardware component within that system. This ability is referred to as fault tolerance.
In the Pathway environment, automatic fault tolerance (that is, fault tolerance that does
not require any additional programming effort on your part) is provided by the use of
process pairs and the actions of the PATHMON process, the TMF subsystem, and the
terminal control process (TCP) provided with the Pathway/TS product.
In the Guardian operating environment, the functions and tasks of an application are
performed by processes, which are running programs. A process pair consists of a
primary process, which does some specific function in the overall work of the
application, and a secondary (backup) process, which remains ready to take over if the
primary process fails. During processing, the primary process keeps the backup process
informed of what it is doing (for example, sending a request) by means of special
interprocess messages, in an activity called checkpointing. Through checkpointing, the
backup process has enough information to take over and continue if the primary process
fails.
Both the PATHMON process and the TCP can be configured as process pairs to support
Pathway applications. When the PATHMON process is configured as a process pair, you
are ensured the ability to control and monitor OLTP system operation even if the
primary PATHMON process fails. When a TCP is configured as a process pair and the
primary TCP fails, terminals controlled by the TCP can still be used.
Pathway server classes provide additional fault tolerance by allowing requests to be
rerouted to surviving server processes in a server class if one server process fails.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -5
Introduction to Pathway Application Programming Other Tandem Fundamentals

Besides process pairs and server classes, fault tolerance in a Pathway application is
ensured by the PATHMON process, the TCP, and the TMF subsystem. Using
information stored in the PATHMON configuration file, the PATHMON process
automatically restarts processes at their initialization level after a failure, allowing these
processes to resume work immediately.

Other Tandem Fundamentals

Besides data integrity and fault tolerance, the Pathway environment also provides the
high performance and low cost, system security, scalability, and distributed processing
of Tandem NonStop systems.

High Performance and Low Cost

The more transactions your system can process (preferably without degrading response
time), the lower the cost of each transaction. The Pathway environment supports fast
response time and high system throughput by allowing:
• Component processes in a Pathway application (for example, requester and server
processes) to reside and execute concurrently in different processors of a
multi-processor system or even a network. This is called multiprocessing.
• More than one Pathway application to run in a Tandem NonStop system.
NonStop TS/MP also supports fast response time and high system throughput by
allowing the replication of processes and programs and the distribution of processes. For
example:
• The PATHMON process can dynamically create additional copies of server
processes at times of peak demand and delete the additional servers when activity
slows again.
• You can add copies of requester and server programs to your Pathway application to
maintain fast response time when the number of users or terminals increases.
• You can distribute processes such as requesters and servers close to the resources
they manage, reducing interprocess communication time within a network.
• You can distribute requesters and servers to less active processors if peak activity on
a particular processor is affecting throughput or response time.

System Security
The Guardian operating environment includes basic mechanisms for controlling access
to files, whether they are data files or program files. Because NonStop TS/MP runs in
the Guardian operating environment, Guardian system security parameters also apply to
Pathway users and processes. In addition, you can supplement the security features of
the Guardian environment with the Safeguard product, which provides authentication,
authorization, and auditing capabilities for Guardian files.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -6
Introduction to Pathway Application Programming Pathway Applications

Scalability
Your organization must be able to expand its transaction processing system as its
operations evolve and its technical requirements change. Tandem NonStop systems are
expressly designed to support incremental, modular expansion, allowing you to increase
the size and processing power of your transaction processing system by:
• Adding hardware and application resources to your existing system
• Linking individual Pathway applications into a single network or adding more
Pathway applications to an existing network
• Supporting an open systems architecture in which standards-based networks as well
as devices and systems from other vendors can be connected to your Tandem system

Distributed Processing
Data communications technology allows organizations to extend their online operations
over long distances to form global networks and to support distributed processing. The
Pathway environment, in conjunction with the Tandem NonStop Kernel operating
system, allows you to distribute application processes within a single system.
Additionally, NonStop TS/MP and NonStop TM/MP, in conjunction with the Expand
networking software, allow you to spread processes, data, and transactions across a
network of Tandem NonStop systems. The coordination of transactions among
application servers residing within an Expand network and possibly accessing different
resource managers (NonStop SQL/MP and Enscribe) is known as distributed transaction
processing (DTP).

Pathway Applications
Pathway applications consist of two types of programs: requester programs and server
programs. This design allows application logic to be distributed near the resources it
manages. For example, presentation services are located near terminal devices or
workstations; database logic resides in server programs near that database. Requesters
and servers communicate by using the Guardian file system or the message system that
is part of the Tandem NonStop Kernel.
Users interact with your application by using devices and processes controlled by your
requester programs. Often these devices are terminals through which the users enter and
retrieve transaction data. They might also, however, be intelligent devices such as
personal computers, workstations, point-of-sale devices, or automatic teller machines
(ATMs). Or, they might be Guardian processes that provide transaction input from a file
or other batch medium.
Server processes receive requests from requester processes to access a database to add,
retrieve, or modify information. Server processes process request messages and send
reply messages with the results of the work on the database.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -7
Introduction to Pathway Application Programming Servers and Server Classes

Servers and Server Classes

You can write Pathway server programs in C, C++, COBOL85, Pascal, the Transaction
Application Language (TAL), the Portable Transaction Application Language (pTAL),
FORTRAN, or Extended BASIC in the Guardian environment. Alternatively, you can
write Pathway server programs in C or COBOL85 in the NonStop Kernel Open System
Services (OSS) environment; you must program such servers to read the Guardian
$RECEIVE file as described in the Open System Services Programmer’s Guide. In both
cases, you configure and manage the servers using the PATHCOM interactive interface
or the Pathway management programming interface (based on the Subsystem
Programmatic Interface, or SPI) in the Guardian environment.
The same server programs, whether developed in the Guardian environment or in the
OSS environment, can be used with several different requester and client interfaces.
These interfaces include SCREEN COBOL, the Pathsend procedures, the Remote Server
Call (RSC) interface, and the Pathway Open Environment Toolkit (POET).
The Pathway environment provides the feature of server classes. A server class is a
collection of replicated Pathway server processes. All server processes in a server class
provide the same set of functions; that is, they execute the same program.

Server Processes
Server processes provide the following benefits:
• Server processes help ensure transaction integrity and, therefore, the integrity of the
database.
• Server code can be reused by many requester programs, and you can separate
presentation services from database functions.
• You can control which transactions can be performed on your node. You can control
the logic of the servers, database names, disk names, and so on.
• In distributed environments, server processes provide high performance by allowing
you to use remote servers instead of performing multiple remote I/O operations,
placing transaction processing close to system resources.

Server Classes
Server classes provide the following benefits:
• You can minimize use of system resources—for example, processes and file opens—
because server classes are shared and highly utilized.
• You can maximize performance because server classes allow multiple copies of
server processes to run concurrently in multiple processors.
• Based on configuration settings determined by the system manager or operator, the
PATHMON process can dynamically create additional server processes within the
server class to maintain acceptable throughput as the workload increases.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -8
Introduction to Pathway Application Programming Requesters

• By temporarily freezing and stopping the server class and changing configuration
parameters, the system manager or operator can adjust the number of servers that are
active at any one time to suit response-time requirements.
• The system manager or operator can balance the workload over multiple processes
and across multiple processors, which provides fault tolerance in addition to load
balancing: if a processor fails, the server class is still available.

Requesters
The Pathway application programming environment provides two programming
interfaces for requesters:
• The Pathsend application program interface (API), provided in the NonStop TS/MP
product
• The SCREEN COBOL language, provided in the Pathway/TS product
Requesters written using these two interfaces are briefly described in the following
paragraphs. In addition, other Tandem products are available to assist you in writing
requesters and clients that communicate with Pathway servers. These products include
the Remote Server Call (RSC) product and the Pathway Open Environment Toolkit
(POET) for workstation clients and the Extended General Device Support (GDSX)
product for front-end and back-end processes.
Section 2, Designing Your Application, provides additional information about how
Pathsend requesters, SCREEN COBOL requesters, RSC and POET clients, and GDSX
processes can be used in Pathway applications.

Pathsend Requesters
The Pathsend procedure calls and the LINKMON process allow Guardian processes to
access Pathway server classes. The Pathsend procedures bring the benefits of Pathway
server classes to a wide range of requesters, providing flexibility in application design.
They also provide high performance for requesters that do not need a complex,
multithreaded interface to terminals or intelligent devices. Finally, they provide support
for both context-free and context-sensitive servers.
Pathsend requesters support the following features:
• Use of the TMF subsystem
• Automatic retry of I/O operations to a server process if the primary process of a
server process pair fails, through use of the Guardian file system
The Extended General Device Support (GDSX) product provides a set of “pseudo
Pathway procedures” that allow you to call Pathsend procedures in the user-supplied part
of a GDSX program. A GDSX process can thus function as a Pathsend requester. GDSX
processes can communicate with devices by means of a number of data communications
protocols, as described in the Extended General Device Support (GDSX) Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1 -9
Introduction to Pathway Application Programming The Pathsend Environment

SCREEN COBOL Requesters

SCREEN COBOL requesters, which are compiled by the SCREEN COBOL compiler
and then interpreted and executed by the terminal control process (TCP), provide ease of
programming if you need to handle large numbers of terminals or intelligent devices or
if you need screen-presentation services. The TCP and the SCREEN COBOL language
produce a high-quality, manageable application. The TCP provides multithreading of
requesters, fault tolerance, terminal device configuration, and operations management so
that you do not need to program these features in your application. Transaction
protection through use of the TMF subsystem, with simplified programming, and
automatic retry of I/O operations are also provided. SCREEN COBOL requesters are
described in the Pathway/TS TCP and Terminal Programming Guide and the
Pathway/TS SCREEN COBOL Reference Manual.
You can use an Extended General Device Support (GDSX) process as a front-end
process to the TCP and SCREEN COBOL requesters to communicate with devices not
directly supported by the TCP. Use of the GDSX product is described in the Extended
General Device Support (GDSX) Manual.

The Pathsend Environment

The Pathsend environment includes Pathsend processes and LINKMON processes:
• Pathsend processes, written as part of your application, use Pathsend procedure calls
to make requests to server classes.
• LINKMON processes, supplied by Tandem, control communication between
Pathsend processes and Pathway server classes.

Pathsend Processes
In writing programs to run as Pathsend processes, you use a set of procedures that are
part of the Guardian procedure library. These procedures allow you to send request
messages to server processes within a server class and to receive the servers’ replies.
You can call the Pathsend procedures from programs written in C, C++, COBOL85,
Pascal, the Transaction Application Language (TAL), or the Portable Transaction
Application Language (pTAL).
Pathsend procedure calls are provided for both context-free and context-sensitive
communication with servers. A context-free server accepts a single message from a
requester, performs the requested tasks, and issues a single reply to respond to the
requester. After the reply message is issued, the server retains no information (context)
that can be used in subsequent requests. A context-sensitive server engages in a
multiple-message communication, or dialog, with a requester. Between messages, the
server retains information (context) pertaining to the dialog.
The use of the Pathsend procedure calls is described in Section 3, Writing Pathsend
Requesters, and their syntax is described in Section 5, Pathsend Procedure Call
Reference. Design considerations related to context-free and context-sensitive servers
are discussed in Section 2, Designing Your Application.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 10
Introduction to Pathway Application Programming LINKMON Processes

LINKMON Processes
LINKMON processes, together with the PATHMON process, perform link-management
functions for Pathsend processes. (A link is a connection to a server process.) A
LINKMON process executes in each processor, or CPU, of a system. As a link manager,
a LINKMON process is responsible for managing links on behalf of all the Pathsend
processes executing in its processor.
If you have the NonStop TS/MP software installed on your system, a LINKMON
process is automatically started in each processor. You cannot start a LINKMON process
with a RUN command.
Figure 1-1 shows a sample Pathsend environment in which Pathsend processes and a
LINKMON process reside in the same processor on system \A. The LINKMON process
sets up communication to the Pathway server class on system \B through the PATHMON
process controlling the server class. The role of the PATHMON process in establishing
this communication is described in Section 3, Writing Pathsend Requesters.
As shown in the figure, only server-class control information is passed to the
LINKMON process; the application data moves directly from the Pathsend requester
process to the server process.

Figure 1-1. Pathsend Interprocess Communication

\A \B

CPU X

Pathsend
Requester

Control
Control Data
Information
Information Server Class X

LINKMON Server
Control Information

004

Although you can obtain some information about LINKMON processes through the
PATHMON process (by means of PATHCOM or SPI), LINKMON processes are not
managed as PATHMON-controlled objects. For details about management of
LINKMON processes, refer to the NonStop TS/MP System Management Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 11
Introduction to Pathway Application Programming Client/Server Capabilities

Client/Server Capabilities
The Remote Server Call (RSC) product and the Pathway Open Environment Toolkit
(POET) bring client/server capabilities to the Pathway environment by allowing you to
move requester functions to a workstation. RSC allows client programs residing on a
workstation to access Pathway server classes in any of three different ways:
• Through a Pathsend requester provided by RSC; this requester works with the
LINKMON process.
• Through a special intelligent device support (IDS) requester supplied with RSC; this
requester works with the terminal control process (TCP) provided in the Pathway/TS
product.
• Through an IDS requester that you develop yourself in the SCREEN COBOL
language; this requester works with the TCP provided in Pathway/TS.
RSC also allows requesters to access Guardian processes directly. To facilitate access to
servers and Guardian processes, RSC consists of multiple components within both the
workstation and Tandem computer environments.
The Pathway Open Environment Toolkit (POET) provides tools for developing RSC
clients for the Microsoft Windows environment. These tools include a simplified
programmatic interface, name mapping, and data conversion mapping.
For information about RSC, refer to the Remote Server Call (RSC) Programming
Manual. For information about POET, refer to the Pathway Open Environment Toolkit
(POET) Programming Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 12
Introduction to Pathway Application Programming Other Transaction Processing Environments

Other Transaction Processing Environments

The NonStop TS/MP product serves as the foundation for open transaction processing
on Tandem NonStop systems. In addition to the Pathway environment, NonStop TS/MP
supports the NonStop TUXEDO transaction processing system. This product allows you
to develop TUXEDO transaction processing applications to run on Tandem NonStop
systems, thus providing these applications with the fundamental advantages of Tandem
NonStop systems.
When using the NonStop TUXEDO system, you work in the NonStop TUXEDO
programming environment; you need not use the requester and server programming
interfaces described in this manual. Note, however, that you can develop applications
that use a combination of modules from the NonStop TUXEDO environment and the
Pathway environment.
You can write a Pathsend requester that also acts as a NonStop TUXEDO client, directly
invoking the services of a NonStop TUXEDO server, by using the NonStop TUXEDO
Application Transaction Monitor Interface (ATMI) functions. For more information
about this mechanism, refer to Section 3, Writing Pathsend Requesters, in this manual
and to the NonStop TUXEDO System Application Development Guide.
Alternatively, you can write a Pathsend or SCREEN COBOL requester that indirectly
invokes the services of a NonStop TUXEDO server by using the Pathway to TUXEDO
translation server provided with the NonStop TUXEDO product. For more information
about this translation server, refer to the NonStop TUXEDO System Pathway Translation
Servers Manual.
You can write a NonStop TUXEDO native System /T client or OSS workstation client
that directly invokes the services of a Pathway server by including calls to the Pathsend
procedures described in this manual. For more information, refer to the NonStop
TUXEDO System Application Development Guide.
Alternatively, you can write a TUXEDO client (or server acting as a client) that
indirectly invokes the services of a Pathway server by using the TUXEDO to Pathway
translation server provided with the NonStop TUXEDO product. This translation server
allows access to Pathway servers from remote TUXEDO requesters (those that use
System /Domain) and non-native TUXEDO workstation clients, neither of which can
make Pathsend procedure calls. For more information about the TUXEDO to Pathway
translation server, refer to the NonStop TUXEDO System Pathway Translation Servers
Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 13
Introduction to Pathway Application Programming Development Tools and Utilities

Development Tools and Utilities

When you are writing requester and server programs for your Pathway application, a
variety of program development tools and utilities are available to you. These tools and
utilities allow you to shorten the amount of time it takes to code, debug, and test your
programs.

Programming Languages and Related Tools

Tandem provides compilers that allow you to write application programs in a number of
programming languages, including C, C++, COBOL85, SCREEN COBOL, Pascal, the
Transaction Application Language (TAL), the Portable Transaction Application
Language (pTAL), FORTRAN, and Extended BASIC. In addition, the Crossref cross-
reference generator is available if you want to supplement the cross-reference listings
provided by the compilers.

The Inspect Symbolic Debugger

The Inspect product is the symbolic program debugging tool for Tandem NonStop
systems. You can use it interactively to examine and modify the execution of Guardian
processes (for example, Pathsend requesters and Pathway servers) as well as SCREEN
COBOL requesters. An online help facility is available for all Inspect commands and
topics.
Using the Inspect product in a Pathway environment requires the use of two terminals or
a terminal emulator with windowing capability. One terminal or window acts as the
application terminal, while the second terminal or window acts as a command or Inspect
terminal.

The Pathmaker Application Generator

The Pathmaker product helps you create Pathway applications consisting of requester
programs written in SCREEN COBOL and server programs written in C or COBOL85.
To create applications with the Pathmaker product, you:
• Enter information about your application into a series of screen-based entry forms,
which the Pathmaker product then stores in a catalog
• Use the Tandem text editor, TEDIT, to create source files containing C or COBOL85
service code
At your command, the Pathmaker product uses the information from the catalog and the
TEDIT file to generate SCREEN COBOL requester code, C or COBOL85 server code,
and command files to configure and start the finished Pathway environment for testing.
The Pathmaker product simplifies the creation of Pathway applications by:
• Generating application code in a uniform structure for all requesters and servers, to
help simplify maintenance and modification
• Producing program statements for tasks that are specific to the Pathway environment

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 14
Introduction to Pathway Application Programming Client/Server Development Tools

• Automatically generating TMF statements in your requester programs when you

indicate that you want your programs to have TMF protection
• Providing a central location for most application information
• Creating error-handling code for the most commonly encountered errors
• Letting you simulate application screens and navigate from one application screen to
another before you write a single line of code
Applications developed with the Pathmaker product can access data from databases
managed by either the NonStop SQL/MP relational database management system or the
Enscribe database record manager. If you are using Pathsend requesters, or clients that
use RSC or POET, you can use the Pathmaker tool to create prototype servers.

Client/Server Development Tools

As mentioned earlier, the Remote Server Call (RSC) product facilitates client/server
computing, allowing workstation applications to access Pathway servers. A large
number of packaged tools and utilities are commercially available for use with RSC,
including Tandem’s Pathway Open Environment Toolkit (POET).

Transaction Processing Scenario

Figure 1-2 and the description that follows it provide an example of how transactions
from Pathsend requesters are processed. The figure shows the path of transactions from
an IBM system to a server class on a Tandem NonStop system.
Note. The figure does not reflect the actual flow of data from the Pathsend requester to the
Pathway server class. Only server-class control information is passed to the LINKMON
process; the application data moves directly from the Pathsend process to the server class.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 15
Introduction to Pathway Application Programming Transaction Processing Scenario

Figure 1-2. Example Application Using a Pathsend Requester

IBM System IBM Terminals

Application

SNA

Tandem NonStop System

SNAX/APN

NonStop TS/MP Environment

PATHMON

Server Class

Request
Pathsend Server
LINKMON
Requester
Reply

Database

CDT017

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 16
Introduction to Pathway Application Programming Transaction Processing Scenario

In this scenario, clerks at an order entry office enter their transactions into terminals
attached to an IBM system. Processing of the transactions, however, requires access to a
database that is linked to a Tandem NonStop system.
1. The clerks enter transactions into their terminals and initiate processing by pressing
function keys. Any preliminary checking or editing is performed by the application
on the IBM system.
2. The IBM system collects the transactions and sends them to a Pathsend requester
located on the Tandem NonStop system. The transactions are sent by using a high-
speed networking product; for example, Tandem’s SNAX Advanced Peer
Networking (SNAX/APN) product.
3. The Pathsend requester accepts the transactions for the Tandem NonStop system and
formats a request message containing the name of the server class and the data
needed by the server to complete its work. The TMF transaction begins.
4. The Pathsend requester forwards the request message to the LINKMON process by
calling the Pathsend SERVERCLASS_SEND_ procedure. (This is a context-free
message.)
5. If the LINKMON process does not have a link to the specified server class, the
LINKMON process asks the PATHMON process for a link to a server process in the
server class. The PATHMON process replies that a server process is available. If the
LINKMON process already has a link to the server class, this step is not performed.
6. The LINKMON process forwards the request to the server process by using
NonStop Kernel interprocess communication.
7. The server process receives and reads the request message.
8. Executing NonStop SQL/MP statements in its program, the server process accesses
the database and updates the appropriate information.
9. The server process formats a reply message, which verifies that the information has
been updated, and replies to the LINKMON process by using NonStop Kernel
interprocess communication.
10. The LINKMON process receives and forwards the reply messages to the Pathsend
requester. The TMF transaction ends.
11. The Pathsend requester returns the reply messages to the IBM system, where the
application displays the information on the terminal screens.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 17
Introduction to Pathway Application Programming Transaction Processing Scenario

NonStop TS/MP Pathsend and Server Programming Manual– 132500

1- 18
 2 Designing Your Application
To develop a functioning Pathway application, you must identify the individual
transactions in your business operations, design and build the application database, and
design and code requester programs and server programs. This section describes the
design of transactions and databases for Pathway applications and the design of
requester and server programs.
To explain these application design tasks, this section uses as an example an application
that processes sales orders for a distributorship. The example shows how the Pathway
environment can be used to create an OLTP application that supports the
distributorship’s order-processing operations.
The distributorship in the example has three offices linked by telecommunications:
• \CORP is a network node at corporate headquarters where the purchasing, accounts
receivable, and accounts payable functions are managed.
• \WHS is a network node in a warehouse where the inventory, shipping, and
receiving functions are performed.
• \REG is a network node in a sales office that is responsible for processing all
customer orders in a particular geographic region. Order-processing functions
consist of entering orders as input and maintaining records of each order. To perform
these two functions, the order processing group:
• Checks with inventory control to determine if items to be ordered are in stock
• Sends inventory control shipping and ordered-items information about each
order
• Gets customer credit information from accounts receivable
• Sends billing information to accounts receivable
• Answers customer inquiries about order status
• Records complete information about each order in the database

Designing Transactions
The first step in developing a Pathway application is to identify and define the
transactions that your application will process. To do this, you isolate the business tasks
you plan to automate, analyze the flow of information within those tasks, list the
transactions that result from the analysis, and then identify the various components of
the transactions. After these tasks are performed, you protect each transaction, and
therefore the integrity and consistency of the database, with the Transaction
Management Facility (TMF) subsystem.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -1
Designing Your Application Analyzing Data Flow

Analyzing Data Flow

Analyzing the flow of data involves identifying what information is required for a
business task, determining the order in which that information is required, and
specifying how the information is to be handled. To automate the order-processing tasks
of the previously described distributorship, for example, you could analyze the flow of
information as follows:
1. Accept the customer’s identification number, a requested delivery date for the order,
and shipping instructions such as the delivery address.
2. Check the customer’s identification number to ensure that the customer is defined in
the \REG database; get the customer’s name and address from the \REG database;
and get a new order identification from the \REG database.
3. Accept a list of order items along with the requested quantity for each order item.
4. Check the current quantity available, in the database on \WHS, of each ordered item
to ensure that sufficient quantity exists to fill the order.
5. Accept any special instructions, such as back-ordering out-of-stock items, required
to process the order.
6. Calculate the total order cost; get the current customer balance and credit limit from
the \REG database; add the total order cost to current customer balance; and ensure
that the new balance does not exceed the customer’s credit limit.
7. Ask the customer to confirm the order.
8. After the customer has confirmed the order, subtract the quantity ordered from the
current quantity available, in the \WHS database, for each ordered item.
9. Add the total order cost to the customer’s current balance in the \REG database.
10. Record the order information in the \REG database.
11. Transmit the order information in the accounts receivable files to the \CORP
database and record the information in the database.
12. Record the order shipping information in inventory files on the \WHS database.
Assume that your analysis of the previous flow of information shows that only two
transactions need to be created to support order processing: an Add New Customers
transaction and an Enter Sales transaction. The Enter Sales transaction, which accepts
and records all the information associated with a customer order, is the example used in
the rest of this section.
The data flow outlined in the previous steps is illustrated in Figure 2-1.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -2
Designing Your Application Analyzing Data Flow

Figure 2-1. Data Flow for a Business Task

1. Accept: 2.
\REG
— Check customer ID.
— Customer ID Customer
— Get customer's name
— Requested shipping date Information
and address.
— Shipping instructions Order IDs
— Get order ID.

3. Accept: 4.
\WHS
—Check quantity available
— List of ordered items
for delivery of each Inventory
— Quantity of each item
ordered item. Data

5. Accept: 6.

— Calculate order cost. \REG

— Back-order information — Get customer balance Customer
and credit limit. Credit
— Subtract order cost from Data
balance and check limit.

7. Accept: 8.
\WHS
— Subtract quantity ordered
— Confirmation from quantity available. Inventory
Data/
Shipping
Data
9, 10.
\REG
— Update customer balance.
— Record order data. Customer
Credit Data/
Order Data

11.
\CORP
— Record order information.
Accounts
Receivable
Data
12.
\WHS
— Record order information.
Inventory
Data/
Shipping
Data
CDT028

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -3
Designing Your Application Identifying Transaction Components

Identifying Transaction Components

After you have identified the Enter Sales transaction for the order-processing
application, you list the functions performed by the transaction and group them either
into data collection and validation operations or into database update operations. For
example, the key functions performed by the Enter Sales transaction during data
collection and validation are:
• Assembling information for the order header, including:
• Obtaining the order-ID
• Accepting the customer-ID
• Accepting the requested delivery date
• Accepting shipping instructions
• Checking the customer-ID
• Obtaining the customer’s name and address from the database
• Assembling the order, including:
• Accepting the list of order items and the quantity of each item
• Checking the current quantity available for each item ordered
• Accepting special instructions
• Calculating total order cost
• Obtaining the customer’s balance and credit limit from the database
• Adding the total cost to the customer’s balance and ensuring that it does not
exceed the credit limit
The key function performed by the Enter Sales transaction during database update
operations is order completion. The order completion function includes:
• Subtracting the quantity ordered from the current quantity available for each ordered
item
• Adding the total order cost to the customer’s current account balance
• Recording the order in the database
• Recording the order invoice in the accounts receivable files
• Recording order shipping information in the inventory files
The relationships of the various functions for the Enter Sales transaction are illustrated
in Figure 2-2. The dark arrows in the figure show the sequence of actions from Step 1
through Step 3. The lighter arrows show the flow of information.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -4
Designing Your Application Identifying Transaction Components

Figure 2-2. Relationships Between Transaction Functions

Order-ID
Order-ID

Customer Information
1 Customer
Customer-ID
Assemble Order Header
Customer-Details Header

Item Detail Customer

Items
Credit
Customer Balance

Item Quantity Available Quantity

2 Available
Item-ID, Quantity
Accept
Display Item Details Items Order

Quantity Ordered

Order Cost
3
Display Totals
Confirm Order Complete Order Totals
"Done" Order 4

Shipping Request Ship-ID

I/C

Invoice Request Invoice-ID

Legend A/R

1 Assemble information for order header; display at terminal and add to database;
optionally change customer information.
2 Accept items in order, check item availability and customer credit; display item details
at terminal and add to database.
3 Display totals at terminal and get confirmation; update item quantity and customer balance;
add totals to database; inform related applications about order.
4 Later, when order is shipped and customer billed, add shipping and invoice numbers to database.
CDT029

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -5
Designing Your Application Protecting Transactions

Protecting Transactions
After listing and grouping the components of the Enter Sales transaction, you protect the
integrity of each transaction, and ultimately the consistency of the database, with the
TMF subsystem. The following pages outline how to integrate the TMF subsystem with
your business transactions. For information about the overall features of the TMF
subsystem, including database file recovery and audit trails, refer to the Introduction to
NonStop Transaction Processing.

Defining TMF Transactions

From a systems perspective, a transaction includes all the steps necessary to transform a
database from one consistent state to another. A TMF transaction must be constructed as
a logical unit of work: that is, all parts of a transaction, which usually consists of
multiple operations, must be handled as a single entity. If any parts of a TMF transaction
are not successfully completed or applied to a database, then none of the transaction
parts are applied to the database. By forcing all components of a transaction to be
handled as a single unit of work, the TMF subsystem prevents inaccurate or partial
updates to the database and protects database consistency.
At the application level, a TMF transaction is defined by special procedure calls or
statements that specify the beginning and end of a transaction. For example, in a
Pathsend requester program, a transaction begins with a call to the
BEGINTRANSACTION procedure and ends with a call to the ENDTRANSACTION or
ABORTTRANSACTION procedure. The procedure calls that define TMF transactions
act as brackets; that is, the statements are placed before and after the add record, update
record, and delete record procedures in your requester program.
Figure 2-3 illustrates the use of the TMF subsystem by a Pathsend requester program
and a Pathway server program. In the illustration, the variable called TRANSID acts as a
transaction identifier.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -6
Designing Your Application Protecting Transactions

Figure 2-3. Pathway Application Programming for the TMF Subsystem

COBOL85, C, COBOL85, C,
C++, Pascal, or C++, Pascal,
TAL TAL, ...

BEGINTRANSACTION Process request by:

•
• TRANSID —Reading files TRANSID
•
Request database —Locking records
services (following TMF
• record-locking rules)
Audited
•
—Changing records Database
Perform checkpointing
in files (changes are Files
as needed
• associated with
• TRANSID supplied
Abort and restart by requester)
transaction if necessary
• —Aborting transaction
• if necessary
ABORTTRANSACTION •
or •
ENDTRANSACTION Reply to request

Requester Server
CDT032

Database Consistency and Concurrency

Potentially, all operations that alter the database are candidates for TMF protection. But
before you can apply TMF protection to your transactions, you need to determine:
• When to begin a TMF transaction
• Whether all of the database update operations have to happen together in the same
TMF transaction or whether they can be parts of different transactions
To answer these issues, you have to establish your criteria for database consistency and
decide how much processing concurrency you can achieve in the application. For
example, the Enter Sales transaction affects several pieces of information: order data,
inventory data, shipping data, customer credit, and receivables. Upon examination of this
transaction, you will see that it is possible to make one general assertion about order
processing and about the Enter Sales transaction in particular: An order is not complete
until every piece of information associated with the order is recorded in the \REG,
\CORP, and \WHS databases.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -7
Designing Your Application Protecting Transactions

To illustrate this assertion, consider a situation where a transaction fails after it changes
the customer’s balance, records the order information, and records the order invoice, but
before it records the shipping information. In this scenario, the customer is going to be
billed for an order never received. Consequently, your basic criterion for database
consistency is as follows: all database updates that are related to the order must be part
of one TMF transaction.
Any record modified or inserted by a database operation that is protected by the TMF
subsystem is locked and unavailable to other transactions until the initial transaction
ends successfully. This type of locking protocol means that you always have a design
tradeoff—consistency versus concurrency—with respect to locking records that are
actively accessed by the application. If records are locked too early, other transactions
cannot access them and the application’s concurrency (its ability to process many
transactions at the same time) suffers.
As the Enter Sales transaction demonstrates, all of the data collection and validation
operations can happen before you begin the TMF transaction—although some
revalidation may be done again as part of the transaction. Assembling the order header
and assembling the order involve reading records in the database but not changing the
records. The rest of the operations change the database and should all be done within a
TMF transaction.
As a general rule, you should design the application’s transactions to maintain
consistency under all circumstances. After the application is installed and running
successfully, you can look for ways to improve its concurrency.

Aborting Transactions
If the requester or the server program detects a problem during the processing of a TMF
transaction, the requester or server causes the transaction to be aborted with a special
procedure call or statement (for example, a call to ABORTTRANSACTION in a
Pathsend program). For requesters, the statement that aborts a transaction is executed in
lieu of the statement that ends a transaction; for example, in a Pathsend program the
requester either completes the transaction with a call to ENDTRANSACTION or causes
it to be backed out, because of an error, with a call to ABORTTRANSACTION.
In the past, program designs typically assigned the task of aborting transactions to
requesters. Current program design often assigns that task to servers. Servers abort
transactions and inform the requesters of those actions, thus ensuring protection of data.
The aborting of transactions by servers is described further under Designing Server
Programs, later in this section.
The TMF subsystem backs out aborted transactions by using information contained in
the TMF audit-trail files. For more information about transaction backout and audit-trail
files, refer to the NonStop TM/MP Application Programmer’s Guide.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -8
Designing Your Application Designing the Database

Designing the Database

The next step in developing a Pathway application is to design the database that will be
accessed and updated by the application. Designing the database, which is a highly
specialized activity typically performed by experienced database administrators,
involves:
• Precisely identifying the meaning and use of the data as it exists in your business
and specifying the database files and records that will store this data. This step is
referred to as logical design.
• Choosing file types and keys for the records. This step is referred to as physical
design.
In addition to completing a logical and physical design of your database, you must also
select a database manager product and ensure that your server programs can interface
with that database manager.

Logical Design
During the logical design process, you determine which classes of data must be
maintained by your application and identify the relationships that exist between the
classes. Each class of data names something that the database will store information
about. For example, in an application that processes sales orders, orders is a class of
data and order-items is a relationship between a particular order and the inventory
items within the order. These data classes and relationships generally become records in
files accessed by the application.
After specifying data classes, you list the attributes (data items) for each class of data.
For example, some of the attributes are order-ID, cust-ID, and order-total.
These attributes become fields in the records of the database. After specifying attributes
for data classes, you diagram the relationships between each of the files in the database
and then normalize your database files. To normalize files is to ensure, at a minimum,
that:
• There are no repeating fields.
• Data is dependent on the entire key (a unique element) of a field.
• Data is dependent on nothing but the key.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2 -9
Designing Your Application Physical Design

Physical Design
You undertake the physical design of your database by selecting the appropriate file
types and record keys for each of the files in the database. Whether you are using the
NonStop SQL/MP software or the Enscribe software as your database management
system (DBMS), these file types can be classified as key-sequenced, relative, entry-
sequenced, or unstructured:
Key-sequenced Each record in the file has a primary key and up to 255 alternate
keys. The primary key is a field or combination of fields within the
record.
Relative Each record in the file has a unique record number, which is the
primary key, and can have up to 255 alternate keys. The record
number is a unique value that corresponds to the physical location
of the record within the file.
Entry-sequenced Each record in the file has a unique record number and can have up
to 255 alternate keys. The record number corresponds to the order
in which a record is stored in the file. The primary key is the
relative byte address of the record.
Unstructured Each record in the file has a unique record number that can be used
as the primary key. Alternate keys are not supported.
Although the file type you choose depends on your application requirements, generally
you should choose key-sequenced files for a database that will be accessed and
maintained by a Pathway application. Key-sequenced files provide more flexibility than
the other file types.

Database Managers
Databases supporting Pathway applications can run under either the NonStop SQL/MP
relational database management system or the Enscribe database record manager. Both
of these products support the creation and use of large databases capable of operating in
local or distributed systems.
The NonStop SQL/MP (Structured Query Language/MP) product is both a database
management system (DBMS) for production environments and a relational database
management system (RDBMS) for decision-making in an information-center
environment. The NonStop SQL/MP product allows you to think about and represent
files in the database as a collection of similarly structured lists. For more information
about designing NonStop SQL/MP databases, refer to the NonStop SQL/MP Reference
Manual.
The Enscribe database record manager provides a record-at-a-time interface between
Pathway servers and your database. For more information about designing Enscribe
databases, refer to the Enscribe Programmer’s Guide.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 10
Designing Your Application Remote Duplicate Database Facility (RDF)

Remote Duplicate Database Facility (RDF)

If disaster recovery of your database is important, the Remote Duplicate Database
Facility (RDF) is available to maintain a copy of the database on a remote system. The
RDF product monitors database updates audited by the TMF subsystem and applies
those updates to the remote copy of the database. For more information about the RDF
product, refer to the Remote Duplicate Database Facility (RDF) System Management
Manual.

Designing Requester Programs

To facilitate the accessing of Pathway server classes from different transaction sources,
you can develop requester programs for a Pathway application that use any of the
following access approaches:
• SCREEN COBOL and the TCP
• SCREEN COBOL and the TCP with the intelligent device support (IDS) facility
• The Pathsend procedure calls
• The Remote Server Call (RSC) product, with or without the Pathway Open
Environment Toolkit (POET)
• The Extended General Device Support (GDSX) product
In Table 2-1, key technical and business considerations are mapped to each way of
accessing Pathway servers. More information about each approach is provided following
the table.

Table 2-1. Considerations for Requester Programs

Support
Large Support for
Server Number for Multi- High Ease of Fault Context
Access of I/O Intelligent Threading Perform- Develop- Toler- TMF Sensi-
Approach Devices Devices Capability ance ment ance Support tivity
TCP X X X X X
TCP with X X X X X X
IDS
Pathsend X X X X
RSC or X X X X X X
POET
GDSX X X X X X X

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 11
Designing Your Application SCREEN COBOL Requesters

SCREEN COBOL Requesters

Screen programs for Pathway terminals perform a variety of front-end functions for your
Pathway application and are typically written as single-threaded programs in the
SCREEN COBOL language. This language offers a simple programming environment
and screen-management system to drive Tandem terminals and IBM 3270 terminals.
SCREEN COBOL supports both conversational mode (for either block-mode or
conversational-mode terminals) and intelligent mode (for intelligent devices and
communications lines).
When you write a screen program in SCREEN COBOL, you can take advantage of the
features of the Pathway/TS TCP. As supplied by Tandem, the TCP supports:
• Fault tolerance
• TMF transactions
• Multitasking of single-threaded screen programs
• Access to server processes with Pathway server classes
• Unsolicited message processing (UMP)
• System management interfaces (that is, PATHCOM or the Pathway management
programming interface)
SCREEN COBOL requester programs do not perform any file I/O operations except to
terminals and server classes. A file I/O operation to a server class, which is in the form
of a request message, is initiated by the requester program by using the SCREEN
COBOL SEND statement.
For information about designing and coding SCREEN COBOL requesters, refer to the
Pathway/TS TCP and Terminal Programming Guide.

IDS Requesters
Standard SCREEN COBOL requesters are screen oriented; they send data back and
forth between the Working-Storage Section of the program and a terminal’s display
screen by way of screen templates defined in the Screen Section. Standard SCREEN
COBOL requesters use SCREEN COBOL ACCEPT and DISPLAY statements in the
Procedure Division to interact with display terminals.
SCREEN COBOL requesters that employ the IDS facility within the TCP send data
back and forth between the Working-Storage Section and an intelligent device (or a
front-end process that controls the device) by way of message templates defined in the
Message Section. IDS requesters use SCREEN COBOL SEND MESSAGE statements
and their associated REPLY clauses in the Procedure Division to interact with the
intelligent devices or front-end processes.
Although IDS sends and receives data through Message Section templates instead of
Screen Section templates, the TCP still provides:
• Link management for access to Pathway server classes
• TMF support to ensure transaction protection and database integrity

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 12
Designing Your Application Pathsend Requesters

• Fault tolerance through process pairs

• Multithreading and multitasking
• Expanded I/O editing support for data streams from intelligent devices
For information about designing and coding IDS requesters, refer to the Pathway/TS
TCP and Terminal Programming Guide.

Pathsend Requesters
As an alternative to writing SCREEN COBOL requesters, you can write Pathsend
requesters in C, C++, COBOL85, Pascal, TAL, or pTAL. In such requesters, you use
Pathsend procedure calls to communicate with Pathway servers. The LINKMON
process manages links to your server processes on behalf of Pathsend requesters.

Design Considerations
Pathsend requesters are a good choice for your applications if you need to do the
following:
• Take a high volume of transactions from a limited number of devices. In this
scenario, there are relatively few requester processes, the requesters are busy, and
configuration and management is minimal.
• Access servers that are shared by Pathway requesters and applications other than
OLTP applications: for example, a security checking server or a logging server. If
such servers are used infrequently or if the workload varies, server processes can be
automatically deleted when not needed and restarted through the PATHMON
process when needed again.
• Access servers from environments containing a mix of online transaction processing
and batch processing: that is, environments where the same set of servers handle
both online requests and requests from batch applications such as NetBatch Plus
processes.
• Write nested servers, which act as requesters by making requests to servers in other
server classes, perhaps server classes managed by a different PATHMON process.
• Write context-sensitive servers, which are discussed later in this section under
“Designing Server Programs.”
Pathsend procedure calls give you more flexibility than WRITEREAD calls for server-
to-server communication. The application gets all the advantages of server classes,
including advantages not readily available with WRITEREAD; for example, load
balancing, adjusting the number of servers to fit response-time requirements, and
configuration and operations management. You can use the Pathsend procedure calls in
C, C++, COBOL85, Pascal, TAL, and pTAL programs.
The Pathsend procedures and the LINKMON process, however, do not provide
multithreading, fault tolerance, device configuration, or operations management for
requesters. Therefore, if you need these capabilities in a Pathsend requester, you must
provide the programming for them.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 13
Designing Your Application Pathsend Requesters

In addition, Pathsend procedure calls that send messages to server classes must be
protected by the TMF subsystem to ensure data integrity in your Pathway application.
The Pathsend procedures and the LINKMON process do not support the checkpointing
of Guardian interprocess message synchronization IDs. This lack of checkpointing
support is an important consideration when writing fault-tolerant requester programs
that do not use the TMF subsystem. Section 3, Writing Pathsend Requesters, provides
more information about writing fault-tolerant Pathsend programs.
The Pathsend procedures allow you to indicate a specific timeout value for each message
sent to a server class. For example, if you perform SERVERCLASS_SEND_ calls to
local and remote systems, you can specify shorter timeout values for the local sends and
longer values for the remote sends.
You can restrict access to Pathway server classes by Pathsend requesters by having the
LINKMON process perform security authorization checks on each send operation.
The Pathsend procedures and the LINKMON process log error messages in the event of
a processing failure. Your Pathsend requester can check for these errors and perform
recovery actions.

Program Structure
The example in Example 2-1 outlines a Pathsend requester program written in TAL.
This program handles data entry for the order-processing application introduced at the
beginning of this section.
Note. The program in Example 2-1 illustrates program structure only; it is not a complete
program. For an example of a complete, running Pathsend requester program, refer to
Appendix B, Examples.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 14
Designing Your Application Clients Using RSC and POET

Example 2-1. Sample Pathsend Requester Program Structure

Declare program variables.

TMFError := BEGINTRANSACTION( TransactionTag ): Begins TMF transaction.

Allocate buffer for request
and reply messages.

PathmonName ‘:=‘ [“$PM”] Set SERVERCLASS_SEND_ parameters.

PathmonNameBytes :=3;
ServerClass ‘:=‘ [“ASERVER”]
ServerClassBytes =7;
Timeout := -1D; Specifies no timeout.
Flags := 1; Specifies nowait send.
SendError := SERVERCLASS_SEND_( PathmonName, Performs send operation.
PathmonNameBytes, ServerClass, ServerClassBytes,
RequestBuffer, RequestBufferBytes, MaxReadCount,
CountRead, Timeout, Flags, SCSendOpNum, Tag);
IF( SendError ) THEN Nowait send request failed.
BEGIN
InfoError := SERVERCLASS_SEND_INFO_ Gets Pathsend and File errors.
( PSError, FSError ); Perform application-dependent
. error logic.
.
.
END;

Filenum:=SCSendOpNum
CALL AWAITIOX ( FileNum, Countread, Tag, -1D ); Wait for completion of send
operation
CALL FILEINFO ( FileNum, SendError ); and get the resulting send error.

IF( SendError ) THEN

BEGIN
INFOERROR := SERVERCLASS_SEND_INFO_ ( PSError, FSError );
.
.
.
END;

TMFError :=RESUMETRANSACTION( TransactionTag ); Resumes TMF transaction.

IF( ReplyError ) THEN Perform application-dependent

. error logic, which may include
. aborting transaction.
.
ELSE
TMFError :=ENDTRANSACTION;

Clients Using RSC and POET

The RSC (Remote Server Call) product facilitates client/server computing, allowing
workstation applications to access Pathway server classes and Guardian processes.
The RSC product supports a number of different transport protocols and workstation
platforms. For detailed information about the supported platforms and protocols, refer
to the Remote Server Call (RSC) Programming Manual.
Transactions are transmitted from the workstation application (the client) to a Pathway
application running on a Tandem NonStop system (the server) by means of a supported
communications protocol, such as NETBIOS, TCP/IP, or an asynchronous connection.
RSC includes a process called the Transaction Delivery Process (TDP), which resides on
the Tandem NonStop system. The TDP is a multithreaded process that can handle
multiple workstations. It routes request messages from workstations to Pathway server
NonStop TS/MP Pathsend and Server Programming Manual– 132500
2- 15
Designing Your Application Requesters Using GDSX

classes by using either the Pathsend API and the LINKMON process or the terminal
control process (TCP) provided in the Pathway/TS product. If the TCP is used, it can
route a request message to a Pathway server by using either the intelligent device
support (IDS) requester supplied as part of RSC or an IDS requester that you develop
yourself. The TDP can also send request messages from a workstation to a Guardian
process.
The Pathway Open Environment Toolkit (POET) provides tools for developing RSC
clients for the Microsoft Windows environment. These tools include a simplified
programmatic interface, name mapping, and data conversion mapping.
For information about designing and coding requesters with the RSC product, refer to
the Remote Server Call (RSC) Programming Manual. For information about using the
POET product, refer to the Pathway Open Environment Toolkit (POET) Programming
Manual.

Requesters Using GDSX

The Extended General Device Support (GDSX) communications subsystem product
simplifies the development of front-end processes and back-end processes for
communication with I/O devices. These devices can be of any type, including
workstations, terminals, ATMs, point-of-sale (POS) devices, and industrial robots.
GDSX supplies code that provides multitasking and other features useful for developing
these front-end and back-end processes.
A GDSX process can act as a front-end process for LINKMON processes or a
Pathway/TS terminal control process (TCP).
A GDSX process contains two primary parts:
• TSCODE, supplied by Tandem
• USCODE, supplied by the application programmer
TSCODE provides generic routines and management services that help you build a
multithreaded, fault-tolerant process. TSCODE provides the following functions:
• Creates new tasks and stops tasks
• Receives all system messages and I/O requests
• Dispatches (wakes up and executes) the appropriate active task to process messages
and requests
• Handles errors
USCODE consists of user exits that are called by TSCODE to handle the application-
specific, data communications-related functions, such as data manipulation, protocol
conversion, and message routing for the I/O process. USCODE is typically written in the
Transaction Application Language (TAL) or the Portable Transaction Application
Language (pTAL) and bound with TSCODE to produce a functional GDSX process.
GDSX provides its own interface to Guardian procedures, NonStop TM/MP procedures,
and Pathsend procedures. The names of the GDSX procedures typically look like their
Guardian, NonStop TM/MP, or Pathsend equivalents, but they have a circumflex (^)
NonStop TS/MP Pathsend and Server Programming Manual– 132500
2- 16
Designing Your Application Requesters Using GDSX

character inserted before the procedure name. For example, SERVERCLASS_SEND_

becomes ^SERVERCLASS_SEND_. The GDSX interface supports both context-free
and context-sensitive Pathsend procedures.
When a GDSX process is used as a front-end process, multiple threads of a user-coded
device handler provide separate tasks to manage the input from I/O devices and provide
functions such as data-stream conversion, implementation of a communications
protocol, and network communications error handling. One instance of the device
handler manages one I/O device.
If the GDSX process is acting as a front-end process for a TCP, the GDSX process
simulates a terminal supported by the TCP; the simulated terminal is typically run by an
IDS requester program. When the IDS facility is used, the GDSX process does not
ordinarily control how data appears to the intelligent devices, nor does it perform any
other device-dependent functions. However, the GDSX process can be designed to
perform device-dependent functions if needed.
A GDSX process can also act as a front-end process to a LINKMON process, as shown
in Figure 2-4. The figure shows the path of a transaction from a general device to a
Pathway server through a GDSX process.
In this example, the GDSX device handler contains the application requester logic and
uses the Pathsend interface to communicate with Pathway servers. Normal interaction
with a server process for each thread is similar to that of a Pathsend requester process.
Note. Figure 2-4 does not reflect the actual flow of data from the GDSX processes to the
Pathway server class. Only server-class control information is passed to the LINKMON
process; the application data moves directly from the GDSX process to the server class.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 17
Designing Your Application Requesters Using GDSX

Figure 2-4. GDSX as a Front-End Process

Tandem NonStop System

PATHMON

Server Class

GDSX LINKMON Server

General Device

NonStop
TM/MP

NonStop
SQL/MP

Database

CDT126

When developing a front-end process using GDSX, consider the following points:
• A GDSX front-end process is a good choice when a specified data communications
protocol is not supported by the Pathway TCP but is supported by GDSX.
• A GDSX front-end process is also a good choice when performance is critical.
SCREEN COBOL might not be efficient enough to handle a large amount of
application function.
• A GDSX process can be designed to be context-sensitive, through use of the
context-sensitive Pathsend pseudo-procedures.
• GDSX processes are managed either through the Subsystem Control Facility (SCF)
interactive interface or through a management application program using the
Subsystem Programmatic Interface (SPI).
For further information about designing and coding GDSX processes, refer to the
Extended General Device Support (GDSX) Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 18
Designing Your Application Dividing Function Between Requester and Server

Dividing Function Between Requester and Server

In designing a Pathway application, you must decide how to divide function between
requester and server. In making this decision, you should consider the type of requester
or client you are writing (SCREEN COBOL, Pathsend, RSC, or GDSX), and you should
also consider performance, maintainability, and other factors.
For example, which program module should check entry fields for validity? If you are
writing a SCREEN COBOL requester, you can easily code it so that the TCP performs
these checks. However, a special edit-checking server could provide better performance.
If your application includes a workstation requester that communicates with servers
using RSC, having the requester check the entry fields would save communications
overhead.
As another example, which program module should change screen field attributes such
as color, blink, brightness or reverse video for such purposes as highlighting an entry
field that contains an error? The SCREEN COBOL language allows such work to be
done by the requester, but it could also be done by the server.
For more considerations about dividing function among modules within an application,
see Packaging Server Functions, under Designing Server Programs, later in this section.

Designing Server Programs

Request validations, security checks, calculations, database inquiries, and database
changes made in response to a request message should be performed by individual units
of code within Pathway server programs. As an application programmer, your task is to
create a server program to perform specific tasks (for example, create a customer
account).
You can write Pathway server programs in C, C++, COBOL85, Pascal, TAL, pTAL,
FORTRAN, or Extended BASIC in the Guardian environment. Alternatively, you can
write Pathway server programs in C or COBOL85 in the NonStop Kernel Open System
Services (OSS) environment; you must program such servers to read the Guardian
$RECEIVE file as described in the Open System Services Programmer’s Guide. In both
cases, you configure and manage the servers using the PATHCOM interactive interface
or the Pathway management programming interface (based on the Subsystem
Programmatic Interface, or SPI) in the Guardian environment.
Regardless of which operating environment or programming language you use, your
Pathway server programs can access database files through the NonStop SQL/MP
relational database management system or the Enscribe database record manager. Refer
to Designing the Database earlier in this section for information about these two
database managers.
You can use the same server programs, whether developed in the Guardian environment
or in the OSS environment, with several different requester and client interfaces. These
interfaces include SCREEN COBOL, the Pathsend procedures, the Remote Server Call
(RSC) interface, and the Pathway Open Environment Toolkit (POET). Requesters or
clients using different interfaces can share the same Pathway server classes if you ensure
that the server program’s request and reply formats are consistent for all requesters.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 19
Designing Your Application Design Considerations

After you code and compile your server program, the server object code and library code
are shared among all processes of the same server class.

Design Considerations
Before structuring and coding a server program, you should consider some design issues
that can affect server performance and efficiency. First, you must decide whether to
program single-threaded or multithreaded servers. Additionally, you should be aware of
the issues related to context-free versus context-sensitive servers, server packaging,
nested servers, aborting transactions, process pairs, early replies, and audited and
unaudited servers. You might also consider the use of a GDSX back-end process.

Single-Threaded Versus Multithreaded Servers

When writing Pathway server programs, you need to consider whether to write them as
single-threaded or multithreaded programs.
Typically, you can get solid performance from single-threaded servers, which are
simpler to design, program, and maintain than multithreaded server programs. For most
applications, single-threaded server design is recommended.
Single-threaded servers generally perform well because they are highly utilized. Server
processes handle requests from many requester programs (for example, a few servers
might support 100 terminals), keeping the server processes highly utilized. Servers are
not idle waiting for input from a single requester or device; they can get work from any
requester or device.
Low server utilization, however, might still result if the server experiences long waits
while processing a request. The main reason for writing multithreaded servers is to
provide resource efficiency in applications where processes have long waits (for
example, when a server creates Guardian processes on demand or communicates with a
remote system from another vendor) or shared access (for example, when requests are
multiplexed over a single link).

Context-Free Servers Versus Context-Sensitive Servers

If your servers are programmed to be context-free, the relationship between an
application requester program (for example, a SCREEN COBOL program) and a
Pathway server process exists only for the duration of a single send operation: that is,
one request message and the server’s reply to it. Subsequent send operations to the same
server class could be serviced by any server process in the server class.
This design allows server processes to be easily shared and serially reused by many
requesters. The sharing and reusing of server processes results in more highly utilized
servers and consequently can require fewer server processes, depending on the service
time required. The assignment of requests to server processes on a per-request basis also
allows the PATHMON process to improve the distribution of work among the available
servers.
However, context-free design requires that your servers not save any information (that is,
context) from previous requests, such as counts, subscripts, or record pointers.
Transaction context must be maintained outside of the server, either by the requester

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 20
Designing Your Application Design Considerations

program or in the database. When you program a server to be context-free, you code the
server to be independent of its previous request. In essence, every request must be
treated as if it were the first request received by the server.
SCREEN COBOL and the TCP support only context-free servers. However, if you use
Pathsend requesters, you can use either context-free or context-sensitive servers. In most
cases, it is preferable to use context-free servers. However, context-sensitive servers
might be a better solution if you need to do one of the following:
• Retrieve large blocks of information (larger than 32 KB) from a database
• Browse a database and repeatedly get the next record, saving the cursor position
Context-sensitive servers require additional programming for both requester and server.
The requester must first establish a dialog with a server class and then send messages
within the dialog. All messages in a dialog will be sent to the same server process in the
server class. Programming details are explained in Section 3, Writing Pathsend
Requesters, and Section 4, Writing Pathway Servers.

Packaging Server Functions

Another major decision that you must make during server design is how to package the
individual functions, or services, that the server performs. This decision is related to the
decision about how to divide function between requester and server, discussed earlier in
this section at the end of the subsection Designing Requester Programs.
Based on various operational considerations (for example, security, management, file
access, throughput, and response-time requirements), you could choose one or more of
the following server packaging alternatives:
• One server class for each database: Called a single-function server, this server
(for instance, a list-item server or a change-item-quantity server) services an entire
database.
• One server class for each file: Each file has its own dedicated server that executes all
I/O (reads, updates, adds, and deletes) against the file. The server performs no
functions beyond file access.
• Similar service times: Transactions with similar I/O rates are grouped and processed
by the same server. Grouping transactions with different service times in separate
servers minimizes server queues and facilitates application tuning.
• One server for each business function: A single server handles all tasks of a specific
business function, including business decisions (for example, loan approvals based
on predefined criteria) and database navigation.
• One server for similar business functions: Similar business transactions (for
instance, enter orders less than $100 in value or delete orders) are handled by a
single server.
• One server for similar business functions and all database functions: A single server
services similar business transactions and handles both business decisions and
database navigation.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 21
Designing Your Application Design Considerations

• Update or read-only server: A single server exclusively handles either update

transactions or inquiry transactions.
One approach to packaging server functions is to first group server functions based on
management considerations (for example, all servers within a server class must freeze
and stop as a unit) and security considerations (for example, the server class must
execute under one user ID). Then, partition server functions based on the database files
that are most frequently accessed.
To ensure acceptable response times for users and allow you to tune your application for
performance, it is very important to partition server functions based on service times. If
the same set of servers handles short and long transactions, some requests for short
transactions will be queued behind long transactions, resulting in poor response times
for the short requests. If the short and long transactions perform different functions, put
those functions in separate server programs. If the short and long transactions perform
essentially the same work—for example, a simple database lookup—but some requests
could be for multiple lookups, you can configure two or more server classes for requests
of different lengths, all using the same program code.

Nested Servers
A server written in C, C++, COBOL85, Pascal, TAL, or pTAL can use the Pathsend
procedures to send a request message to a server in another server class and receive a
reply. In such a case, the server is acting as a requester. Servers communicating with
each other in this manner are called nested servers.
For example, consider a situation where a requester on one node requires the services of
two server classes on another node. Instead of sending to server class A, waiting for a
reply, and then sending to server class B, the requester could send to server class A, and
server class A could send to server class B, get the response, and then reply to the
requester. This use of nested servers reduces the number of messages sent across data
communications lines and enables application logic to be distributed near the resources
it manages.
Consider the following when considering the use of nested server programs:
• Single-threaded servers that send to other server classes can cause process
deadlocks. A process deadlock is a situation in which two processes cannot proceed
because each is waiting for a reply from the other.
For example, if a process in server class A sends a request to server class B and the
process in server class B then sends a request to server class A, a deadlock might
occur. Even if there is more than one process in server class A, there is no guarantee
that the second request would not be sent to the same process that sent the original
request.
To avoid this problem, the server program for server class A should keep a read
operation posted on $RECEIVE and wait for completion of either the send operation
or the read operation. Although this multithreading increases the complexity of the
program, it is necessary in order to prevent deadlock.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 22
Designing Your Application Design Considerations

• Single-threaded servers that send to other server classes can cause low server
utilization in the same way that any single-threaded process that calls another
process can: the server process sending the request is idle until it receives a reply
from the server to which it sent the request.
• Single-threaded servers that send to other server classes can, therefore, result in
longer queues for a server class, and these longer queues can affect application
performance.

Aborting Transactions
A request sent to the server can have one of three outcomes:
• All the work for the request was completed successfully.
• None of the work for the request was completed.
• The work for the request was only partially completed.
In the first case, the requester can commit the transaction. In the second case, the
requester can commit the transaction and then retry it. In both of these cases, the
information in the server’s reply is sufficient to ensure the integrity of the transaction.
However, if the transaction work was only partially completed, as in the third case, the
server needs to ensure that the transaction is not committed, so that the incomplete work
can be backed out. To ensure transaction backout, the server should call the
ABORTTRANSACTION procedure after it reads the server’s request and before it sends
its reply. A call to ABORTTRANSACTION by the server does not end the transaction—
only the requester can end it—but it ensures that the transaction is aborted. The
requester should then call either ABORTTRANSACTION or ENDTRANSACTION
after it replies to the server. (If the requester calls ENDTRANSACTION in this
situation, the ENDTRANSACTION call returns an error because the transaction has
already been aborted. However, either call ensures that the resources associated with the
transaction are released.)

Fault-Tolerant Process Pairs

You should not write your server processes as fault-tolerant process pairs when you are
using the TMF subsystem to protect transactions. The additional programming and
functional overhead required to do so is unnecessary.

Early Replies
You should program your servers so that when a server process reads a request message,
it completely processes the request before it replies to it. This practice is often called the
no-early-reply rule. If you do not follow this rule, the server process loses the TMF
transaction identifier it requires to finish processing a transaction. After the reply, any
further attempts to lock, write, or delete records in the same audited files will fail. In
addition, failure to follow the no-early-reply rule can create a queue of incomplete
transactions for single-threaded servers. The queue occurs when the requester, which has
been replied to prematurely, sends one more request to the server and the server

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 23
Designing Your Application Design Considerations

increases its potential queue by one request. A single-threaded server queue can result in
poorer performance for the application system.

Audited and Nonaudited Servers

If your Pathway application uses a database that is a combination of TMF audited files
and nonaudited files, write separate servers to process the two types of files. Updates to
audited files must occur within a TMF transaction; updates to nonaudited files should
not occur within a transaction, because the transaction imposes unnecessary overhead.

Use of a GDSX Back-End Process

The Extended General Device Support (GDSX) communications subsystem product,
described under Requesters Using GDSX, earlier in this section, can also be used as a
back-end process for Pathway server classes.
A GDSX back-end process receives input from multiple processes on a Tandem
NonStop system and provides access to a limited number of I/O devices. Common uses
of GDSX back-end processes include implementing communications protocols; message
switching; and coordinating access to shared resources or I/O devices, such as log files,
terminals, or remote ports.
In a back-end process, each thread of the device handler has no direct association with
an I/O device. Figure 2-5 shows a typical message-switching application in which
processes on the Tandem system send messages to the communications process on the
remote system. Here, the GDSX process might run a line-handler task to handle data on
the communications lines and a device-handler task for each server process. The device
handlers forward data to the line handler for forwarding to the remote system.

Figure 2-5. GDSX as a Back-End Process

Tandem NonStop System Remote System

Server Class

GDSX Comm
Server
Process

CDT129

For further information about designing and coding GDSX processes, refer to the
Extended General Device Support (GDSX) Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 24
Designing Your Application Server Program Structure

Server Program Structure

Example 2-2 illustrates the structure of a single-function server program written in
COBOL85. The program in the following example handles data entry for the order-
processing application introduced at the beginning of this section.
Note. The program in Example 2-2 illustrates program structure only; it is not a complete
program. For an example of a complete, running Pathway server program, refer to Appendix B,
Examples.

Example 2-2. COBOL85 Server Program Example (page 1 of 3)

IDENTIFICATION DIVISION. Declares server name.

PROGRAM-ID. ORDER-SERVER.

ENVIRONMENTAL DIVISION.

CONFIGURATION SECTION.
SPECIAL NAMES. Defines source library file
FILE “$src.srvlib.orderlib” IS COBOL-LIB. names.
FILE “$obj.srvlib.ocmgrobj” IS COMM-MGR.

INPUT-OUTPUT SECTION. Selects logical names for

FILE CONTROL. $RECEIVE and all database
SELECT MSG-IN files accessed by server.
ASSIGN TO $RECEIVE $RECEIVE is a Guardian
FILE STATUS IS RCV-STAT IN WS-RCV-INFO. file that receives and stores
messages transmitted between
SELECT MSG-OUT requesters and servers.
ASSIGN TO $RECEIVE
FILE STATUS IS RCV-STAT IN WS-RCV-INFO.

SELECT LAST-ID
ASSIGN TO $DATA.ORDER.LASTID
ORGANIZATION IS RELATIVE
ACCESS IS SEQUENTIAL
RECORD KEY IS WS-LASTID-REL-KEY
FILE STATUS IS FILE-STAT IN WS-FILE-INFO.
.
.
.

RECEIVE-CONTROL. Declares number of

TABLE OCCURS 5 TIMES concurrent opens and
SYNCDEPTH IS 1 maximum size of largest
REPLY CONTAINS 204 CHARACTERS. request message.

DATA DIVISION.

FILE SECTION. Declares record data

FD MSG-IN structures for logical file
RECORD CONTAINS 204 CHARACTERS names.
LABEL RECORDS ARE OMITTED

01 ORDER-MSG.
.
.
.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 25
Designing Your Application Server Program Structure

Example 2-2. COBOL85 Server Program Example (page 2 of 3)

FD MSG-OUT
RECORD CONTAINS 36 TO 204 CHARACTERS
LABEL RECORDS ARE OMITTED

01 ORDER-REPLY.
.
.
.
01 ERROR-STATUS-REPLY.
.
.
.
FD LAST-ID
RECORD CONTAINS 12 CHARACTERS
LABEL RECORDS ARE OMITTED

01 LAST-ID-RECORD.
02 LAST-ID PIC 9(12).
.
.
.

WORKING STORAGE SECTION. Declares data structures of

variables used by server.
PROCEDURE DIVISION.

DECLARATIVES. Declares error procedures to

be used when an I/O statement
returns an error.

MAIN SECTION. Contains main program logic.

PERFORM START. Server program begins and
PERFORM PROCESS-REQUEST ends here.
UNTIL last-requester-close.
PERFORM STOP.
STOP RUN.

START SECTION. Contains logic that opens all

OPEN INPUT msg-in. files used by server.
OPEN OUTPUT msg-out SYNCDEPTH 1.
OPEN I/O last-id SHARED SYNCDEPTH 1.
.
.
.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 26
Designing Your Application Designing Applications for Batch Processing

Example 2-2. COBOL85 Server Program Example (page 3 of 3)

PROCESS-REQUEST SECTION. Contains logic that reads
PERFORM GET-MSG IN RCV-MGR. requests in $RECEIVE, services
IF NOT last-requester-close requests, and replies to
PERFORM DO-REQUEST. requests.

DO-REQUEST.
IF function-code OF
order-check-msg = ORDER-CHECK
PERFORM DO-ORDER-CHECK
ELSE IF function-code OF
order-check-msg = ORDER-COMIT
PERFORM DO-ORDER-COMIT
ELSE PERFORM BAD-REQUEST IN ERROR-MGR.

DO-ORDER CHECK.
.
.
.

RCV-MGR SECTION. Provides $RECEIVE I/O

services.

DB-MGR SECTION. Provides disk-file I/O

services.

ERROR-MGR SECTION. Provides error-processing

services.

STOP SECTION. Contains logic that closes

CLOSE msg-in. all files used by server.
CLOSE msg-out.
CLOSE last-id.
.
.
.

Designing Applications for Batch Processing

If your Pathway application includes batch processing, consider the different needs of
this type of processing in your design.
For example, you might code a Pathsend program that takes its input from a file rather
than from a terminal, then sends requests to a server to make updates to a database. This
program could be configured as a server, thus operating as a nested server. Its input file
might be TMF protected, and the Pathsend program might make updates to it.
An application that does several updates to a database, with each update coded as a
separate TMF transaction, could be slow when it performs these updates as a batch job
rather than performing them online. For batch processing, it is usually faster to group a
number of updates in a single transaction. However, if your batch jobs are very large,
note that you should not try to group more than about one thousand updates in one TMF
transaction.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 27
Designing Your Application Designing Applications for Batch Processing

NonStop TS/MP Pathsend and Server Programming Manual– 132500

2- 28
 3 Writing Pathsend Requesters
This section explains how to write programs that use Pathsend procedure calls to make
requests to Pathway servers. These programs can be either one of the following:
• Standard requesters—programs that initiate application requests
• Nested servers—servers that act as requesters by making requests to servers in other
server classes
Nested servers are described further in Section 4, Writing Pathway Servers. However,
this section (Section 3) describes the use of Pathsend procedure calls to perform
requester functions, whether by a standard requester or a nested server. The terms
Pathsend program and Pathsend process are used to refer to any program or process that
uses Pathsend calls, whether it is a standard requester or a nested server.
Pathsend programs are Guardian requesters, as described in the Guardian Programmer’s
Guide. You can write Pathsend programs in C, C++, COBOL85, Pascal, pTAL, or TAL.
You should be familiar with the Guardian requester/server model as implemented in the
programming language you are using.
In addition, if you are using the Transaction Management Facility (TMF) subsystem,
you should be familiar with the programming guidelines and considerations for TMF
requesters, as described in the NonStop TM/MP Application Programmer’s Guide. Note
that the NonStop TM/MP Application Programmer’s Guide describes requesters that use
calls to Guardian procedures rather than Pathsend procedures. In Pathsend requesters
that use the TMF subsystem, calls to Pathsend procedures are used instead of calls to
WRITEREAD and associated Guardian procedures, and there are other differences as
described in this section.
An example of a Pathsend requester program that is also a nested server is given in
Example B-1 on page B-2.

The Pathsend Procedure Calls

Pathsend programs use six procedures that are part of the Guardian procedure library:
SERVERCLASS_SEND_
SERVERCLASS_SEND_INFO_
SERVERCLASS_DIALOG_ABORT_
SERVERCLASS_DIALOG_BEGIN_
SERVERCLASS_DIALOG_END_
SERVERCLASS_DIALOG_SEND_
Later parts of this section describe how to use these procedures in context-free and
context-sensitive Pathsend programs. Section 5, Pathsend Procedure Call Reference,
gives detailed syntax and usage considerations for the procedures.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -1
Writing Pathsend Requesters Interprocess Communication in the Pathsend
Environment

Interprocess Communication in the Pathsend

Environment
Communication between requesters and servers in the Pathsend environment differs
from communication between Guardian requesters and servers. Rather than directly
opening a particular server process, the requester opens the LINKMON process, which
in turn opens a server process selected by the PATHMON process. The LINKMON
process’s open of the server process is shared among all requesters running in the same
processor as the LINKMON process.
The communication begins as follows:
1. The requester calls a Pathsend procedure (for example, SERVERCLASS_SEND_)
to request a server-class send operation.
2. The request goes to the LINKMON process running in the same processor as the
Pathsend requester process.
3. The LINKMON process checks for a link to a server process in the server class
specified in the request. If the LINKMON process has no available link to the server
class, the actions performed depend on the settings of the PATHMON configuration
parameters NUMSTATIC and MAXLINKS. The following steps might occur:
a. The LINKMON process sends a “get-link” request to the PATHMON process
that manages the server class specified in the procedure call.
b. The PATHMON process selects a server process in the specified server class. If
necessary, it starts and initializes a new server process in the server class.
c. The PATHMON process sends back to the LINKMON process the Guardian
process name of a server process within the specified server class.
d. The LINKMON process opens the server process of that name. This open is
shared among all the Pathsend requester processes running in the same
processor as that LINKMON process; therefore, a close operation does not
necessarily occur when the communication with a particular requester process is
finished.
For more information about the NUMSTATIC and MAXLINKS parameters and
their effects on get-link requests, refer to the NonStop TS/MP System Management
Manual.
4. The LINKMON process forwards the send request to the server process.
5. When the server process replies, the LINKMON process replies to the requester
process, and the server-class send operation is complete.
The action of the PATHMON process in step 3 is called granting a link. The LINKMON
process, which requests links and provides access to the server process after the link is
granted, is called the link manager. (For SCREEN COBOL requesters, the terminal
control process (TCP) serves as the link manager.)

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -2
Writing Pathsend Requesters Basic Pathsend Programming

Basic Pathsend Programming

The simplest type of Pathsend program is a context-free program. This subsection
provides the information you need to write a context-free Pathsend program. It also
provides information common to both context-free and context-sensitive programming.
The next subsection provides information about the additional tasks required of a
context-sensitive Pathsend program.
Context-free Pathsend programming consists simply of sending messages to a server in a
server class and receiving the replies. The sending of a message to a Pathway server
class is called a server-class send operation. In a context-free program, only two
Pathsend procedures are used: SERVERCLASS_SEND_ to send the messages (and to
receive the replies if in waited mode) and SERVERCLASS_SEND_INFO_ to get
additional information about the results of the last SERVERCLASS_SEND_ call if the
call failed.
The SERVERCLASS_SEND_ procedure enables Pathsend requesters to send data to
and receive replies from a specified Pathway server class. This procedure communicates
with the LINKMON process in the processor where the Pathsend requester is running,
passing information that enables the LINKMON process to choose a link to a server
process. After a link to the server has been obtained, the requester’s data is sent directly
from the requester’s data space to the server.
Subsequent server-class send operations to the same server class might not be sent to the
same server process; therefore, the requester must provide all necessary context on each
send.
When a server process replies to the request, the LINKMON process completes the
server-class send operation. If you use waited I/O, the reply data is in the reply buffer
when the SERVERCLASS_SEND_ call is completed. If you use nowait I/O, you
complete the I/O operation with a call to the AWAITIOX procedure.
A Pathsend requester can use the SERVERCLASS_SEND_INFO_ procedure to get
detailed information about server-class send initiation and completion errors.

Programming for Failure Recovery

A Pathsend requester must provide a mechanism for failure recovery. It can do so in one
or both of the following ways:
• By using TMF level recovery through the Transaction Management Facility (TMF)
subsystem
• By providing fault tolerance through process pairs and checkpointing
TMF level recovery is the recovery of the database to a consistent state through the use
of the TMF subsystem. When a failure occurs, the TMF subsystem allows the
application to back out (abort) the entire transaction, returning the contents of the
database to the values it held when the transaction was started. The application can then
retry the transaction.
In general, TMF level recovery is recommended for use with Pathsend programs: this
method is easier and faster to program than process pairs, and it handles nonretryable

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -3
Writing Pathsend Requesters Programming for Failure Recovery

requests more smoothly. However, because the TMF subsystem guarantees only the
consistency of the database and not fault tolerance for other operations (such as
messages sent over data communications lines), your application might need to use
process pairs along with TMF level recovery.
TMF programming is described in the NonStop TM/MP Application Programmer’s
Guide. The Guardian Programmer's Guide discusses process pairs and checkpointing.
The following paragraphs provide information specific to the use of these features in a
Pathsend program.

Server Process Failures

A server process can fail for reasons that include the following: the server process calls
ABEND, a processor fails, or someone stops the process.
If the server process fails while there are outstanding SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ calls to
it, error 904 (FEScServerLinkConnect) is returned to the requester. If the server-class
send operation is protected by a TMF transaction, the requester should abort the
transaction and reissue the request or the dialog. If the request is repeatable, the
requester should just retry the request. Context-free requesters need only retry the call to
SERVERCLASS_SEND_, but context-sensitive requesters must retry the entire dialog.
If the server process fails while there are no outstanding SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ calls to it
but a dialog is in progress, the dialog is aborted and the requester receives error 929
(FEScDialogAborted) on its next call to SERVERCLASS_DIALOG_SEND_.
If the server process is in transaction mode and fails, the TMF transaction is
automatically aborted. This treatment of the TMF transaction is a feature of the TMF
subsystem.

LINKMON Limit Errors

In some cases, you can recover from a LINKMON limit error by retrying the Pathsend
procedure call. Whether a retry will work depends on the design and operating
environment of your application, including the configuration of static and dynamic links.
Static links between a LINKMON process and a PATHMON process generally persist
for some time, depending on the application and the system workload. Dynamic links
and server classes come and go more frequently, again depending on the application and
the system workload. The number of concurrent server-class send operations is very
dynamic.
Therefore, it might be appropriate to retry a call to SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ (after a
short wait) if the concurrent calls limit is exceeded. Conversely, it might not be
appropriate to retry a call if the limit for the maximum number of PATHMON processes
is exceeded.
For more information about PATHMON configuration and performance, refer to the
NonStop TS/MP System Management Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -4
Writing Pathsend Requesters Programming for Failure Recovery

Pathsend Programming and the TMF Subsystem

The Pathsend procedure calls support TMF transactions. Calling a Pathsend procedure to
start a server-class send operation propagates the current transaction identifier, if any, to
the server process. If the Pathsend procedure call fails, use of the TMF subsystem allows
your application to abort the transaction and retry the call. As an application developer,
you do not have to be concerned about the role of LINKMON processes in the
propagation of transaction identifiers.
If you use the TMF subsystem, you should check for errors after each call to the
BEGINTRANSACTION, ENDTRANSACTION, ABORTTRANSACTION, and
RESUMETRANSACTION procedures. Failure to perform these checks could cause
important parts of your application to fail. If a server error occurs during a TMF
transaction, the requester should explicitly abort the transaction (even though in some
cases, the transaction might already have been aborted). For further information,
including a list of the file-system errors that can be returned by each of the TMF
procedure calls, refer to the NonStop TM/MP Application Programmer’s Guide.
When designing and coding your application, you should pay attention to whether or not
your Pathsend requests are in transaction mode. For example, your application should
not be in transaction mode while using Pathsend procedure calls to make requests to
subsystems whose operations are not TMF protected, such as the spooler.
Instead, do the following:
1. Save the value of the transaction identifier.
2. Suspend the transaction by calling the RESUMETRANSACTION procedure with a
tag value of 0.
3. Make the subsystem request.
4. Resume the transaction by calling the RESUMETRANSACTION procedure with a
tag value equal to the saved transaction identifier.
Attempting to send to a server class configured with the TMF parameter set to OFF fails
with Pathsend error 917 (FEScServerClassTmfViolation) if the requesting process has a
current transaction. The NonStop TS/MP System Management Manual describes how to
set the TMF parameter in the PATHCOM SET SERVER command.

Fault-Tolerant Programming
The following paragraphs describe issues related to the use of Pathsend procedure calls
in fault-tolerant programs and discuss the appropriate levels of recovery that your
application can perform after takeover by a backup Pathsend process.
For Pathsend calls that are protected by the TMF subsystem, on takeover the new
primary process must determine whether the current transaction ended or aborted. If the
transaction aborted, the new primary process can retry the entire TMF transaction. That
is, the backup process can begin a new TMF transaction and reissue the Pathsend calls.
This retry mechanism relies on the TMF subsystem’s capability to back out all work that
the server process carried out on the original request.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -5
Writing Pathsend Requesters Security Issues

For Pathsend calls not protected by the TMF subsystem, the proper recovery depends on
the nature of the request:
• Retryable requests that are not protected by the TMF subsystem can be repeated
many times without adverse effect. An example of this kind of request is a request to
read a bank account balance. Requests to retrieve data from a database are retryable
requests.
For these requests, on backup takeover, the backup can simply reissue the request.
The request could be processed more than once by different server processes without
resulting in data corruption.
• Nonretryable requests that are not protected by the TMF subsystem cannot be
processed more than once without having adverse effects. An example of this kind
of request is a request to subtract $50.00 from a bank account balance.
For these requests, there is no way for the server class to detect duplicate requests;
Pathsend does not support checkpointing of Guardian sync IDs. Therefore, the
backup process cannot send the request again because the operation might be
processed more than once. Because the request cannot be safely retried, the
Pathsend process cannot ensure that the request gets processed at least once.
Because the request thread suspends while a checkpoint is in progress, checkpointing
large buffers can affect the performance of your application. You should checkpoint the
entire context of nonretryable requests, but avoid checkpointing unnecessary data: for
example, data from retryable requests or data that has not changed since the last
checkpoint.
The LINKMON process opens servers that are configured with the TMF parameter OFF
with a sync depth of 1, and I/O operations to the server process are automatically retried
if the primary process of a server process pair fails.
See the Guardian Programmer’s Guide for detailed information about checkpointing
and sync IDs.

Security Issues
There are two levels of security to consider for Pathsend processes: security at the
network level and security at the server-class level. In addition, if you are using the
Remote Server Call (RSC) product, you can provide additional security to control access
to servers.

Network Security
If your Pathsend process is to access a Pathway server class on another system, the user
ID of the PATHMON process controlling the server class has to have corresponding user
IDs and remote passwords with the following systems:
• The system where the requesting process is running
• The system where the PATHMON process is running
• The system where the server class is running

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -6
Writing Pathsend Requesters Avoiding Coded PATHMON Names

This level of security is required because the LINKMON process must be able to open
the PATHMON process (to make link requests); the LINKMON process must be able to
open the server processes (to send user requests); and the PATHMON process must be
able to open the server processes (to send startup messages). All of these opens are
performed with the PATHMON user ID.
Note. The user ID of the Pathsend process need not have remote passwords to the
PATHMON system or to the server-class system to access the server class. Moreover, the
Pathsend-process user ID need not be known on the PATHMON or server-class systems.

Server-Class Security
LINKMON processes perform authorization checks on each server-class send operation
to make sure that the user ID of the Pathsend process at the time of the send conforms to
the server class’s OWNER and SECURITY attributes. You set these attributes for server
classes at configuration time if those server classes are to be accessed by Pathsend
processes.
The NonStop TS/MP System Management Manual describes how to set the SERVER
OWNER and SERVER SECURITY parameters in PATHCOM.

RSC Client Security

Remote Server Call (RSC) workstation clients can be allowed or disallowed to
communicate with specified Pathway servers. You can set up security validation by
creating an Access Control Server (ACS). For more information about creating an ACS,
refer to the Remote Server Call (RSC) Programming Manual.

Avoiding Coded PATHMON Names

SCREEN COBOL requesters can send requests to Pathway server classes without
having to specify the name of the PATHMON process controlling the server class,
because the TCP has a default PATHMON process to send to. Pathsend processes,
however, must specify the PATHMON name of the server class to send to, because the
Pathsend procedures provide no default PATHMON name.
It is possible, however, to avoid coding PATHMON names in Pathsend programs. For
example, you can use ASSIGNs containing the PATHMON system and process name.
Or, if the Pathsend process is a Pathway server, the process can use the name of its
creator as the default PATHMON name to send to. This method of avoiding coding the
PATHMON name is reliable as long as the sending server is not associative, in which
case its creator might not be a PATHMON process.
The Pathsend program examples BREQ and PATHSRV in Appendix B, Examples, use
ASSIGNs to avoid coding PATHMON names. The PATHSRV example also uses the
creator default method just described to avoid coding the PATHMON names.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -7
Writing Pathsend Requesters Context-Sensitive Pathsend Programming

Context-Sensitive Pathsend Programming

If you are writing a context-sensitive Pathsend program, you must follow the guidelines
in the previous subsection, Basic Pathsend Programming, and also perform additional
programming tasks. This subsection describes these additional tasks and other
considerations for context-sensitive programming.
Context-sensitive Pathsend programming involves establishing a dialog between a
requester and a server process in a server class, and then sending messages within the
dialog. After the dialog is established, the same server process is used for all the
messages in the dialog; therefore, the server can retain context between send operations.
A requester starts a dialog by calling SERVERCLASS_DIALOG_BEGIN_. This
procedure returns a dialog identifier to be used on subsequent server-class send
operations, which are made by calling SERVERCLASS_DIALOG_SEND_. The
requester can use multiple calls to SERVERCLASS_DIALOG_BEGIN_, and the
resulting dialog identifiers, to engage in multiple simultaneous dialogs. As in the
context-free case, the requester can call SERVERCLASS_SEND_INFO_ after a server-
class send operation to get detailed information about send initiation and completion
errors.
Either the requester or the server can abort the dialog, but only the server can end it.
(The server aborts the dialog by returning file-system error 1 (FEEOF) in its reply; it
ends the dialog by returning file-system error 0 (FEOK).) To abort the dialog, the
requester calls SERVERCLASS_DIALOG_ABORT_. The requester calls
SERVERCLASS_DIALOG_END_ to clean up resources after the server has ended or
aborted the dialog.
As in context-free programming, the requester can perform context-sensitive server-class
send operations either waited or nowait. The requester receives an error indication if the
server process has terminated or if it has ended or aborted the dialog.
To participate in a dialog with a context-sensitive Pathway requester, a server must
perform additional tasks besides those required of all servers. These additional tasks are
described under Writing Context-Sensitive Servers in Section 4.

Using Context-Sensitive Requesters With Context-Free Servers

Context-sensitive requesters can perform single-send dialogs with Pathway servers that
are coded to be context-free. However, if these servers check for system messages and
they use the Common Run-Time Environment (CRE), they must be modified to
recognize and respond to Pathsend dialog abort system messages. Section 4, Writing
Pathway Servers, describes how to code servers to handle these system messages.

Resource Utilization
On a requester’s first SERVERCLASS_DIALOG_BEGIN_ call, an extended segment is
added to the requester’s segment space for use as a workspace. This segment can be up
to 64 KB in size. It is deallocated when the requester process is terminated.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -8
Writing Pathsend Requesters Programming for Failure Recovery

Programming for Failure Recovery

The SERVERCLASS_DIALOG_BEGIN_ procedure allows two types of TMF
protection for dialogs:
• The one-transaction-per-dialog model
• The any-transaction-per-dialog model
You use bit 14 of the flags parameter in the SERVERCLASS_DIALOG_BEGIN_
procedure call to select which model will be used.
A value of 0 for bit 14 of flags (the default) selects the one-transaction-per-dialog
model. In this model, the dialog records the transaction identifier that is current at the
time of the SERVERCLASS_DIALOG_BEGIN_ call. Subsequent Pathsend calls for
this dialog must use this transaction identifier, or the calls will fail; that is, the current
transaction at the time of all Pathsend calls for the dialog must be the same as when the
dialog was started.
With the one-transaction-per-dialog model, the transaction cannot commit—that is, calls
to ENDTRANSACTION will fail—until the server ends (not aborts) the dialog and the
requester calls SERVERCLASS_DIALOG_END_. In other words, when bit 14 of
flags is set to 0, the TMF subsystem treats a dialog like an I/O operation: the
ENDTRANSACTION operation fails until the dialog has finished. The same restriction
applies to nested servers: if a server receives a message in a dialog and then initiates a
dialog with another server, it must complete the entire initiated dialog before replying to
the message from the received dialog.
A value of 1 for bit 14 of flags selects the any-transaction-per-dialog model. In this
model, all server-class send operations within the dialog will contain the transaction
identifier that is current at the time of the send, and there are no restrictions on
ENDTRANSACTION other than those associated with calls to the WRITEREAD
procedure (as described in the NonStop TM/MP Application Programmer’s Guide).
When the one-transaction-per-dialog model is used, the dialog must be ended rather than
aborted for the transaction to be completed successfully. The requester can ask the server
to end the dialog by specifying a user-defined END REQUEST command that is
recognized by the server.
If the dialog is aborted, the requester receives error 233, FEScError, and
SERVERCLASS_SEND_INFO_ returns Pathsend error 929, FEScDialogAborted.
The file-system error returned by SERVERCLASS_SEND_INFO_ provides information
about the reason for the abort. If the server explicitly aborts the dialog, the file-system
error is FEEOF (1).
If the server abends and the dialog is not in an ended state, the dialog is aborted, and the
requester receives error 233, FEScError, and SERVERCLASS_SEND_INFO_ returns
Pathsend error 929, FEScDialogAborted.
When the one-transaction-per-dialog model is used, the current TMF transaction, if any,
is automatically aborted when the server returns an error value (FEEOF, or any other
value besides FEOK or FEContinue) in the reply. When the any-transaction-per-dialog
model is used (when bit 14 of the dialog flags is equal to 1), no automatic transaction

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3 -9
Writing Pathsend Requesters Cancellation of Server-Class Send Operations

abort occurs. In either case, the requester should abort the transaction when it receives
an error reply.
If the server is shut down by the operator, the following occurs:
• For dialogs that currently have an I/O operation outstanding, the I/O operation is
first completed, and action is then taken depending on the error value returned:
• If the I/O operation is completed with an error value of FEOK (0), then the
FEOK value is passed back to the requester along with all user data.
• If the I/O operation is completed with an error value of FEContinue (70), the
requester receives error 233, FEScError, and SERVERCLASS_SEND_INFO_
returns Pathsend error 929, FEScDialogAborted. No user data is passed back to
the requester.
• If the I/O operation is completed with any other file-system error, the requester
receives error 233, FEScError, and SERVERCLASS_SEND_INFO_ returns the
file-system error that occurred. No user data is passed back to the requester.
After completion of the I/O operation, the dialog is aborted and the LINKMON
process closes the server process. The server process should interpret the close
operation as a dialog abort.
• For dialogs that have no I/O operation outstanding, the dialog is aborted and the
LINKMON process closes the server process. The server process should interpret
the close operation as a dialog abort.
When writing requesters as process pairs with or without the TMF subsystem, note that
it is not possible to checkpoint dialogs. The dialog and the transaction, if any, are
aborted when the requester fails.

Cancellation of Server-Class Send Operations

The requester can explicitly cancel an outstanding server-class send operation if it has
used the nowait option in the Pathsend procedure call and has not yet sensed completion
of the server-class send operation through the AWAITIOX procedure. In this case, the
server receives a Pathsend dialog abort system message (as described in Section 4) even
if it has already replied to the last send operation.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3- 10
Writing Pathsend Requesters Writing Requesters That Interoperate With
NonStop TUXEDO Servers

Writing Requesters That Interoperate With

NonStop TUXEDO Servers
Pathsend requesters can call on the services of NonStop TUXEDO servers in either of
two ways:
• Directly, by making calls to the NonStop TUXEDO Application Transaction
Monitor Interface (ATMI) functions
• Indirectly, by using the Pathway to TUXEDO (PWY2TUX) translation server
To communicate directly with a NonStop TUXEDO server by calling the ATMI
functions, a Pathsend requester must also act as a NonStop TUXEDO client, as follows:
• It must be compiled as a NonStop Kernel Open System Services (OSS) process;
therefore, it must be written in a Tandem language that supports Open System
Services (such as C).
• It must link in the ATMI functions. For example, if the requester is a C program, it
can be linked to the ATMI functions by being compiled with the buildclient
command.
• It must join a NonStop TUXEDO application by calling the tpinit() ATMI function
or one of the ATMI functions that implicitly calls tpinit(), such as tpalloc() or
tpcall().
• It must follow the restrictions on Pathsend procedure calls that apply to NonStop
TUXEDO clients: namely, it cannot make nowait calls (calls with bit 15 of the
flags parameter set to 1) to the SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, and SERVERCLASS_DIALOG_SEND_
procedures.
A requester using this method of interoperation can use request/response messages,
conversations, or both.
For more information about writing a NonStop TUXEDO client and calling the ATMI
functions, refer to the NonStop TUXEDO System Application Development Guide. For
the syntax of the ATMI function calls, refer to the NonStop TUXEDO System Reference
Manual.
To communicate indirectly with a NonStop TUXEDO server by using the Pathway to
TUXEDO translation server, a Pathsend requester program (or a SCREEN COBOL
requester program) must be written according to the guidelines in the NonStop TUXEDO
System Pathway Translation Servers Manual. Refer to that manual for further
information. A Pathsend requester using the Pathway to TUXEDO translation server
must use context-free Pathsend calls (calls to SERVERCLASS_SEND_); the translation
server does not support dialogs (conversations).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3- 11
Writing Pathsend Requesters Writing Requesters That Interoperate With
NonStop TUXEDO Servers

NonStop TS/MP Pathsend and Server Programming Manual– 132500

3- 12
 4 Writing Pathway Servers
This section explains how to write server programs that service requests from Pathway
requesters. Such requesters can be Pathsend requesters, SCREEN COBOL requesters, or
clients that use the Remote Server Call (RSC) product or the Pathway Open
Environment Toolkit (POET) product. Pathsend requesters are described in this manual.
SCREEN COBOL requesters are described in the Pathway/TS TCP and Terminal
Programming Guide; the SCREEN COBOL language is described in the Pathway/TS
SCREEN COBOL Reference Manual. For information about writing RSC clients, refer
to the Remote Server Call (RSC) Programming Manual. For information about writing
POET clients, refer to the Pathway Open Environment Toolkit (POET) Programming
Manual.
Pathway server programs read requests from $RECEIVE, as described in the Guardian
Programmer’s Guide and in the manuals describing Tandem programming languages.
You can write Pathway server programs in C, C++, COBOL85, Pascal, pTAL, TAL,
FORTRAN, or Extended BASIC. You should be familiar with the Guardian requester/
server model and with the $RECEIVE mechanism as implemented in the programming
language you are using.
Note. This section describes how to write Pathway servers in the Guardian environment. You
can also write a Pathway server program in the NonStop Kernel Open System Services (OSS)
environment. The basic design considerations in this section apply also to Pathway servers in
the Open System Services environment; however, additional Open System Services
programming considerations also apply. For information about these programming
considerations, refer to the Open System Services Programmer’s Guide.

If you are using the Transaction Management Facility (TMF) subsystem, you should
also be familiar with general programming guidelines and considerations for TMF
servers, as described in the NonStop TM/MP Application Programmer’s Guide.

Basic Pathway Server Programming

The simplest type of Pathsend server is a context-free server. This subsection provides
information related to writing context-free Pathway servers, as well as information that
applies to all Pathway servers. Writing Context-Sensitive Servers, later in this section,
provides information about the additional tasks required of a context-sensitive Pathway
server.
In X/Open and NonStop TUXEDO system terminology, a context-free server is called a
request/response server, and a context-sensitive server is called a conversational server.

Servers Shared by Different Types of Requesters

The protocol for Pathway server processes is essentially the same regardless of the type
of requester they work with. Therefore, Pathway servers can be used by more than one
type of requester; for example, by both Pathsend requesters and SCREEN COBOL
requesters, or by both Guardian requesters and clients from client/server environments.
If servers are used by several types of requester, the server program request and reply
formats must be consistent with that of all the requesters.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -1
Writing Pathway Servers Guardian Servers and Pathway Servers

Guardian Servers and Pathway Servers

Like a Guardian server, a Pathway server receives messages by reading the Guardian
$RECEIVE file. However, unlike a Guardian server, it does not receive its messages
directly from a requester program, but instead receives them from an intermediate
process: either a LINKMON process or a terminal control process (TCP). Whereas a
Guardian server receives open messages, I/O messages, and close messages from the
requester, a Pathway server receives all these messages from the LINKMON process or
the TCP. A Pathway server receives no information about the identity of the requester
process that initiated the communication, unless the requester provides that information
in the messages it sends.
The LINKMON process’s open of the server process is shared among all the Pathsend
requester processes running in the same processor as that LINKMON process.
Therefore, a close does not necessarily occur when the communication with a particular
requester process is finished.

Server Stop Protocol

A Pathway server must stop itself when the LINKMON process or the TCP closes it on
behalf of the last requester that has a link to it. Otherwise, the PATHMON process
considers the server process to be in the PENDING state.

Handling of Messages from $RECEIVE

COBOL85 servers always perform all of the I/O for the request message most recently
read from $RECEIVE and always reply to that message before reading another message.
For simplicity, this is the protocol servers should follow in most cases.
If you do need to write a multithreaded server or a context-sensitive server that performs
queuing of messages from $RECEIVE, you must write this portion of your server
program in a language other than COBOL85.

Pathsend Requester Failures

A Pathsend requester process can fail for reasons that include the following: the
requester process calls ABEND, a processor fails, or someone stops the process.
When a Pathsend requester process fails with outstanding I/O operations, the effect is
similar to that of a failure of any Guardian process with outstanding I/O; that is, all
outstanding I/O operations are canceled, including those for outstanding server-class
send operations. As a result, the target server process might get cancellation messages
related to the outstanding calls to SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_.
However, the target server process does not get a close message, as it would have from a
Guardian file-system open. Except in the case of a processor failure, the links that the
LINKMON process established with server processes on behalf of the Pathsend
requester are maintained, because these links are shared. Therefore, servers must
monitor cancel messages, rather than close messages, to determine when a pending reply
is no longer needed.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -2
Writing Pathway Servers LINKMON Process Failures

When a requester fails, all transactions initiated by that requester but not yet completed
at the time of the failure are automatically aborted.

LINKMON Process Failures

If a halt occurs in the processor where a LINKMON process is running, all the Pathsend
requesters in that processor fail. (See the previous subsection for details about requester
failures.) The links that the LINKMON process established with the server processes are
relinquished.

Linkage Space Considerations

If the server allocates too little space for $RECEIVE messages, system performance can
be adversely affected, and the requester can get errors at peak workload times. To avoid
this problem, the number of links specified by the server (for example, in the COBOL85
RECEIVE-TABLE OCCURS clause) should be greater than the value specified in the
SET SERVER MAXLINKS parameter in PATHCOM. The server’s specified number of
links should be large enough for all openers, including requesters from other
PATHMON environments that specify associative servers and from outside the Pathway
environment (for example, operations applications).
The default value for MAXLINKS is an unlimited number of links; therefore, to avoid
this problem, MAXLINKS must be set to a value. The error returned to a Pathsend
requester is error 905 (FEScNoServerLinkAvailable) or 923
(FEScTooManyServerLinks). The error returned to a SCREEN COBOL requester is
error 4 (link denied).

Considerations for Servers Used With SCREEN COBOL Requesters

SCREEN COBOL requesters require that the first two bytes of a server reply message
contain an integer reply-code value. For more information about reply-code values, refer
to the description of the REPLY CODE clause of the SCREEN COBOL SEND
statement in the Pathway/TS SCREEN COBOL Reference Manual.
The checkpointing requirements of the Pathway/TS terminal control process (TCP) can
be reduced significantly if TMF protected servers read outside of transaction mode—
that is, perform their read operations while no transaction is in progress—before
updating the database.
You can improve the performance of a server used by SCREEN COBOL requesters by
taking advantage of the TCP’s checkpointing strategy for TMF protected servers, as
follows:
• Do not use transaction mode for a server with read-only access to a database if the
requester displays the data before making any attempt to change it. In the event of a
failure, the read operations are retryable and fault-tolerant operation is maintained.
• Do not use transaction mode for a server that writes to an entry-sequenced logging
file in which duplicates are acceptable. In the event of a failure, the write operations
can be rewritten.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -3
Writing Pathway Servers Consideration for Servers Used With Remote Server
Call (RSC) Clients

For more information about the checkpointing strategy used by the TCP, refer to the
Pathway/TS TCP and Terminal Programming Guide.

Consideration for Servers Used With Remote Server Call (RSC) Clients
Some Remote Server Call (RSC) clients are written to check for an integer reply-code
value in the first two bytes of the server’s reply message. If the RSC client program calls
the RscSetOption function to set TMF_OPTION to either 2 (RSC_END_TRANS) or 3
(RSC_BEGIN_END_TRANS), the RSC transaction delivery process (TDP) monitors
this reply code from the server to determine whether the client should end the
transaction. The reply code is a feature that can allow optimization to reduce network
traffic. For further information about use of the RSC reply code, refer to the Remote
Server Call (RSC) Programming Manual.

Nested Servers
A Pathway server can use Pathsend procedure calls to make requests to servers in other
server classes. In such a case, the server is acting as a server with respect to the requester
that sends requests to it, but it is also acting as a Pathsend requester with respect to
another server. Servers communicating with each other in this way are known as nested
servers. Because they are Pathsend programs, nested servers must be written in C, C++,
COBOL85, Pascal, pTAL, or TAL.
If you use nested servers, you should try to prevent queues of requests from developing,
because a single-threaded server that calls Pathsend procedures waits for a response
before proceeding. Servers usually just wait for disk I/O, a high-priority activity. Waiting
for a low-priority server might tie up system resources.
For information about how to code the requester functions of nested servers (that is, how
to use the Pathsend procedure calls), refer to Writing Pathsend Requesters. An example
of a nested server, called PATHSRV, is given in Example B-2 on page B-53.

Using Context-Free Servers With Context-Sensitive Requesters

Dialogs using context-sensitive requesters are described under Context-Sensitive
Pathsend Programming in Section 3. A context-free server-class send operation (initiated
by a call to SERVERCLASS_SEND_) is, in effect, a dialog consisting of a single send
operation. Pathway servers that are coded to be context-free can participate in single-
send dialogs with context-sensitive requesters. However, if these servers check for
system messages, you must modify them so that they recognize Pathsend dialog abort
system messages, as described under Writing Context-Sensitive Servers later in this
section.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -4
Writing Pathway Servers Considerations for Servers That Use the TMF
Subsystem

Considerations for Servers That Use the TMF

Subsystem
If you are writing a Pathway server that uses the TMF subsystem for transaction
management, a number of additional programming considerations apply, related to the
following topics:
• Recommended application structure
• Writing a server to use the TMF subsystem if your application does not follow the
recommended structure
• Using audited and nonaudited files
• Record locking
• Grouping transaction operations

Recommended Structure for Applications

The recommended structure for applications that use the TMF subsystem is described in
this subsection. If your current or planned application has these characteristics,
programming the application to use the TMF subsystem is relatively straightforward;
otherwise, refer to Writing a Server to Use the TMF Subsystem, later in this section, for
a checklist of changes to make.
One process (usually the requester) coordinates all of the work required to do a single
TMF transaction; this process identifies the beginning and ending points of each
transaction. Additionally, if the server replies to a request message by indicating that it
failed to complete all of the changes, this process can either abort and abandon the
transaction or abort and retry the transaction.
The communication between requesters and servers is by standard interprocess I/O. The
requester does the send operation, and the server does the READUPDATE call for
$RECEIVE and the REPLY call. Each request message and the server’s reply to the
message is for a single transaction.
Any disk I/O request is for a single transaction. The TMF subsystem appends the
process’s current transaction identifier to each disk-request message so that the audit
trails can include the identity of the transaction responsible for each database change.
How concurrency control is performed depends on which relational database
management system (RDBMS) is being used. If the NonStop SQL/MP relational
database management system is used, concurrency control is done by means of
NonStop SQL/MP access options. Use of the access options causes NonStop SQL/MP
software to generate the appropriate TMF transactions as required; therefore, servers that
use NonStop SQL/MP databases are not required to include TMF procedure calls or
statements. For information about use of the NonStop SQL/MP access options, refer to
the NonStop SQL/MP Reference Manual. For servers that use the Enscribe database
record manager, concurrency control is done by using the Enscribe locking facilities,
and you must program the transactions by using TMF procedure calls or statements. For

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -5
Writing Pathway Servers Writing a Server to Use the TMF Subsystem

information about use of the Enscribe locking facilities, refer to the Enscribe
Programmer’s Guide.
Servers do not reply to request messages until all work for the request has been
completed. The contents of the reply message indicate the outcome of the request, which
is one of the following:
• All the work for the request was completed successfully.
• None of the work for the request was completed.
• The work for the request was only partially completed.
In the first case, the requester can commit the transaction. In the second case, the
requester can commit the transaction and then retry it. In both of these cases, the
information in the server’s reply is sufficient to ensure the integrity of the transaction.
However, if the transaction work was only partially completed, as in the third case, the
server needs to ensure that the transaction is not committed so that the incomplete work
can be backed out. To ensure transaction backout, the server should call the
ABORTTRANSACTION procedure after reading the server’s request and before
sending its reply. A call to ABORTTRANSACTION by the server does not end the
transaction—only the requester can end it—but such a call imposes the requirement that
the requester also call ABORTTRANSACTION, rather than ENDTRANSACTION,
after the requester’s reply.

Writing a Server to Use the TMF Subsystem

If, for some reason, your application does not follow the structure described in
Recommended Structure for Applications, earlier in this section, you should consult the
checklist that follows.
Writing a Pathway server to use the TMF subsystem requires that you do the following:
1. Decide which files in the database should be audited.
2. Determine any modifications that are necessary to convert the application to follow
the TMF subsystem locking rules.
3. Decide how to group sequences of the application operations into TMF transactions
(that is, units of recovery).
4. Ensure that any fault-tolerant servers respond correctly to file-system error 75:
REQUESTING PROCESS HAS NO CURRENT-TRANSACTION IDENTIFIER.
A backup server that takes over in mid-transaction does not have a current-
transaction identifier to send to the disk process; therefore, the disk process returns
error 75 to the server, which passes the error to the requester. If the requester aborts
and retries the transaction, the new request has a current-transaction identifier. The
preferred solution is to change the servers so they are not fault-tolerant servers.
5. Determine whether any new transaction deadlock situations are introduced as a
result of TMF implicit record locking and modify the application to avoid the
deadlock. (Transaction deadlock is a situation in which two transactions cannot

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -6
Writing Pathway Servers Using Audited and Nonaudited Files

proceed because they are each waiting for the other to release a lock.) One way to
cope with deadlock is to use timeout.
In addition, your requester program must use the necessary transaction-control
procedure calls or statements to begin and end the transaction and to abort or restart the
transaction if necessary.
Note. Whenever your server begins work on a new queued message, it must call the
ACTIVATERECEIVETRANSID procedure to change the current transaction identifier, as
described in the NonStop TM/MP Application Programmer’s Guide.

Using Audited and Nonaudited Files

TMF recovery strategy involves backing out the aborted transaction changes; backing
out those changes enables the transaction to be reexecuted from the beginning (with a
new transaction identifier). This strategy means that if you decide to have a mixture of
audited and nonaudited files in the database, you must be careful: only changes to
audited files are backed out. If a transaction works on a mix of audited and nonaudited
files, the operations on the nonaudited files must be retryable.
A retryable operation is an operation that can be interrupted and repeated an indefinite
number of times without affecting the consistency of the database; for example, all
reading operations are retryable. Whether or not a writing operation (on a nonaudited
file) is retryable depends on your criteria for consistency of the data in the database. If
the transaction changes both audited and nonaudited files, you should analyze the
transaction to determine whether backing it out and reexecuting it affects consistency.
For example, consider a transaction that extracts records from a database, computes
some aggregates like averages or means, and then uses the aggregates to extract a subset
of the extracted records from the database for summary reporting. This transaction can
be implemented by doing the extraction twice, the first time to compute the aggregates
and the second time to extract the subset. You can place the extracted records in a
nonaudited scratch file (each server can have its own scratch file, to avoid conflict
among them). If the transaction is aborted and restarted, the transaction starts writing the
scratch file from the beginning and there is no need for the scratch file to be audited.
Another example is logging all input messages to a server, which allows examination of
them after a failure. It is self-defeating to designate the log file as an audited file; the
message that caused the failure would be backed out.
The following restrictions apply to applications that use the TMF subsystem:
• When working with audited files, do not use the COBOL85 feature that provides
record blocking on an unstructured disk file.
• When changing audited files, do not use a server that provides its own record
blocking, its own record caching, or its own form of record locking.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -7
Writing Pathway Servers Locking Records

Locking Records
If your application uses the TMF subsystem, your servers must follow the TMF locking
rules. Locking gives the TMF subsystem the control required to ensure that transactions
are presented with a consistent view of the database. With respect to the locking of
records, you must consider the following aspects of your application:
• Repeatable reads.
• Errors that result from locks being held by the transaction identifier instead of the
process identifier and OPENID of the file opener.
• Errors that result from reading deleted records.
• Batch updates by a transaction that acquires a large number of locks. You should use
file locks instead of record locks for batch updating.
For details about how to handle these aspects of your application, refer to the
NonStop TM/MP Application Programmer’s Guide.

Grouping Transaction Operations

Your application can view the transaction as a logical unit of work; for example, the
order header and all of the detail items in a purchase order might be such a work unit.
The TMF subsystem, however, treats the transaction as a physical unit of recovery.
When you use the TMF subsystem in your application, you must consider this
difference.
Basically, you need to answer certain questions. What is the logical unit of work that
you want to accomplish within an application? How can the work be divided into a
number of transactions that can be recovered by the TMF subsystem?
Factors that influence the answers to these questions are:
• Concurrency: How long will record locks be held by a transaction?
• Performance: How much server activity and how many display screens are involved
in the choice of one conversion strategy over another?
• Consistency: Are the units of recovery large enough to ensure that your criteria for
consistency will be maintained?
In view of these factors, two guidelines can help you decide how to group the database
accesses made by an application into a single transaction:
• Any group of accesses that together modify the database from one consistent state to
another consistent state should be a single TMF transaction.
• Any group of accesses that require a consistent view of the database should be a
single TMF transaction.
The following examples demonstrate how you might apply these guidelines.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -8
Writing Pathway Servers Grouping Transaction Operations

Example 1
Some logical transactions do not have to be identified as TMF transactions. For
example, a logical transaction locates a single record and displays the record contents.
Because this transaction changes nothing in the database, it does not affect consistency
and does not have to be a TMF transaction.

Example 2
A data-entry transaction with a group of accesses that insert new data into the database
should be a TMF transaction. For example, a logical transaction records receipt of some
items for a stockroom by accepting the stock codes and quantity received from a data-
entry operator and then updates the records (in an audited file) for the items.
Because the first guideline applies, you should arrange to begin a TMF transaction after
the data is accepted and to end the transaction after the last record is updated. The TMF
subsystem ensures that all changes resulting from the one operator entry either are
permanent or are backed out in case the transaction aborts. Note that because any change
to an audited file requires a transaction identifier, this example is also true if the
transaction inserts only one record in the file.

Example 3
An update transaction should be a TMF transaction. For example, assume a logical
transaction does the following:
1. Accepts a specification from the operator
2. Performs the equivalent of an inquiry operation to find the data that will be updated
3. Releases the locks obtained for the inquiry
4. Displays the data for the operator
5. Accepts modifications to the displayed data (saving a copy of the original displayed
data)
6. Performs the inquiry a second time
7. Verifies that the results of the first inquiry and the second inquiry are the same
8. Writes the modified record to the database
The transaction should be implemented as two TMF transactions. The first should begin
after the data is accepted and should end (rather than release the locks) after the last
record is read. The second should begin after the modifications to the displayed data
have been accepted and should end after the last modified record is written to the
database. If the inquiry part of the transaction is just a single read, however, there is no
need for the first inquiry to be part of a TMF transaction.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4 -9
Writing Pathway Servers Servers as Process Pairs

Servers as Process Pairs

When you are using the TMF subsystem, you should not write your server processes as
fault-tolerant process pairs. The additional programming and functional overhead
required to do so is unnecessary. If your servers are already coded as process pairs,
however, it is not usually necessary to change them back to ordinary servers. For such
servers, note that if the primary server process fails, the backup process (on takeover)
does not have a current-transaction identifier. This means that the server process receives
error 75: NO CURRENT-TRANSACTION IDENTIFIER on the first I/O request to an
audited file. You should insert code into such a server either to recognize this error and
report it as a failure to the requester, or to terminate when it receives this error.
Because the COBOL85 run-time library recognizes the PARAM named NONSTOP, you
can prevent a COBOL85 server from running as a process pair by having a PARAM
NONSTOP OFF in effect when the server is started. For Pathway servers, you can
accomplish this task by including PARAM NONSTOP OFF with the parameters in the
definition of the server class during PATHMON configuration. When PARAM
NONSTOP OFF is in effect when the server is started, the COBOL85 run-time library
ignores the STARTBACKUP and CHECKPOINT verbs and stores the successful
completion code in the PROGRAM-STATUS special register.
There is no comparable mechanism to indicate that a server process coded in TAL,
pTAL, or FORTRAN should not run as a process pair. If you have server processes in
these languages that are coded as process pairs, you can either implement a custom
PARAM or recode your server.

Transaction Deadlocks
An application that uses the TMF subsystem might hold more record locks and hold
them longer than it would without the TMF subsystem because:
• Implicit locks are held on the keys of deleted records.
• Implicit locks are held for inserted records.
• Locks are held until the transaction is either committed or aborted and backed out.
The increased locking could cause new possibilities for transaction deadlock. If
transaction deadlock might become a problem, consider implementing the methods for
coping with deadlock discussed in the Guardian Programmer’s Guide.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 10
Writing Pathway Servers Considerations for Debugging Pathway Servers

Considerations for Debugging Pathway Servers

When you are running a Pathway server in debug mode (that is, with the server class
configured with DEBUG ON) or when the server falls into debug mode because of an
error, some situations might cause errors to be returned to the requester communicating
with that server, and in some cases also might cause errors to be returned to other
requesters. Those situations are described in the following subsections.

LINKMON Process and TCP Timeouts

The LINKMON process and the TCP have a built-in five-minute timeout for process
opens and I/O operations such as get-link requests. During normal operation, such
operations are generally completed in less than five minutes unless there is a problem.
However, if the server process is in debug mode, the server can be stopped (for example,
at a breakpoint) for longer than five minutes. In certain cases, this can cause a process
open or I/O operation from a LINKMON process or TCP to time out. The effects of the
timeout depend on what type of operation timed out.
If an I/O operation to the PATHMON process times out, the LINKMON process or the
TCP behaves as if there were a problem with the PATHMON process: that is, it shuts
down communication with the PATHMON process, relinquishing all links granted by
that PATHMON process.
In this situation, if the requester that sent the request that timed out was a Pathsend
requester, this request (as well as all other server-class send requests queued under that
PATHMON process) fails with Pathsend error 902, FEScPathmonConnect, with
file-system error 40 (timeout error). Subsequent send requests to server classes under
that PATHMON process fail with Pathsend error 915, FEScPathmonShutDown.
If the timed-out request came from a SCREEN COBOL requester, this request (as well
as all other server-class send requests queued under that PATHMON process) fails with
a SEND error with a TERMINATION-STATUS value of 18 (PATHMON I/O error) and
a TERMINATION-SUBSTATUS value of 40. Subsequent send requests to server classes
under that PATHMON process fail with a SEND error with a TERMINATION-STATUS
value of 18 (PATHMON I/O error) and a TERMINATION-SUBSTATUS value of 40.
If a timeout occurs on the server process open, the LINKMON process or the TCP
returns the link to the PATHMON process. If other links to the server class exist, no send
requests fail as a result of the open failure. If no other links exist, the send request (as
well as all other server-class send requests queued for that LINKMON process or TCP)
fails. A Pathsend requester receives Pathsend error 904, FEScServerLinkConnect, with
file-system error 40 (timeout error). A SCREEN COBOL requester receives a SEND
error with a TERMINATION-STATUS value of 12 (I/O error) and a TERMINATION-
SUBSTATUS value of 40 (timeout error).
Note also that if the server class is configured so that it can be opened more than once
(that is, with MAXLINKS greater than 1), a LINKMON process or TCP can attempt to
open a server process in that server class at any time. This open attempt can time out if
the server process is being debugged.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 11
Writing Pathway Servers PATHMON Process Timeouts

When Pathsend error 915 or a SEND error with TERMINATION-STATUS 12 occurs

with a timeout error during server debugging, use the PATHCOM STATUS PATHMON
command to find the server classes that are in the LOCKED state. Identify the server
program file for each locked server class, and issue the TACL command STATUS *,
PROG object-file-name to list all running processes. Stop these processes by
using the TACL STOP command. The PATHMON process then unlocks the server class,
the LINKMON process or the TCP completes the shutdown logic, and error 915 is no
longer returned.
The LINKMON process now has no more communication with that PATHMON
process. When a subsequent send request comes in for a server controlled by that
PATHMON process, the LINKMON process or TCP opens the PATHMON process, and
waits to be opened by that PATHMON process in turn, before initiating any get-link
requests to that PATHMON process.

PATHMON Process Timeouts

Similarly, the PATHMON process has a built-in five-minute timeout for I/O operations,
such as the opening of a a server process and the sending of a startup message to it that
can occur as a result of a get-link request. If such an operation takes longer than five
minutes (for example, because the server process is in debug mode), the get-link request
from the LINKMON process or the TCP times out. A Pathsend requester receives
Pathsend error 902, FEScPathmonConnect, with file-system error 40 (timeout error).
A SCREEN COBOL requester receives a SEND error with a TERMINATION-STATUS
value of 18 (PATHMON I/O error) and a TERMINATION-SUBSTATUS value of 40
(timeout error).
By default, the COBOL85 and C run-time libraries handle initialization automatically
before the first line of server code is executed. Therefore, PATHMON timeout errors on
server initialization are most likely to occur when you are debugging servers written in
TAL or pTAL, or other servers that are explicitly programmed to monitor system
messages such as open messages.

Server Timeouts
Another kind of timeout problem that can occur for servers being debugged is related to
the configured timeout value for the server class (set in the PATHCOM SET SERVER
TIMEOUT command). This timeout covers only the time taken by the I/O to the server
process. If a timeout occurs, the LINKMON process or the TCP cancels the send
operation, and the Pathsend call or SCREEN COBOL SEND request fails. Changing the
value of the timeout parameter during debugging might help prevent this problem from
occurring.

Avoiding Timeout Errors

A good way to avoid most timeout errors during the debugging of servers is to write an
Inspect command file that automatically continues server operation before a timeout
occurs.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 12
Writing Pathway Servers Writing Context-Sensitive Servers

Writing Context-Sensitive Servers

If you are writing a context-sensitive Pathway server, you must follow the guidelines in
Basic Pathway Server Programming earlier in this section and also perform additional
programming tasks. This subsection describes these additional tasks and other
considerations for programming context-sensitive servers.
When you use context-sensitive Pathway servers, the requester and server must be
designed to work together. The context-sensitive Pathsend procedure calls used by the
requester, described in Section 3, Writing Pathsend Requesters, and Section 5, Pathsend
Procedure Call Reference, convey dialog information to the server, and the server
conveys dialog information in its reply.
Note. The SCREEN COBOL language does not support context sensitivity; therefore, to take
advantage of context sensitivity, you must use either a Pathsend requester or a GDSX
front-end process that uses the GDSX pseudo Pathsend procedures.

Functions of a Context-Sensitive Server

In addition to the functions performed by all Pathway servers, a context-sensitive server
must do the following:
• Detect a newly established dialog
• Receive, service, and reply to messages associated with a dialog
• Correlate messages with a dialog
• Continue a dialog
• Terminate a dialog
• Abort a dialog
• Detect an aborted dialog
It is simpler to code a context-sensitive server if you allow only one dialog at a time.
To impose this restriction, you must configure the server class with a MAXLINKS value
of 1. If MAXLINKS is set to a value other than 1, you must code your server to save
multiple dialog contexts and to switch context, if needed, on each incoming request.
When a server receives a message on $RECEIVE, it checks dialog flag bits <12:13>
returned by the file-system procedure FILE_GETRECEIVEINFO_ or the Common
Run-Time Environment (CRE) procedure CRE_Receive_Read_ to determine whether
this is the first message of a new dialog or a message within an existing dialog. A value
of zero in these two bits indicates a context-free send operation. In addition, the server
can check dialog flag bit 14 to determine the model used by the requester for associating
transactions with dialogs.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 13
Writing Pathway Servers Detecting a Newly Established Dialog

For the syntax of the FILE_GETRECEIVEINFO_ procedure, refer to the Guardian

Procedure Calls Reference Manual. For the syntax of the CRE_Receive_Read_
procedure, refer to the Common Run-Time Environment (CRE) Programmer’s Guide.
Note. CRE_Receive_Read_ and the other CRE routines are callable only from TAL, pTAL,
FORTRAN, and COBOL85 programs and are used when you are writing mixed-language
programs. You cannot use CRE routines along with Guardian procedure calls. For more
information about the use of the CRE routines, refer to the Common Run-Time Environment
(CRE) Programmer’s Guide.

Detecting a Newly Established Dialog

A context-sensitive server detects a newly established dialog by checking bits <12:13>
of the dialog flags returned by the FILE_GETRECEIVEINFO_ or the
CRE_Receive_Read_ procedure. If the value in these bits is 1, the message is the first in
a new dialog. This message corresponds to the requester’s call to the
SERVERCLASS_DIALOG_BEGIN_ procedure; this call starts the dialog and usually
also sends data.

Receiving, Servicing, and Replying to Messages in a Dialog

The server receives, services, and replies to messages in a dialog by doing the following:
1. Reads a message from $RECEIVE by using the $RECEIVE reading mechanism for
the programming language, such as a call to the Guardian READUPDATE or
READUPDATEX procedure.
2. Checks bits <12:13> of the dialog flags returned by the FILE_GETRECEIVEINFO_
or the CRE_Receive_Read_ procedure. If the value in these bits is 1, the message is
the first in a new dialog. If the value is 2, the message is for an existing dialog.
3. Checks the value of bit 14 of the dialog flags if the TMF subsystem is being used,
the server is enforcing transaction commit protection during the dialog, and the
value of bits <12:13> of the dialog flags is 1 (first message in a new dialog). If the
value of bit 14 is 1, aborts the dialog by replying with FEEOF (1).
4. Processes the message and performs the requested services.
5. Issues a reply by using the language’s $RECEIVE writing mechanism, such as a call
to the REPLY or REPLYX procedure.

Dialog Control
The server controls the dialog by means of a file-system error code it returns in its reply.
Table 4-1 shows the effect of the reply error codes.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 14
Writing Pathway Servers Receiving, Servicing, and Replying to
Messages in a Dialog

Table 4-1. Meaning of Error Codes Returned by Context-Sensitive Server in Reply

Error Returned
by Server Meaning
70 (FEContinue) Allow the dialog to continue.
0 (FEOK) End the dialog. After the server replies with this code, the associated
TMF transaction (if any) will be allowed to commit, and the server will
receive no further sends associated with this dialog.
1 (FEEOF) Abort the dialog. After the server replies with this code, there will be no
further sends associated with this dialog.
Any other value1 Abort the dialog. After the server replies with this code, there will be no
further sends associated with this dialog.
The requester receives this error as the file-system error of the
FEScServerLinkConnect error.
1Use of other error values is not recommended because of their effect on the link to the server. Refer to
accompanying text for details.

Use of error values other than 1 (FEEOF) is not recommended. Any other error value
causes the LINKMON process to close the link to the server and return it to the
PATHMON process. If there are no other links to that server process, the server process
is deleted; then, if that process is later needed to service subsequent requests, it will need
to be re-created, thereby affecting system performance. It is recommended, therefore,
that servers always reply with FEEOF to abort a dialog. The server should return any
additional information in the message itself rather than in the reply code.
Whether the current TMF transaction, if any, is also aborted when an error value is
returned depends on the setting of dialog flag bit 14 in the call to
SERVERCLASS_DIALOG_BEGIN_, as explained in the following subsection.

Use of TMF Transactions With Dialogs

A context-sensitive server participates in a TMF transaction in the same manner as
context-free servers would do. The server inherits the transaction identifier of the
received message. The server can abort the transaction when it has the transaction
identifier.
When the server calls FILE_GETRECEIVEINFO_ or CRE_Receive_Read_ and detects
the first message in a dialog, it can also check the dialog flags used to initiate the dialog.
If you want to enforce transaction commit protection during the dialog—that is, ensure
that the requester cannot call ENDTRANSACTION until the server has ended the
dialog—you can code your server to verify that bit 14 of the dialog flags is 0 whenever
the value of bits <12:13> is 1. If the requester’s value for bit 14 is 1, the server can abort
the dialog by replying to the message with an error value of FEEOF (1).
When the one-transaction-per-dialog model is used (when bit 14 of the dialog flags is
equal to 0), the server must end the dialog by replying FEOK to a message before the
requester can commit the associated TMF transaction. To allow requesters to ask the
server to end the dialog, you can program your server to recognize a user-defined “end
request” command.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 15
Writing Pathway Servers Correlating Messages With a Dialog

When the one-transaction-per-dialog model is used, the transaction is automatically

aborted when the server returns an error value (FEEOF, or any other value besides
FEOK or FEContinue) in the reply. When the any-transaction-per-dialog model is used
(when bit 14 of the dialog flags is equal to 1), no automatic transaction abort occurs. In
either case, the requester should abort the transaction when it receives an error reply.

Correlating Messages With a Dialog

After a message with a “dialog begin” indication has been received, all messages
received on that open of $RECEIVE will pertain to that dialog until one of the following
occurs:
• The server terminates the dialog.
• A Pathsend dialog abort system message (system message -121) is received,
indicating that the dialog has been aborted. To receive Pathsend dialog abort
messages, the server must be monitoring system messages of this type. For more
information about Pathsend dialog abort messages, refer to “Detecting an Aborted
Dialog” later in this section.
• A path error (such as CPU down or network down) occurs on the link to the
LINKMON process.

Continuing a Dialog
To indicate that the dialog should continue, the server replies to a message in the dialog
with an error value of FECONTINUE (70).

Aborting a Dialog
The server can abort a dialog by replying to any message in the dialog with an error
value of FEEOF (1).

Terminating a Dialog
After the dialog has started, either the requester or the server can abort it, but only the
server can end it. To end the dialog, the server replies to a message in the dialog with an
error value of FEOK (0).

Detecting an Aborted Dialog

When a dialog is aborted, either explicitly or by termination of the requester, the server
receives a Pathsend dialog abort system message (system message -121), if the server is
monitoring system messages of this type. Context-sensitive Pathsend servers must
ensure that Pathsend dialog abort system messages are monitored on $RECEIVE. The
Pathsend dialog abort system message is described in the Guardian Procedure Errors
and Messages Manual.
When a server receives a Pathsend dialog abort system message, the server should call
FILE_GETRECEIVEINFO_ to obtain the file number and process handle associated
with the message. The server can then use these parameters to identify the dialog. The
server should reply to the Pathsend dialog abort system message with an error value of
NonStop TS/MP Pathsend and Server Programming Manual– 132500
4- 16
Writing Pathway Servers Writing Pathway Servers That Interoperate With
TUXEDO Requesters

either FEOK (0) or FEEOF (1); these values direct the LINKMON process to release the
link for re-use.
If a processor or network failure occurs, it is possible to have a dialog abort but not
receive a Pathsend dialog abort message. Therefore, to monitor all aborted dialogs, your
server must also monitor CPU down, remote CPU down, and loss of communication
with network node system messages if it uses FILE_GETRECEIVEINFO_, or CLOSE
system messages if it uses CRE_Receive_Read_ or the COBOL85 or FORTRAN
language. The server should treat such messages as dialog aborts.
Servers can monitor CPU down, remote CPU down, and loss of communication with
network node system messages by calling the Guardian procedures MONITORCPUS
and MONITORNET, which are described in the Guardian Programmer’s Guide.
It is also possible to get Pathsend dialog abort messages that do not represent actual
aborted dialogs. A server can receive such a message if the requester calls
SERVERCLASS_DIALOG_BEGIN_ and then cancels the message. In that case, the
LINKMON process sends the Pathsend dialog abort message even though the server
might not have received the first message in the dialog. Similarly, if the requester calls
SERVERCLASS_SEND_ with the nowait option and then cancels the operation before
being notified of completion of the operation through the AWAITIOX procedure, the
server receives the Pathsend dialog abort message even if it has already replied to the
last send operation.
A cancel operation can also occur if the requester abends while a server-class send
operation is in progress, whether or not the send operation was invoked with the nowait
option.

Writing Pathway Servers That Interoperate With

TUXEDO Requesters
A Pathway server can service requests from TUXEDO requesters (TUXEDO clients or
TUXEDO servers acting as clients). Such requesters can make their requests to a
Pathway server in either of two ways:
• Directly, by making calls to the Pathsend procedures described in this manual
• Indirectly, by using the TUXEDO to Pathway (TUX2PWY) translation server
Only those TUXEDO requesters that are NonStop TUXEDO native System /T clients
can make direct requests by using Pathsend procedure calls. For more information, refer
to the NonStop TUXEDO System Application Development Guide.
Native System /T clients, remote TUXEDO requesters using System /Domain, and non-
native TUXEDO workstation clients can indirectly invoke the services of a Pathway
server by using the TUXEDO to Pathway translation server. If you use this translation
server, special guidelines apply, and you might need to modify the code of existing
Pathway servers. For further information, refer to the NonStop TUXEDO System
Pathway Translation Servers Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 17
Writing Pathway Servers Writing Pathway Servers That Interoperate With
TUXEDO Requesters

NonStop TS/MP Pathsend and Server Programming Manual– 132500

4- 18
 5 Pathsend Procedure Call
Reference
Pathsend programs use six procedures that are part of the Guardian procedure library.
This section provides the syntax and semantics of these procedure calls, preceded by
information about how to call the procedures from each of the languages supported by
Pathsend. The descriptions are in alphabetic order by procedure name.
For each procedure, the reference information includes:
• A brief description of what the procedure does
• The syntax of the procedure call (in TAL)
• Parameter definitions
• Additional usage considerations for the call, where applicable
Lengthy usage considerations that apply to several different Pathsend procedures are
grouped at the end of the section.
You invoke all of the procedures by using a function-type statement in which the
procedure returns a numeric completion code (as an integer value).
Table 5-1 lists the Pathsend procedure calls by name and gives the purpose of each one.

Table 5-1. Summary of Pathsend Procedure Calls

Procedure Purpose
SERVERCLASS_DIALOG_ABORT_ Requests that the specified dialog be aborted
SERVERCLASS_DIALOG_BEGIN_ Initiates a dialog with a server process in a server
class and sends the first message in the dialog
SERVERCLASS_DIALOG_END_ Ends the specified dialog
SERVERCLASS_DIALOG_SEND_ Initiates a send within the specified dialog
SERVERCLASS_SEND_ Initiates a context-free send operation to a server
process in the specified server class
SERVERCLASS_SEND_INFO_ Gets error information about the last call to any of
the other Pathsend procedures

The procedures whose names begin with SERVERCLASS_DIALOG_ are used for
requests to context-sensitive servers. The SERVERCLASS_SEND_ procedure is used
for requests to context-free servers. The SERVERCLASS_SEND_INFO_ procedure is
used to obtain information about both context-free and context-sensitive requests.
The method for accessing these procedures from a Pathsend program depends on the
programming language you use. You can write Pathsend programs in C, C++,

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -1
Pathsend Procedure Call Reference Calls From C or C++

COBOL85, Pascal, pTAL, or TAL. The topics that follow explain how to call the
Pathsend procedures from each of the supported programming languages.
Note. For general information about calling these and other Tandem system procedures from
programs written in various programming languages, refer to the information about accessing
Guardian procedures in the Guardian Programmer’s Guide.

None of the Pathsend procedures set the condition-code register. Therefore, language
restrictions on procedures that set this register do not apply to the Pathsend procedures.

Note. For some Pathsend procedures, some parameters in the TAL calling syntax contain an
embedded colon and are of the form name:length. These parameters are TAL parameter
pairs. For further information about parameter pairs and their use in mixed-language
programming (for example, calls to these procedures from languages other than TAL or pTAL),
refer to the TAL Programmer’s Guide.

Calls From C or C++

To invoke any of the procedures from within a C or C++ program, you execute a
statement of the following form:

error = SERVERCLASS_SEND_ ( pathmon-process-name

,pathmon-process-name-len
,server-class-name
,server-class-name-len
,message-buffer
,request-len
,maximum-reply-len
,actual-reply-len
,timeout
,flags
,scsend-op-num
,tag );

error
is an integer variable defined earlier in your data declarations.

pathmon-process-name, pathmon-process-name-len,
server-class-name, server-class-name-len, message-buffer,
request-len, maximum-reply-len, actual-reply-len, timeout,
flags, scsend-op-num, and tag
are variables defined earlier in your data declarations. The types of these variables
should be the C types that correspond to the TAL variable types specified in the
Pathsend procedure-call description later in this section. For a table of these
corresponding data types, refer to the information about mixed-language
programming in the C/C++ Programmer’s Guide.
To use the Pathsend procedures in a C program, you must first have named them in an
#include <cextdecs> preprocessor directive.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -2
Pathsend Procedure Call Reference Calls From COBOL85

For further information, refer to the C/C++ Programmer’s Guide.

Calls From COBOL85

To invoke any of the procedures from within a COBOL85 program, you execute a
statement of the following form:

ENTER "SERVERCLASS_SEND_" USING pathmon-process-name

pathmon-process-name-len
server-class-name
server-class-name-len
message-buffer
request-len
maximum-reply-len
actual-reply-len
timeout
flags
scsend-op-num
tag
GIVING error

pathmon-process-name, pathmon-process-name-len,
server-class-name, server-class-name-len, message-buffer,
request-len, maximum-reply-len, actual-reply-len, timeout,
flags, scsend-op-num, and tag
are variables defined in the WORKING-STORAGE SECTION of the DATA
DIVISION. The types of these variables should be the COBOL85 types that
correspond to the TAL variable types specified in the Pathsend procedure-call
description later in this section. For a table of these corresponding data types, refer
to the information about invoking non-COBOL routines in the COBOL85 Manual.
If the length of a string parameter is declared in a separate parameter (as in
SERVER_CLASS_SEND_), this parameter must be passed to the procedure. If the
length is declared as part of the string parameter in the form name:length (as in
SERVERCLASS_DIALOG_BEGIN_ and SERVERCLASS_DIALOG_SEND_),
the length must not be passed explicitly.

error
is an integer variable (USAGE NATIVE-2) defined in the WORKING-STORAGE
SECTION of the DATA DIVISION.
For further information, refer to the COBOL85 Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -3
Pathsend Procedure Call Reference Calls From Pascal

Calls From Pascal

To invoke any of the procedures from within a Pascal program, you execute a statement
of the following form:

error := SERVERCLASS_SEND_ ( pathmon-process-name

,pathmon-process-name-len
,server-class-name
,server-class-name-len
,message-buffer
,request-len
,maximum-reply-len
,actual-reply-len
,timeout
,flags
,scsend-op-num
,tag )

error
is a variable of type IO_Error_Number defined earlier in your data declarations.

pathmon-process-name, pathmon-process-name-len,
server-class-name, server-class-name-len, message-buffer,
request-len, maximum-reply-len, actual-reply-len, timeout,
flags, scsend-op-num, and tag
are variables defined earlier in your data declarations. The types of these variables
should be the Pascal types that correspond to the TAL variable types specified in the
Pathsend procedure-call description later in this section. For definitions of these
corresponding data types, refer to the information about mixed-language
programming and data-type correspondence in the Pascal Reference Manual.
To use the Pathsend procedures in a Pascal program, you must first have named them in
a SOURCE PEXTDECS compiler directive. For further information, refer to the Pascal
Reference Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -4
Pathsend Procedure Call Reference Calls From TAL or pTAL

Calls From TAL or pTAL

To invoke any of the procedures from within a TAL or pTAL program, you execute a
statement of the following form:

error := SERVERCLASS_SEND_ ( pathmon-process-name

,pathmon-process-name-len
,server-class-name
,server-class-name-len
,message-buffer
,request-len
,maximum-reply-len
,actual-reply-len
,timeout
,flags
,scsend-op-num
,tag );

error
is an integer variable defined earlier in your data declarations.

pathmon-process-name, pathmon-process-name-len,
server-class-name, server-class-name-len, message-buffer,
request-len, maximum-reply-len, actual-reply-len, timeout,
flags, scsend-op-num, and tag
are variables defined earlier in your data declarations, with types as specified in the
Pathsend procedure-call description later in this section.
To use the Pathsend procedures in a TAL or pTAL program, you must first have named
them in a SOURCE EXTDECS compiler directive. For further information, refer to the
TAL Programmer’s Guide.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -5
Pathsend Procedure Call Reference SERVERCLASS_DIALOG_ABORT_ Procedure

SERVERCLASS_DIALOG_ABORT_ Procedure
The SERVERCLASS_DIALOG_ABORT_ procedure aborts the specified dialog.
A call to SERVERCLASS_DIALOG_BEGIN_ to begin a dialog must be matched by a
call to SERVERCLASS_DIALOG_ABORT_ or SERVERCLASS_DIALOG_END_ at
the end of the dialog.

Syntax
The syntax of the SERVERCLASS_DIALOG_ABORT_ procedure is:

error := SERVERCLASS_DIALOG_ABORT_ ( dialog-id ); ! i

error returned value

INT
returns an error word containing one of the following values:
0 (FEOK) indicates that the call was successful.
233 (FESCErr) indicates that an error occurred. You can call the
SERVERCLASS_SEND_INFO_ procedure to get more detailed
information about the error.

dialog-id input
INT(32):value
is the dialog identifier previously returned from the
SERVERCLASS_DIALOG_BEGIN_ call that began the dialog.
This parameter is required.

Considerations
The following considerations apply to the SERVERCLASS_DIALOG_ABORT_
procedure:
• If the server has opened $RECEIVE for system message handling and is using the
Common Run-Time Environment (CRE), aborting the dialog will cause the server to
receive a system message -121, PATHSEND DIALOG ABORT.
• SERVERCLASS_DIALOG_SEND_ operations in progress within this dialog at the
time of the call to SERVERCLASS_DIALOG_ABORT_ are canceled.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -6
Pathsend Procedure Call Reference SERVERCLASS_DIALOG_BEGIN_ Procedure

SERVERCLASS_DIALOG_BEGIN_ Procedure
The SERVERCLASS_DIALOG_BEGIN_ procedure initiates a dialog with a server
process in a server class and sends the first message in the dialog.
The procedure identifies the server class to the system and returns a dialog identifier for
subsequent dialog operations. A SERVERCLASS_DIALOG_BEGIN_ call must be
matched by a SERVERCLASS_DIALOG_ABORT_ or
SERVERCLASS_DIALOG_END_ call at the end of the dialog.
The completion of this processing—that is, getting the final outcome (success or failure)
and, if successful, the reply data—occurs in one of two ways, depending on whether the
send operation is initiated as waited or nowait:
• For waited send operations, initiation and completion are both performed by the
SERVERCLASS_DIALOG_BEGIN_ procedure.
• For nowait send operations, initiation is performed by the
SERVERCLASS_DIALOG_BEGIN_ procedure, and completion is performed by
calling the AWAITIOX procedure.

Syntax
The syntax of the SERVERCLASS_DIALOG_BEGIN_ procedure is:

error := SERVERCLASS_DIALOG_BEGIN_
( dialog-id ! o
,pathmon-process-name:length !
i:i
,server-class-name:length !
i:i
,message-buffer !
i,o
,request-len ! i
,maximum-reply-len ! i
,[ actual-reply-len ] ! o
,[ timeout ] ! i
,[ flags ] ! i
,[ scsend-op-num ] ! o
,[ tag ] ); ! i

error returned value

INT

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -7
Pathsend Procedure Call Reference Syntax

returns an error word containing one of the following values:

0 (FEOK) indicates that the call was successful and the server has ended the
dialog.
70 (FEContinue) indicates that the call was successful and the server is ready
for the next message in the dialog.
233 (FESCErr) indicates that an error occurred. You can call the
SERVERCLASS_SEND_INFO_ procedure to get more detailed
information about the error.

dialog-id output
INT:ref:EXT(32):1
returns an identifier that can be used for subsequent operations on this dialog.
This parameter is required.

pathmon-process-name:length input:input
STRING:ref:EXT:*, INT:value
contains the external Guardian process name of the PATHMON process controlling
the server class (for example, $PM or \AB.$PMN). The process name portion can
have up to five characters after the dollar sign ($) if it is a local process name and up
to four characters after the dollar sign if it is a process on a remote system. The
name cannot include an optional first or second name qualifier, must be left justified
in the buffer, and can contain trailing blanks.
The integer value is the byte length of the pathmon-process-name string. This
value can range from 2 through 15.
This parameter is required.

server-class-name:length input:input
STRING:ref:EXT:*, INT:value
contains the name of the server class to send to (for example, EMP-SERVER). This
name must conform to the Pathway server-class naming rules, must be left justified
in the buffer, and can contain trailing blanks. This server-class name, along with the
pathmon-process-name, uniquely identifies a server class.
The integer value is the byte length of the server-class-name string. This
value can range from 1 through 15.
This parameter is required.

message-buffer input, output

STRING:ref:EXT:*
contains the message to send to the server class. On successful completion of the
send operation, message-buffer contains the reply from the server class.
NonStop TS/MP Pathsend and Server Programming Manual– 132500
5 -8
Pathsend Procedure Call Reference Syntax

This parameter is required.

request-len input
INT:value
is the byte length of the data contained in message-buffer. The range of
acceptable values is 0 through 32767 bytes.
This parameter is required.

maximum-reply-len input
INT:value
is the maximum number of bytes that the reply message from the server class can
contain. The range of acceptable values is 0 through 32767 bytes.
No more than maximum-reply-len bytes of the actual reply are placed into
message-buffer upon successful completion of a send.
It is not an error if the server replies with a byte count not equal to the maximum-
reply-len value specified by the requester in the call to this procedure. If the
server replies with a byte count greater than the maximum-reply-len value, the
actual bytes transferred are truncated to maximum-reply-len.
This parameter is required.

actual-reply-len output
INT:ref:EXT:1
returns a count of the number of bytes returned in the server process reply. This
parameter is for waited I/O only and can be omitted for nowait I/O. The return value
of this parameter is 0 if nowait I/O is used. For nowait I/O, the actual reply length is
returned by AWAITIOX.

timeout input
INT(32):value
specifies the maximum amount of time, in hundredths of a second, that the
LINKMON process waits for the completion of this send. This value must be either
-1D or a value greater than 0D. The default is -1D (wait indefinitely).
If there is an outstanding I/O operation to a server process when a
SERVERCLASS_DIALOG_BEGIN_ operation times out, the I/O operation is
canceled.
See Timeout Considerations for Pathsend Programming later in this section for
details about timeout for waited and nowait operations.

flags input
INT:value

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5 -9
Pathsend Procedure Call Reference Syntax

flags.<15>
with a value of 1 indicates that this operation is to be performed nowait. A value
of 0 indicates that this operation is to be performed waited. The default is 0.

flags.<14>
if set to 0, selects the one-transaction-per-dialog model. In this model, the
dialog records the transaction identifier that is current at the time of the
SERVERCLASS_DIALOG_BEGIN_ call. Subsequent
SERVERCLASS_DIALOG_SEND_, SERVERCLASS_DIALOG_ABORT_,
and SERVERCLASS_DIALOG_END_ calls that use the returned dialog-id
must specify this transaction identifier, or the calls will fail.
ENDTRANSACTION will fail unless the dialog has been ended (not aborted).
When this bit is set to 0, the TMF subsystem treats a dialog like an I/O
operation: the ENDTRANSACTION operation fails until the dialog has
finished. The same restriction applies to a nested server (a server that receives a
request and then becomes a requester to other servers): if a server receives a
message in a dialog and then initiates a dialog with another server, it must
complete the entire initiated dialog before replying to the message from the
received dialog.
A value of 1 selects the any-transaction-per-dialog model. In this model, all
server-class send operations within the dialog will contain the transaction
identifier that is current at the time of the send, and there are no restrictions on
ENDTRANSACTION other than those associated with calls to the
WRITEREAD procedure.
The default is 0.

flags.<0:13>
must be 0.

scsend-op-num output
INT:ref:EXT:1
returns the server-class send operation number. You can use the server-class send
operation number in place of the file-number parameter in calls to CANCEL,
CANCELREQ, and AWAITIOX for nowait sends, and in calls to FILEINFO for
waited and nowait sends, to indicate that the calls refer to server-class send
operations. The value of scsend-op-num is determined on the first successfully
initiated nowait send. This value is returned on every subsequent nowait send that is
initiated successfully. A value of -1 is returned for nowait sends that are not
initiated successfully. A value of -1 is always returned for waited sends.
See Server-Class Send Operation Number later in this section for more information
about the server-class send operation number.

tag input
INT(32):value

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 10
Pathsend Procedure Call Reference Considerations

is used for nowait I/O only. The tag is stored by the system and then passed back
to the application by the AWAITIOX procedure when the nowait operation is
completed. You can use the tag parameter to identify multiple nowait I/O
operations. For waited I/O, this parameter is not used and can be omitted.
The default is 0D.

Considerations
If the SERVERCLASS_DIALOG_BEGIN_ procedure fails but does not return a valid
dialog identifier, the dialog was never created. In this case, there is no need to abort the
dialog.
For additional considerations, refer to Usage Considerations for Pathsend Procedures at
the end of this section.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 11
Pathsend Procedure Call Reference SERVERCLASS_DIALOG_END_ Procedure

SERVERCLASS_DIALOG_END_ Procedure
The SERVERCLASS_DIALOG_END_ procedure cleans up resources for the specified
dialog after the server has ended it.
A call to SERVERCLASS_DIALOG_BEGIN_ must be matched by a call to
SERVERCLASS_DIALOG_ABORT_ or SERVERCLASS_DIALOG_END_ at the end
of the dialog. For the SERVERCLASS_DIALOG_END_ procedure to work correctly,
the server must previously have ended the dialog by replying with an error value other
than FEContinue (70).
This procedure does not perform any I/O operations.

Syntax
The syntax of the SERVERCLASS_DIALOG_END_ procedure is:

error := SERVERCLASS_DIALOG_END_ ( dialog-id ); ! i

error returned value

INT
returns an error word containing one of the following values:
0 (FEOK) indicates that the call was successful.
233 (FESCErr) indicates that an error occurred. You can call the
SERVERCLASS_SEND_INFO_ procedure to get more detailed
information about the error.

dialog-id input
INT(32):value
is the dialog identifier previously returned from the
SERVERCLASS_DIALOG_BEGIN_ call that began the dialog.
This parameter is required.

Considerations
None.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 12
Pathsend Procedure Call Reference SERVERCLASS_DIALOG_SEND_ Procedure

SERVERCLASS_DIALOG_SEND_ Procedure
The SERVERCLASS_DIALOG_SEND_ procedure initiates a send within the specified
dialog.
The completion of this processing—that is, getting the final outcome (success or failure)
and, if successful, the reply data—occurs in one of two ways, depending on whether the
send operation is initiated as waited or nowait:
• For waited send operations, initiation and completion are both performed by the
SERVERCLASS_DIALOG_SEND_ procedure.
• For nowait send operations, initiation is performed by the
SERVERCLASS_DIALOG_SEND_ procedure, and completion is performed by the
AWAITIOX procedure.
This procedure is similar to the context-free SERVERCLASS_SEND_ procedure, with a
few differences as described under “Considerations.”

Syntax
The syntax of the SERVERCLASS_DIALOG_SEND_ procedure is:

error := SERVERCLASS_DIALOG_SEND_
( dialog-id ! i
,message-buffer !
i,o
,request-len ! i
,maximum-reply-len ! i
,[ actual-reply-len ] ! o
,[ timeout ] ! i
,[ flags ] ! i
,[ scsend-op-num ] ! o
,[ tag ] ); ! i

error returned value

INT
returns an error word containing one of the following values:
0 (FEOK) indicates that the call was successful and the server has ended the
dialog.
70 (FEContinue) indicates that the call was successful and the server is ready
for the next message in the dialog.
233 (FESCErr) indicates that an error occurred. You can call the
SERVERCLASS_SEND_INFO_ procedure to get more detailed
information about the error.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 13
Pathsend Procedure Call Reference Syntax

dialog-id input
INT(32):value
is an identifier, previously returned from SERVERCLASS_DIALOG_BEGIN_, that
specifies the dialog for this send operation.
This parameter is required.

message-buffer input, output

STRING:ref:EXT:*
contains the message to send to the server class. On successful completion of the
send operation, message-buffer contains the reply from the server class.
This parameter is required.

request-len input
INT:value
is the byte length of the data contained in message-buffer. The range of
acceptable values is 0 through 32767 bytes.
This parameter is required.

maximum-reply-len input
INT:value
is the maximum number of bytes that the reply message from the server class can
contain. The range of acceptable values is 0 through 32767 bytes.
No more than maximum-reply-len bytes of the actual reply are placed into
message-buffer upon successful completion of a send.
It is not an error if the server replies with a byte count not equal to the maximum-
reply-len value specified by the requester in the call to this procedure. If the
server replies with a byte count greater than the maximum-reply-len value, the
actual bytes transferred are truncated to maximum-reply-len.
This parameter is required.

actual-reply-len output
INT:ref:EXT:1
returns a count of the number of bytes returned in the server process reply. This
parameter is for waited I/O only and can be omitted for nowait I/O. The return value
of this parameter is 0 if nowait I/O is used. For nowait I/O, the actual reply length is
returned by AWAITIOX.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 14
Pathsend Procedure Call Reference Syntax

timeout input
INT(32):value
specifies the maximum amount of time, in hundredths of a second, that the
LINKMON process waits for the completion of this send. This value must be either
-1D or a value greater than 0D. The default is -1D (wait indefinitely).
If there is an outstanding I/O operation to a server process when the
SERVERCLASS_DIALOG_SEND_ operation times out, the I/O operation is
canceled.
See Timeout Considerations for Pathsend Programming later in this section for
details about timeout for waited and nowait operations.

flags input
INT:value

flags.<15>
with a value of 1 indicates that this operation is to be performed nowait. A value
of 0 indicates that this operation is to be performed waited. The default is 0.

flags.<0:14>
must be 0.

scsend-op-num output
INT:ref:EXT:1
returns the server-class send operation number. You can use the server-class send
operation number in place of the file-number parameter in calls to CANCEL,
CANCELREQ, and AWAITIOX for nowait sends, and in calls to FILEINFO for
waited and nowait sends, to indicate that the calls refer to server-class send
operations. The value of scsend-op-num is determined on the first successfully
initiated nowait send. This value is returned on every subsequent nowait send that is
initiated successfully. A value of -1 is returned for nowait sends that are not
initiated successfully. A value of -1 is always returned for waited sends.
See Server-Class Send Operation Number later in this section for more information
about the server-class send operation number.

tag input
INT(32):value
is used for nowait I/O only. The tag is stored by the system and then passed back
to the application by the AWAITIOX procedure when the nowait operation is
completed. You can use the tag parameter to identify multiple nowait I/O
operations. For waited I/O, this parameter is not used and can be omitted.
The default is 0D.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 15
Pathsend Procedure Call Reference Considerations

Considerations
The SERVERCLASS_DIALOG_SEND_ procedure is similar to the context-free
SERVERCLASS_SEND_ procedure, with the following differences:
• The dialog identifier, obtained from the SERVERCLASS_DIALOG_BEGIN_ call
that started the dialog, is used to identify the dialog, which is associated with a
particular server class.
• The SERVERCLASS_DIALOG_SEND_ call fails if the current transaction
identifier does not match the transaction identifier used for the
SERVERCLASS_DIALOG_BEGIN_ call, unless this feature has been overridden
by setting the flags.<14> bit to 1 in the call to
SERVERCLASS_DIALOG_BEGIN_.
• The send fails if there is already an outstanding send in the dialog from a previous
nowait call to this procedure.
For additional considerations, refer to Usage Considerations for Pathsend Procedures at
the end of this section.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 16
Pathsend Procedure Call Reference SERVERCLASS_SEND_ Procedure

SERVERCLASS_SEND_ Procedure
The SERVERCLASS_SEND_ procedure initiates a context-free send operation to a
server process in the specified server class.
The completion of this processing—that is, getting the final outcome (success or failure)
and, if successful, the reply data—occurs in one of two ways, depending on whether the
send operation is initiated as waited or nowait:
• For waited send operations, initiation and completion are both performed by the
SERVERCLASS_SEND_ procedure.
• For nowait send operations, initiation is performed by the SERVERCLASS_SEND_
procedure, and completion is performed by calling the AWAITIOX procedure.

Syntax
The syntax of the SERVERCLASS_SEND_ procedure is:

error := SERVERCLASS_SEND_ ( pathmon-process-name ! i

,pathmon-process-name-len ! i
,server-class-name ! i
,server-class-name-len ! i
,message-buffer ! i,o
,request-len ! i
,maximum-reply-len ! i
,[ actual-reply-len ] ! o
,[ timeout ] ! i
,[ flags ] ! i
,[ scsend-op-num ] ! o
,[ tag ] ); ! i

error returned value

INT
returns an error word containing one of the following values:
0 (FEOK) indicates that the call was successful.
233 (FESCErr) indicates that an error occurred. You can call the
SERVERCLASS_SEND_INFO_ procedure to get more detailed
information about the error.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 17
Pathsend Procedure Call Reference Syntax

pathmon-process-name input
STRING:ref:EXT:*
contains the external Guardian process name of the PATHMON process controlling
the server class (for example, $PM or \AB.$PMN). The process name portion can
have up to five characters after the dollar sign ($) if it is a local process name and up
to four characters after the dollar sign if it is a process on a remote system. The
name cannot include an optional first or second name qualifier, must be left justified
in the buffer, and can contain trailing blanks.
This parameter is required.

pathmon-process-name-len input
INT:value
is the byte length of the pathmon-process-name string. This value can range
from 2 through 15.
This parameter is required.

server-class-name input
STRING:ref:EXT:*
contains the name of the server class to send to (for example, EMP-SERVER). This
name must conform to the Pathway server-class naming rules, must be left justified
in the buffer, and can contain trailing blanks. This server-class name, along with the
pathmon-process-name, uniquely identifies a server class.
This is a required parameter.

server-class-name-len input
INT:value
is the byte length of the server-class-name string. This value can range from
1 through 15.
This parameter is required.

message-buffer input, output

STRING:ref:EXT:*
contains the message to send to the server class. On successful completion of the
send operation, message-buffer contains the reply from the server class.
This parameter is required.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 18
Pathsend Procedure Call Reference Syntax

request-len input
INT:value
is the byte length of the data contained in message-buffer. The range of
acceptable values is 0 through 32767 bytes.
This parameter is required.

maximum-reply-len input
INT:value
is the maximum number of bytes that the reply message from the server class can
contain. The range of acceptable values is 0 through 32767 bytes.
No more than maximum-reply-len bytes of the actual reply are placed into
message-buffer upon successful completion of a send.
It is not an error if the server replies with a byte count not equal to the maximum-
reply-len value specified by the requester in the call to this procedure. If the
server replies with a byte count greater than the maximum-reply-len value, the
actual bytes transferred are truncated to maximum-reply-len.
This parameter is required.

actual-reply-len output
INT:ref:EXT:1
returns a count of the number of bytes returned in the server process reply. This
parameter is for waited I/O only and can be omitted for nowait I/O. The return value
of this parameter is 0 if nowait I/O is used. For nowait I/O, the actual reply length is
returned by AWAITIOX.

timeout input
INT(32):value
specifies the maximum amount of time, in hundredths of a second, that the
LINKMON process waits for the completion of this send. This value must be either
-1D or a value greater than 0D. The default is -1D (wait indefinitely).
If there is an outstanding I/O operation to a server process when a
SERVERCLASS_SEND_ operation times out, the I/O operation is canceled.
See Timeout Considerations for Pathsend Programming later in this section for
details about timeout for waited and nowait operations.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 19
Pathsend Procedure Call Reference Considerations

flags input
INT:value

flags.<15>
with a value of 1 indicates that this operation is to be performed nowait. A value
of 0 indicates that this operation is to be performed waited. The default is 0.

flags.<0:14>
must be 0.

scsend-op-num output
INT:ref:EXT:1
returns the server-class send operation number. You can use the server-class send
operation number in place of the file number parameter in calls to CANCEL,
CANCELREQ, and AWAITIOX for nowait sends, and in calls to FILEINFO for
waited and nowait sends, to indicate that the calls refer to server-class send
operations. The value of scsend-op-num is determined on the first successfully
initiated nowait send. This value is returned on every subsequent nowait send that is
initiated successfully. A value of -1 is returned for nowait sends that are not
initiated successfully. A value of -1 is always returned for waited sends.
See Server-Class Send Operation Number later in this section for more information
about the server-class send operation number.

tag input
INT(32):value
is used for nowait I/O only. The tag is stored by the system and then passed back
to the application by the AWAITIOX procedure when the nowait operation is
completed. You can use the tag parameter to identify multiple nowait I/O
operations. For waited I/O, this parameter is not used and can be omitted. The
default is 0D.

Considerations
Refer to Usage Considerations for Pathsend Procedures at the end of this section.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 20
Pathsend Procedure Call Reference SERVERCLASS_SEND_INFO_ Procedure

SERVERCLASS_SEND_INFO_ Procedure
The SERVERCLASS_SEND_INFO_ procedure retrieves error information about the
last SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_,
SERVERCLASS_DIALOG_SEND_, SERVERCLASS_DIALOG_END_, or
SERVERCLASS_DIALOG_ABORT_ operation that was initiated or completed with
return error 233 (FEScError) or 0 (FEOK).
If the return error from the previous Pathsend call is 0 (FEOK), both the Pathsend error
and the file-system error will always be 0, so you do not need to call
SERVERCLASS_SEND_INFO_.

Syntax
The syntax of the SERVERCLASS_SEND_INFO_ procedure is:

error := SERVERCLASS_SEND_INFO_ ( pathsend-error ! o

,file-system-error ); ! o

error returned value

INT
returns an error word. This error is associated with the call to
SERVERCLASS_SEND_INFO_ and not with the previous Pathsend call. The error
word contains one of the following file-system errors:
0 (FEOK) indicates that no errors occurred in the call to
SERVERCLASS_SEND_INFO_.
2 (FEInvalOp) is returned if the caller has an invalid segment in use. Error 2 is
also returned if the caller has no extended data segment in use and one of the
reference parameters is an extended address.
22 (FEBoundsErr) is returned if a reference parameter is out of bounds.
29 (FEMissParam) is returned if a required parameter is missing.
These errors are programming errors.

pathsend-error output
INT:ref:1
returns the Pathsend error. See Section 6, Pathsend Errors, for descriptions of
Pathsend errors.

file-system-error output
INT:ref:1
returns the file-system error. See the Guardian Procedure Errors and Messages
Manual for descriptions of file-system errors.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 21
Pathsend Procedure Call Reference Considerations

Considerations
The following considerations apply to the SERVERCLASS_SEND_INFO_ procedure:
• The condition code setting has no meaning following a call to
SERVERCLASS_SEND_INFO_.
• A call to SERVERCLASS_SEND_INFO_ before a call is ever made to
SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ results in return
error 0 (FEOK), Pathsend error 906 (FEScNoSendEverCalled), and file-system error
0 (FEOK).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 22
Pathsend Procedure Call Reference Usage Considerations for Pathsend Procedures

Usage Considerations for Pathsend Procedures

The following subsections discuss usage considerations that apply to several of the
Pathsend procedure calls.

Condition Code
The condition-code setting has no meaning following a Pathsend procedure call.

Waited I/O
The following considerations apply to waited Pathsend procedure calls:
• The tag parameter has no meaning and can be omitted.
• On a successful completion of a waited Pathsend procedure call, the
actual-reply-len parameter indicates the number of bytes in the reply.
For an example of issuing a waited call to SERVERCLASS_SEND_, see the Pathsend
server program example PATHSRV, Example B-2 on page B-53. In this COBOL85
program, paragraph 460-SEND-TO-SUBSIDIARY-SERVER performs a waited
SERVERCLASS_SEND_ call.

Nowait I/O
The following considerations apply to nowait Pathsend procedure calls:
• The maximum nowait depth for Pathsend procedure calls is 255 per process.
In other words, a Pathsend requester process can have no more than 255 outstanding
nowait server-class send operations at any one time.
• You complete a nowait Pathsend procedure call with a call to AWAITIOX.
You cannot use the AWAITIO—without the X—procedure to complete a nowait
server-class send operation. The SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, and SERVERCLASS_DIALOG_SEND_
procedures use a 32-bit extended message buffer address, and AWAITIO cannot be
used to complete extended I/O operations.
If a nowait Pathsend procedure call returns an error, the send operation was not
initiated and therefore does not need to be completed with AWAITIOX.
• The actual-reply-len parameter has no meaning for nowait server-class send
operations and can be omitted. The count of the number of bytes in the server-class
reply is returned in the count-transferred parameter of the AWAITIOX
procedure.
• After calling SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ with the nowait option, do not modify the I/O
buffer before the I/O operation is completed as indicated by AWAITIOX.
• Nowait server-class send operations must not use buffers that are currently in use for
other outstanding nowait I/O operations. The file system requires that these buffers
not be modified.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 23
Pathsend Procedure Call Reference Calls Within a TMF Transaction

• SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, and

SERVERCLASS_DIALOG_SEND_ return the server-class send operation number
in the scsend-op-num parameter. See Server-Class Send Operation Number
later in this section for more information about the server-class send operation
number.
For an example of issuing a nowait call to SERVERCLASS_SEND_, see the Pathsend
program example BREQ, Example B-1 on page B-2. In this TAL program, the
procedure Initiate^IO initiates a nowait SERVERCLASS_SEND_ call.

Calls Within a TMF Transaction

The following considerations apply when calling Pathsend procedures during a TMF
transaction:
• If SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ is called during a TMF transaction, the
transaction identifier is propagated to the server process in the same way as it is in
all interprocess communication (for example, calls to the WRITEREAD procedure).
As an application developer, you do not have to be concerned about the role of the
LINKMON process in the propagation of transaction identifiers.
• For context-sensitive server-class send operations, two types of TMF protection are
available, depending on the setting of bit 14 in the flags parameter on the call to
SERVERCLASS_DIALOG_BEGIN_. For details, refer to the discussion of
context-sensitive requesters in Section 3, Writing Pathsend Requesters.
• If a send is made to a server class that is configured with the TMF parameter set to
OFF while there is a currently active transaction identifier, the send is completed
with return error 233 (FEScError), Pathsend error 917
(FEScServerClassTmfViolation), and file-system error 0 (FEOK). See the
NonStop TS/MP System Management Manual for details about the TMF parameter
of the SET SERVER command.

Server-Class Send Operation Number

The server-class send operation number is returned in the scsend-op-num parameter
on the first successfully initiated nowait server-class send operation (through a call to
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND). The same value is returned on every subsequent
nowait send that is initiated successfully.
You can use the server-class send operation number in the following calls:
• In calls to AWAITIOX to wait for completion of any SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_
operation
• In calls to CANCEL and CANCELREQ to cancel an outstanding
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ operation
• In calls to FILEINFO

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 24
Pathsend Procedure Call Reference Server-Class Send Operation Number

A value of -1 is returned for nowait send operations that are not initiated successfully. A
value of -1 is always returned for waited send operations.
Note. Passing the server-class send operation number as a filenum parameter to Guardian
procedures other than AWAITIOX, CANCEL, CANCELREQ, and FILEINFO results in file-
system error 2 (FEInvalOp).

Calling AWAITIOX
You can use the server-class send operation number as the file number in calls to the
AWAITIOX procedure to wait for completion of any outstanding
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ operation. Alternatively, you can specify -1 as the
file number to AWAITIOX to wait for completion of any outstanding I/O operation,
including calls to SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_. When completion of one of these calls occurs,
you can identify the specific call by the tag returned by AWAITIOX. (You cannot use
the AWAITIO—without the X—procedure to complete a nowait server-class send
operation. The SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, and
SERVERCLASS_DIALOG_SEND_ procedures use a 32-bit extended message buffer
address, and AWAITIO cannot be used to complete extended I/O operations.)
For an example of one way to use AWAITIOX, see the Pathsend program example
BREQ, Example B-1 on page B-2. In this TAL example, the procedure complete^io
waits for completion of an I/O operation on any file and identifies the I/O operation by
the tag returned by AWAITIOX. You can also see how this procedure calls AWAITIOX.
If AWAITIOX returns with an error (condition code CCL), it calls FILEINFO to retrieve
the error code. If the error returned is 233 (FEScError),
SERVERCLASS_SEND_INFO_ is called to get the specific Pathsend error and
file-system error.

Canceling a Server-Class Send Operation

You cancel an outstanding SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_
operation by using the Guardian CANCEL or CANCELREQ procedure if you have used
the nowait option in the Pathsend procedure call and have not yet received notification of
completion of the server-class send operation with the AWAITIOX procedure. Note that
if the AWAITIOIX call was completed with a Pathsend error, such as a server-class send
timeout error (described in the following subsection), this is still a completion; in such
cases, you should not call CANCEL or CANCELREQ.
Calling CANCEL or CANCELREQ has the following effects on your Pathsend
program:
• The program is not affected by any reply to the SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ call.
• The program cannot determine whether the request message has been sent to a
server process, and if it was sent, whether or not it was canceled before the server
process finished processing the request.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 25
Pathsend Procedure Call Reference Server-Class Send Operation Number

• If the canceled call was to SERVERCLASS_DIALOG_BEGIN_ or

SERVERCLASS_DIALOG_SEND_, the dialog is aborted and the server receives a
Pathsend dialog abort system message, even if the server has already replied to the
server-class send operation.
• If the SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ operation was performed within a TMF
transaction, the transaction is automatically aborted.
On a call to CANCELREQ, you supply:
• The server-class send operation number as the file-number parameter.
• The tag identifying the specific operation to be canceled. (The tag is optional.)
If you use the tag parameter, the system cancels the oldest incomplete
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ call with that tag value. If you do not provide a
tag, the system cancels the oldest incomplete SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ call.
On a call to CANCEL, you supply the server-class send operation number as the
file-number parameter.
Refer to the Guardian Procedure Calls Reference Manual for descriptions of the
CANCEL and CANCELREQ procedures.
A cancel operation can also occur if the requester abends while a server-class send
operation is in progress, whether or not the send operation was invoked with the nowait
option.

Calling FILEINFO
You can use the server-class send operation number returned by
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ as the file-number parameter in calls to FILEINFO
to get the return error associated with the last waited or nowait
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ call. FILEINFO returns the same error that the
server-class procedure call returned. See the error parameter in the procedure-call
syntax descriptions earlier in this section for details about the return errors.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 26
Pathsend Procedure Call Reference Timeout Considerations for Pathsend Programming

Timeout Considerations for Pathsend Programming

When you design a Pathsend application, you can decide which of two timeout methods
that you want the LINKMON process to use:
• A server TIMEOUT attribute that applies only to the server process I/O
• A SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ parameter value that applies to the entire
processing of the send request
Because the errors for each are different, programs can differentiate between the two
kinds of timeout.

Server Timeout
You can specify a TIMEOUT attribute for your Pathway servers. You specify the server
TIMEOUT value when you configure the server. If a timeout occurs during an I/O
operation to a server, the I/O operation is canceled and any TMF transaction is aborted.
Unlike the timeout parameter to a SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ call,
described in the following subsection, the TIMEOUT attribute does not include waiting
for the server link; the TIMEOUT attribute applies only to the I/O to the server.
Server timeout returns Pathsend error 904 (FEScServerLinkConnect) and file-system
error 40 (FETimedOut).

Server-Class Send Timeout

Pathsend allows you to specify a timeout value on calls to SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, and SERVERCLASS_DIALOG_SEND_. You
can specify a different timeout value for each call. For example, if you perform
SERVERCLASS_SEND_ calls to local and remote systems, you can specify a shorter
timeout value for the local sends and a longer value for the remote send operations.
Within a dialog, if flags.<14> was set to 0 in the call to
SERVERCLASS_DIALOG_BEGIN_ and a timeout occurs, both the dialog and the
transaction (if any) are automatically aborted.
If a waited SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ request is not completed before the specified
timeout value expires, it is completed with return error 233 (FEScError), and a
subsequent call to SERVERCLASS_SEND_INFO_ returns Pathsend error 918
(FeScSendOperationAborted) and file-system error 40 (FETimedOut). If there is an
outstanding I/O operation to a server process when a SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_
operation times out, that I/O operation is canceled and the transaction, if any, is
automatically aborted.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 27
Pathsend Procedure Call Reference Timeout Considerations for Pathsend Programming

Nowait server-class send operations are completed with a call to AWAITIOX. The
timeout considerations for nowait operations are more complex, because a time-limit
parameter can also be set in the AWAITIOX call. The error returned by AWAITIOX
depends on which time limit was reached:
• If a timeout value is specified in a nowait call to SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ and
the request is not completed within the specified time, the AWAITIOX call returns
error 233, and a subsequent call to SERVERCLASS_SEND_INFO_ returns
Pathsend error 918 (FeScSendOperationAborted) and file-system error 40
(FETimedOut).
• If a time-limit value is specified in the call to AWAITIOX and this time limit is
reached, the AWAITIOX call returns file-system error 40 (FETimedOut).
If you use a time-limit value on AWAITIOX to wait on I/O operations other than the
Pathsend calls SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, and
SERVERCLASS_DIALOG_SEND_ (for example, WRITEREAD calls, which do not
provide a timeout parameter) and you also use a timeout value on the Pathsend calls,
timeouts can occur in one of the preceding two ways, depending on which timer expires
first. To avoid this complex error handling, it is recommended that you use a timeout
value only on the AWAITIOX call.
For programs whose only I/O operations are Pathsend procedure calls, it is not useful to
use both a timeout on the Pathsend procedure calls and a time limit on AWAITIOX,
because you would be timing the same operation twice. In this case, choose either
Pathsend timeouts or AWAITIOX time limits.
When an AWAITIOX time limit is reached, whether an I/O operation is canceled
depends on the filenum parameter used in the AWAITIOX call. If filenum is set
equal to scsend-op-num and the AWAITIOX time-limit value is greater than zero,
the oldest outstanding send operation is canceled. Iffilenum is set to-1 or if the
AWAITIOX time-limit value is zero or less, no operation is canceled. When a Pathsend
timeout is reached, no send operation is canceled regardless of parameter settings.
Note that any time a send is canceled, the current TMF transaction, if any, is
automatically aborted. Any time a send timeout occurs when an I/O operation is
outstanding on a server process, the I/O operation to the server is canceled and the
transaction, if any, is aborted. However, you should code the requester to call the
ABORTTRANSACTION procedure so that the appropriate cleanup is done on the
requester’s side.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

5- 28
 6 Pathsend Errors
This section describes the error codes that can be returned by the Pathsend procedure
calls. These errors can be returned by processes that call the Pathsend procedures
directly and also by software that uses these procedures internally, such as the NonStop
TUXEDO system.

Types of Errors Returned by the

Pathsend Procedures
Three errors are associated with each call to a Pathsend procedure (except
SERVERCLASS_SEND_INFO_):
• A return error
• A Pathsend error
• A file-system error
The return error is the error returned by the Pathsend procedure or by AWAITIOX. For
details about the return errors for a particular procedure call, see the error parameter
in the syntax description for that call in Section 5, Pathsend Procedure Call Reference.
For waited send operations, the return error is returned from the Pathsend procedure call.
For nowait send operations, the return error is returned from the Pathsend procedure call
if it is a send-initiation error. For errors other than send-initiation errors, the return error
is returned by the AWAITIOX procedure and can be retrieved by calling the FILEINFO
procedure.
If the return error is error 233 (FEScError), you can call
SERVERCLASS_SEND_INFO_ to retrieve the Pathsend error and the file-system error.
(If the return error is 0 (FEOK), both the Pathsend error and the file-system error are
always 0, so you do not need to call SERVERCLASS_SEND_INFO_.)
The Pathsend errors are described in this section. The file-system errors are described in
the Guardian Procedures Errors and Messages Manual; however, specific Pathsend
considerations for some of the file-system errors are given in this section in the
descriptions of the associated Pathsend errors.

Descriptions of Pathsend Errors

The Pathsend error codes are described in numeric order on the following pages.
Each description includes the following:
• The error number and corresponding error literal
• The cause of the error
• Typical file-system errors that can be encountered with the Pathsend error
• The effect of the error
NonStop TS/MP Pathsend and Server Programming Manual– 132500
6 -1
Pathsend Errors Descriptions of Pathsend Errors

• How to recover from the error

These errors can be returned to Pathsend requesters. They can also be returned to
NonStop TUXEDO clients that invoke NonStop TUXEDO request/response services or
conversational services. When Pathsend errors are returned to a NonStop TUXEDO
client, the necessary recovery actions might differ from those listed here; for information
about recovery from errors in the NonStop TUXEDO environment, refer to the NonStop
TUXEDO System Messages Manual.
For examples of how errors are returned to a Pathsend program, refer to the example
programs in Appendix B, Examples. The error literals that follow the error numbers—
for example, 233 (FEScError)—are used in the example programs.
You can recover from some errors by retrying the Pathsend procedure call; the errors
you should retry depend on the requirements of your application. The programming
examples in Appendix B, Examples, illustrate retrying the call for certain errors. You
can use those examples and the error descriptions in this section to decide how to
program for retrying errors.

900
FEScInvalidServerClassName

Cause. The server-class name specified in a call to SERVERCLASS_SEND_ or

SERVERCLASS_DIALOG_BEGIN_ is not syntactically correct. This is a
programming error.
Typical file-system error: 2 (FEInvalOp)
Effect. The send initiation fails with an error.
Recovery. Correct the server-class name syntax. See the NonStop TS/MP System
Management Manual for a description of the correct syntax for server-class names.

901
FEScInvalidPathmonName

Cause. The PATHMON process name specified in a call to SERVERCLASS_SEND_

or SERVERCLASS_DIALOG_BEGIN_ is not syntactically correct. This name must be
a valid external process name. This is a programming error.
Typical file-system error: 2 (FEInvalOp)
Effect. The send initiation fails with an error.
Recovery. Correct the PATHMON name syntax. Refer to Section 5, Pathsend
Procedure Call Reference, for details about the correct syntax for PATHMON process
names.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -2
Pathsend Errors Descriptions of Pathsend Errors

902
FEScPathmonConnect

Cause. An error has occurred in the requester’s communication with the PATHMON
process. For example, an open operation has failed, an I/O error has occurred, or the
PATHMON process has failed.
Typical file-system errors: 12, 14, 40, 48, 201, or one of the path errors between 240
and 255.
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server process.
Recovery. Recovery depends on the file-system error:
• Error 12 (FEInUse) indicates that the PATHMON process was unable to open the
LINKMON process. Specific causes of this situation include (but are not limited to)
the following:
• The maximum number of LINKMON processes that the PATHMON process
can communicate with has been exceeded. See the NonStop TS/MP System
Management Manual for information about setting the MAXLINKMONS
parameter.
• The remote password on the system where the SERVERCLASS_SEND_
request originated was not set for the system where the server class was running.
The PATHMON process receives a file-system error 48 but converts this error
into a file-system error 12. To recover, ensure that all remote passwords are
properly set, as described in the Expand Network Management Guide.
• Error 14 (FENoSuchDev) indicates the PATHMON process does not exist. Start the
PATHMON process or use an existing PATHMON process.
• Error 40 (FETimeout) indicates that a timeout error occurred, possibly because a
server was in debug mode. See Considerations for Debugging Pathway Servers in
Section 4 for more information about timeout errors for servers in debug mode.
• Error 48 (FESecViol) indicates there was a security violation. See Section 3, Writing
Pathsend Requesters, for information about network and server-class security.
• Error 201 (FEPathDown) or error 240 through 255 indicates that a path error
occurred (for example, the PATHMON process failed). Restart the PATHMON
process.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -3
Pathsend Errors Descriptions of Pathsend Errors

903
FEScPathmonMessage

Cause. The LINKMON process received an unrecognizable message from the

PATHMON process while processing a SERVERCLASS_SEND_ or
SERVERCLASS_DIALOG_BEGIN_ request. You might be using incompatible
versions of the LINKMON and PATHMON processes, or this could be a LINKMON or
PATHMON process internal error.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server process.
Recovery. This is a nonrecoverable error.

904
FEScServerLinkConnect

Cause. An error has occurred with the link to the server. For example, an open
operation has failed or there is an I/O problem. This error could occur on a call to
SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_.
Typical file-system errors: 14, 40, 48, 201, or one of the path errors between 240
and 255.
Effect. The call is completed with an error. The message might or might not have been
sent to the server process, depending on the file-system error.
If the file-system error is a path error, any transaction associated with the call is aborted.
If this error is returned from SERVERCLASS_DIALOG_BEGIN_ or
SERVERCLASS_DIALOG_SEND_, the dialog is ended on the server side.
Recovery. If this error is returned from SERVERCLASS_SEND_, recovery depends on
the file-system error:
• Error 14 (FENoSuchDev) indicates that the server process does not exist. Retry the
SERVERCLASS_SEND_ to cause the LINKMON process to use a different link to
the server process or to receive additional links from the PATHMON process. The
success of the retry depends on why the server process stopped.
• Error 40 (FETimedout) indicates that the I/O to the server process timed out because
it exceeded the configured SERVER TIMEOUT value for the server class.
• Error 48 (FESecViol) indicates there was a security violation. Section 3, Writing
Pathsend Requesters, for information about network and server-class security.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -4
Pathsend Errors Descriptions of Pathsend Errors

• Error 201 (FEPathDown) or error 240 through 255 indicates that a path error
occurred (for example, the processor where the server process was running has
failed).
If this error is returned from SERVERCLASS_DIALOG_BEGIN_ or
SERVERCLASS_DIALOG_SEND_, use SERVERCLASS_DIALOG_END_ or
SERVERCLASS_DIALOG_ABORT_ to terminate the requester’s portion of the dialog.

905
FEScNoServerLinkAvailable

Cause. The LINKMON process had no links to the server class and was unable to get a
link from the PATHMON process to satisfy this request. For example, MAXLINKS
links for each server class are already allocated to other LINKMON processes and
TCPs. This is a PATHMON configuration problem.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server process.
Recovery. Increase the maximum number of servers in the server class (with the
PATHCOM MAXSERVERS parameter) or increase the number of links available to the
server (with the PATHCOM MAXLINKS parameter). For details about the
MAXSERVERS and MAXLINKS parameters, see the NonStop TS/MP System
Management Manual.

906
FEScNoSendEverCalled

Cause. The SERVERCLASS_SEND_INFO_ procedure was called before

SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ was ever called by
this program. This is a programming error.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_INFO_ call is completed with return error 0,
Pathsend error 906, and file-system error 0.
Recovery. Ensure that your application program does not call
SERVERCLASS_SEND_INFO_ before it calls SERVERCLASS_SEND_ or
SERVERCLASS_DIALOG_BEGIN_.
Note. Error 907, FEScInvalidSegmentId, and error 908, FEScNoSegmentInUse, are not
returned on D-series systems. Error 908 has been replaced by error 912,
FEScParameterBoundsError.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -5
Pathsend Errors Descriptions of Pathsend Errors

909
FEScInvalidFlagsValue

Cause. The caller set bits in the flags parameter that are reserved and must be 0.
This is a programming error.
Typical file-system error: 2 (FEInvalOp)
Effect. The send initiation fails with an error.
Recovery. Set the reserved bits in the flags parameter to 0.

910
FEScMissingParameter

Cause. A required parameter was not supplied. This is a programming error.

Typical file-system error: 29 (FEMissParam)
Effect. The send initiation fails with an error.
Recovery. Check the syntax of the procedure call and supply the required parameters.

911
FEScInvalidBufferLength

Cause. The buffer length in the request-len or maximum-reply-len parameter

for a call to SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ is invalid. This is a programming error.
Typical file-system error: 21 (FEBadCount)
Effect. The send initiation fails with an error.
Recovery. Check the buffer lengths allowed for the request-len and maximum-
reply-len parameters in the syntax description for the procedure call, and specify the
correct buffer lengths.

912
FEScParameterBoundsError

Cause. The address specified by a reference parameter is out of bounds, or the caller
supplied a reference parameter that is an extended address but does not have an extended
segment in use. This is a programming error.
Effect. The send initiation fails with an error.
Recovery. Correct the programming error.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -6
Pathsend Errors Descriptions of Pathsend Errors

913
FEScServerClassFrozen

Cause. The server class the process tried to send to is frozen.

Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server class.
Recovery. Resend after the system manager or operator has thawed the server class.

914
FEScUnknownServerClass

Cause. The server class is not configured through the specified PATHMON process.
The program has specified an incorrect server-class name or specified the wrong
PATHMON process.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server class.
Recovery. Check the server-class name and the PATHMON process name. Or check if
the server class has been configured yet (the PATHMON-controlled objects could be in
the process of being cold started).

915
FEScPathmonShutDown

Cause. The send operation has been denied for one of the following reasons:
• The PATHMON process for the server class is shutting down.
• A timeout occurred on an I/O operation to a server in debug mode.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_ call is completed with an error. The message is
not sent to the server class.
Recovery. If the PATHMON process is shutting down, determine the reasons for the
shutdown and perform appropriate recovery actions.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -7
Pathsend Errors Descriptions of Pathsend Errors

If a server process is in debug mode and a timeout error (file-system error 40) occurred,
do the following:
• Use the PATHCOM STATUS PATHMON command to find the server classes that
are in the LOCKED state.
• Identify the server program file for each locked server class.
• Issue the TACL command STATUS *, PROG object-file-name to list all
running processes.
• Stop these processes by using the TACL STOP command.
For more information about timeout errors for servers in debug mode, refer to
Considerations for Debugging Pathway Servers in Section 4.

916
FEScServerCreationFailure

Cause. The LINKMON process was unable to get a link to the server class due to a
server creation failure. This is usually a server-class configuration problem, or a server
might have a problem that causes it to fail.
Typical file-system error: 0 (FEOK)
Effect. The send initiation fails with an error.
Recovery. Verify that the server class has been configured correctly. If the problem is
not in the configuration, correct the error in the server.

917
FEScServerClassTmfViolation

Cause. The transaction mode of the Pathsend program does not match that of the server
class. The process has a current transaction ID at the time of the send, and the server
class is configured with the TMF parameter set to OFF. This is a programming error or
a server-class configuration error.
Typical file-system error: 0 (FEOK)
Effect. The call is completed with an error. The message is not sent to the server class.
Recovery. Correct your program or change the server-class configuration setting to
TMF ON.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -8
Pathsend Errors Descriptions of Pathsend Errors

918
FEScSendOperationAborted

Cause. The send operation has been terminated at an indeterminate point.

Typical file-system error: 40 (FETimedOut)
Effect. The send fails. A message might or might not have been sent to the server
process, depending on when the send was aborted.
Recovery. The recovery action depends on which file-system error has occurred. With
error 40 (FETimedOut), you might want to try a larger timeout value.

919
FEScInvalidTimeoutValue

Cause. The caller supplied an invalid timeout value in a call to

SERVERCLASS_SEND_, SERVERCLASS_DIALOG_BEGIN_, or
SERVERCLASS_DIALOG_SEND_. This is a programming error.
Typical file-system error: 2 (FEInvalOp)
Effect. The send initiation fails with an error.
Recovery. Specify a valid timeout value. See the procedure call syntax description for
details about valid timeout values.

920
FEScPFSUseError

Cause. The caller’s process file segment (PFS) could not be accessed.
Typical file-system error: 31 (FENoBufSpace)
Effect. The send initiation fails with an error.
Recovery. Code the process to stop itself if this error occurs.

921
FEScTooManyPathmons

Cause. A call to SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_

specifies a PATHMON process not known to the LINKMON process, and the
LINKMON process is already communicating with the maximum number of
PATHMON processes allowed. The maximum number is 256.
Typical file-system error: 0 (FEOK)

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6 -9
Pathsend Errors Descriptions of Pathsend Errors

Effect. The call is completed with an error. The message is not sent to the server
process.
Recovery. In some cases, you can recover from this error by retrying the call. Whether
a retry will work depends on the design and operating environment of your application.
If the PATHMON processes in your application are frequently created and stopped, retry
the call. Otherwise, investigate the cause of the large number of PATHMON processes
and eliminate some processes.
For more information about LINKMON limits, refer to LINKMON Limit Errors on
page 3-4 and also to the NonStop TS/MP System Management Manual.

922
FEScTooManyServerClasses

Cause. The LINKMON process already has links to the maximum number of server
classes allowed for all PATHMON processes. The maximum number is 1024.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server process.
Recovery. In some cases, you can recover from this error by retrying the call. Whether
a retry will work depends on the design and operating environment of your application.
For more information about LINKMON limits, refer to LINKMON Limit Errors on
page 3-4 and also to the NonStop TS/MP System Management Manual.

923
FEScTooManyServerLinks

Cause. The LINKMON process already has the maximum number of concurrent links
to server processes allowed for all PATHMON processes. The maximum number
is 1750.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server process.
Recovery. In some cases, you can recover from this error by retrying the call. Whether
a retry will work depends on the design and operating environment of your application.
For more information about LINKMON limits, refer to LINKMON Limit Errors on
page 3-4 and also to the NonStop TS/MP System Management Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6- 10
Pathsend Errors Descriptions of Pathsend Errors

924
FEScTooManySendRequests

Cause. The maximum number of concurrent server-class send operations allowed has
been exceeded. The maximum number is 255 per requester and 512 for all requesters
running in a processor. This error can occur on a call to SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_.
Typical file-system error: 0 (FEOK)
Effect. The call is completed with an error. The message is not sent to the server
process.
Recovery. In some cases, you can recover from this error by retrying the call. Whether
a retry will work depends on the design and operating environment of your application.
For more information about LINKMON limits, refer to LINKMON Limit Errors on
page 3-4 and also to the NonStop TS/MP System Management Manual.

925
FEScTooManyRequesters

Cause. The LINKMON process is already communicating with the maximum number
of requesters allowed. The maximum number of concurrently active Pathsend
requesters per processor is 256. This error can occur only on the requester’s first call to
SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_.
Typical file-system error: 0 (FEOK)
Effect. The SERVERCLASS_SEND_ or SERVERCLASS_DIALOG_BEGIN_ call is
completed with an error. The message is not sent to the server process.
Recovery. Retry if the number of Pathsend requesters fluctuates in this processor.

926
FEScDialogInvalid

Cause. The specified dialog identifier is not valid. This error can occur on a call to
SERVERCLASS_DIALOG_SEND_, SERVERCLASS_DIALOG_END_, or
SERVERCLASS_DIALOG_ABORT_.
Typical file-system error: 0 (FEOK)
Effect. The send initiation fails with an error.
Recovery. Use a valid dialog identifier.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6- 11
Pathsend Errors Descriptions of Pathsend Errors

927
FEScTooManyDialogs

Cause. The requester cannot start a new dialog, because it already has the maximum
number of dialogs open. The maximum number of dialogs per requester is 256.
This error can occur on a call to SERVERCLASS_DIALOG_BEGIN_.
Typical file-system error: 0 (FEOK)
Effect. The send initiation fails with an error.
Recovery. Reduce the number of dialogs.

928
FEScOutstandingSend

Cause. The requester has an outstanding send operation on this dialog. This error can
occur on a call to SERVERCLASS_DIALOG_SEND_.
Typical file-system error: 0 (FEOK)
Effect. The send initiation fails with an error.
Recovery. Complete the current send before starting another.

929
FEScDialogAborted

Cause. The dialog has been aborted for one of the following reasons:
• The server requested a dialog abort.
• The server terminated between send operations.
• The server terminated immediately following send completion (with a reply of
FEContinue), but before the LINKMON process had replied to the requester.
If the server terminated, the termination could be due to a server error or a network error.
This error can occur on a call to SERVERCLASS_DIALOG_BEGIN_ or
SERVERCLASS_DIALOG_SEND_.
Typical file-system error: if the server requested the abort, 1 (FEEOF); if the server
terminated, the file-system error returned by the LINKMON process
Effect. The dialog is aborted. If flags.<14> was not set to 1 on the call to
SERVERCLASS_DIALOG_BEGIN_, the transaction is also aborted. The procedure
(if waited) or AWAITIOX (if nowait) returns with an error.
Recovery. Use SERVERCLASS_DIALOG_END_ to terminate the requester’s portion
of the dialog.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6- 12
Pathsend Errors Descriptions of Pathsend Errors

930
FEScChangedTransid

Cause. A call to SERVERCLASS_DIALOG_SEND_,

SERVERCLASS_DIALOG_END_, or SERVERCLASS_DIALOG_ABORT_ was done
under a different transaction identifier than the previous call to
SERVERCLASS_DIALOG_BEGIN_ (which had specified flags.<14> = 0). This is
a programming error.
Typical file-system error: 0 (FEOK)
Effect. The operation fails with an error.
Recovery. Use RESUMETRANSACTION to make the correct transaction identifier
current, and reissue the call that failed.

931
FEScDialogEnded

Cause. A call to SERVERCLASS_DIALOG_SEND_ failed because the server had

already ended the dialog, either by replying with FEOK or FEEOF or by terminating
while a send was still outstanding.
Typical file-system error: the file-system error of the last server response. Any number
other than 0 (FEOK) indicates one of the following:
• The dialog has been aborted.
• The previous call to SERVERCLASS_DIALOG_BEGIN_ or
SERVERCLASS_DIALOG_SEND_ failed with an error that indicated an abort,
but the requester has not yet aborted the dialog.
Effect. The procedure initiation fails with an error.
Recovery. None. Use SERVERCLASS_DIALOG_END_ to end the specified dialog or
SERVERCLASS_DIALOG_ABORT_ to abort it.

933
FEScDialogOutstanding

Cause. A call to SERVERCLASS_DIALOG_END_ was made, but the server has not
ended the dialog.
Typical file-system error: 0 (FEOK)
Effect. The procedure initiation fails with an error.
Recovery. Have the server end the dialog by replying FEOK, or use
SERVERCLASS_DIALOG_ABORT_ to abort the dialog.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6- 13
Pathsend Errors Descriptions of Pathsend Errors

934
FEScTransactionAborted

Cause. The transaction associated with the dialog has been aborted.
Typical file-system error: 0 (FEOK)
Effect. The procedure initiation fails with an error. The dialog is now aborted.
Recovery. Use SERVERCLASS_DIALOG_END_ or
SERVERCLASS_DIALOG_ABORT_ to terminate the requester’s portion of the dialog.

947
FEScLinkmonConnect

Cause. There is a problem communicating with the LINKMON process in this

processor.
Typical file-system errors: 14 (FENOSUCHDEV) and 43 (FENODISCSPACE)
Effect. The send initiation fails with an error.
Recovery. Recovery depends on the file-system error.
File-system error 14 indicates that there is no LINKMON process executing in the
processor.
File-system error 43 indicates that the LINKMON process was unable to initialize itself;
in this case, the LINKMON process writes a message to $0 that indicates the reason for
the initialization failure. Each subsequent SERVERCLASS_SEND_,
SERVERCLASS_DIALOG_BEGIN_, or SERVERCLASS_DIALOG_SEND_ call in
this processor causes the LINKMON process to reattempt initialization; after the
condition is corrected, the LINKMON process can complete initialization.
For information about initialization and limits for the LINKMON process, refer to
LINKMON Limit Errors on page 3-4 and also to the NonStop TS/MP System
Management Manual.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

6- 14
A NonStop TS/MP Limits for
Pathsend Requesters
Table A-1 lists the NonStop TS/MP product limits that apply to the Pathsend
programming environment.
For limits related to the configuration of a PATHMON environment, refer to the
NonStop TS/MP System Management Manual.

Table A-1. Limits for Pathsend Requesters

Item Limit
Buffer length Maximum of 32,767 bytes per server-class send operation
Dialogs Maximum of 256 per requester
Maximum of 1750 for all requesters in a processor
Per server, as many as the number of allowed links. The limits
on allowed links are described in the NonStop TS/MP System
Management Manual.
Links to server processes Maximum of 1750 concurrent links per processor
Nowait depth Maximum of 255 for nowait Pathsend procedure calls
PATHMON processes Maximum of 256 PATHMON processes with which a
LINKMON process can communicate
Pathsend requesters Maximum of 256 concurrently active Pathsend requesters per
processor
Server classes Maximum of 1024 server classes to which all requesters in a
processor can have outstanding links
Server-class send operations Maximum of 255 concurrent outstanding send requests per
requester
Maximum of 512 concurrent outstanding send requests per
processor
TMF transactions Maximum of 100 outstanding TMF transactions per requester.
(This limit is imposed by the TMF subsystem.)

NonStop TS/MP Pathsend and Server Programming Manual– 132500

A- 1
NonStop TS/MP Limits for Pathsend Requesters

NonStop TS/MP Pathsend and Server Programming Manual– 132500

A- 2
 B Examples
This appendix shows the source code for two example programs designed to help you
understand how to write Pathsend programs and Pathway servers. The following
examples are included:
• A context-free Pathsend requester coded in TAL
• A context-free Pathsend nested server example coded in COBOL85
Edit files containing the source code for these examples are provided on the site update
tape (SUT). The source code for the Pathsend requester is in the file BREQS; the source
code for the nested server is in the file PATHSRVS.
The examples allow you to see some of the programming concepts, described in
previous sections of this manual, put into practice. These programs, however, require
several other files provided on the SUT to run. You should read the file README
(located on the SUT) carefully before attempting to run these programs.
Note. The program examples included in this manual and on the SUT are for reference
purposes only and are intended to illustrate usage of Pathsend procedures. The program
examples are not intended to form the basis of production programs. The examples do not
demonstrate the definitive way to write Pathsend programs. There are other methods that
might be more suitable for your requirements or application.

Pathsend Requester Example

Example B-1, BREQ, is a Pathsend context-free requester program coded in TAL.
BREQ issues nowait SERVERCLASS_SEND_ calls to two server classes. The program
demonstrates how to write a Pathsend program that issues nowait
SERVERCLASS_SEND_ calls to two servers and waits for successful completion of
both send operations before committing the transaction.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 1
Examples Pathsend Requester Example

Example B-1. Context-Free Pathsend Requester Program

?PAGE "BROADCAST REQUESTER (BREQ) OVERVIEW"
! @@@ START COPYRIGHT @@@
! Tandem Confidential: Need to Know only
! Copyright (c) 1980-1985, 1987-1995, Tandem Computers Incorporated
! Protected as an unpublished work.
! All Rights Reserved.
!
! The computer program listings, specifications, and documentation
! herein are the property of Tandem Computers Incorporated and shall
! not be reproduced, copied, disclosed, or used in whole or in part
! for any reason without the prior express written permission of
! Tandem Computers Incorporated.
! @@@ END COPYRIGHT @@@
?SETTOG 1
?IFNOT 1

BREQ is a single-threaded program. It gets its input from the input

file, which contains data used to make two Pathsend sends. In
addition to doing the two sends, BREQ writes the input record to the
message log file. The two sends and the write are treated as one
logical transaction.

The processing flow is like this: read one record from the input file,
initiate the I/O nowaited, complete the I/O, and write the reply from
the two servers to the output file before reading the next record from
the input file.

An input record is made up of the following data for each Pathsend

send: either a PATHMON ASSIGN name or a PATHMON system and process
name, followed by a server class name.

The PATHMON ASSIGN names get passed to BREQ at startup. If a trace

file ASSIGN is present, then BREQ outputs a msg after it does any of
the following:

. READs one record from the input file

. Initiates a SERVERCLASS SEND
. Initiates a WRITE to the msg log file
. Executes BEGINTRANSACTION
. Executes ENDTRANSACTION
. Executes ABORTTRANSACTION
. CANCELs outstanding I/O
. After an I/O completes successfully
. After an I/O times out
. After an I/O fails
. When a transaction is retried
. When max-retries is exceeded

BREQ does a SERVERCLASS_SEND_ to the first server, a WRITE to the

message log file, and a SERVERCLASS_SEND_ to the second server. All
three I/O's are NOWAIT and initiated in this sequence.

The reply from each server named in the request msg is simply its
process id. BREQ puts together the reply for each server and writes
it to the output file.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 2
Examples Pathsend Requester Example

The following files are used by BREQ and must exist before running the
program:

INFILE -- each record is one transaction

OUTFILE -- the outcome of the transaction

ERROR LOG -- Entry sequenced; record length 132; not audited.

The physical file name is read from the ASSIGN
'ERROR-LOG-FILE'.

MESSAGE LOG -- Entry sequenced; record length 132; audited.

The physical file name is read from the ASSIGN
'MESSAGE-LOG-FILE'.

TRACE (OPTIONAL) -- Entry sequenced; record length 80; not

audited; may be a terminal, may not be a
spooler location ($S.#SOME.LOC). The physical
file name is read from the ASSIGN
'TRACE-FILE'.

The following ASSIGNS passed in:

ERROR-LOG-FILE
-- Used to record error msgs.

MESSAGE-LOG-FILE
-- A write to this file is part of a transaction. It contains
request msgs received by BREQ from DRIVER.

TRACE-FILE (OPTIONAL)
-- Used to log certain READs, SENDs, and TMF operations.

There can also be up to 50 ASSIGNs to give logical names to PATHMONs

that may be referenced in the input record.

The following PARAM is passed:

MAX-RETRIES
-- BREQ uses this number to calculate the number of times
it will try to recover from PATHSEND SEND failures.

Example run-line:

RUN BREQ/ NAME $BREQ, IN infile, OUT outfile/

The procedures on the following pages are listed in alphabetical

order.

?ENDIF 1

?PAGE "COMPILER DIRECTIVES, GLOBALS, AND EXTDECS"

?INSPECT, SYMBOLS, NOCODE
?LIST

?DATAPAGES 64

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 3
Examples Pathsend Requester Example

?PAGE "VERSION PROCEDURE DECLARATION"

?SOURCE VersProc( PS^EXAMPLE^MODULE )

?PAGE "GLOBAL DECLARATIONS FOR BREQ"

! The following are GLOBAL defines

DEFINE def = DEFINE#;

DEF lit = LITERAL#;

DEF str = STRING#;

DEF dbl = INT(32)#;

?PAGE "STRUCTS BREQ"

! These are the STRUCTS used when compiling the BREQ program.
! Comments appears to the right of the code.

?PAGE "STRUCTURE NEEDED TO STARTUP A PROCESS"

?SECTION STARTUP

! The following is used when starting up

STRUCT .CI^STARTUP (*);

BEGIN
INT MSGCODE;
STRUCT DEFAULT;
BEGIN
INT VOLUME [0:3],
SUBVOL [0:3];
END;
STRUCT INFILE;
BEGIN
INT VOLUME [0:3],
SUBVOL [0:3],
DNAME [0:3];
END;
STRUCT OUTFILE;
BEGIN
INT VOLUME [0:3],
SUBVOL [0:3],
DNAME [0:3];
END;
STRING PARAM [0:564];
END; ! CI STARTUP

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 4
Examples Pathsend Requester Example

?PAGE "STRUCTURE NEEDED TO PROCESS ASSIGN MSGS"

?SECTION ASSIGN

! The following is used to process ASSIGN msgs:

STRUCT CI^ASSIGN (*); !ASSIGN msg

BEGIN !
INT MSG^CODE; ![0] -2
STRUCT LOGICALUNIT; !PARAMS to ASSIGN command
BEGIN !
STRING PROGNAMELEN, ![1]
PROGNAME[0:30], !
FILENAMELEN, ![17]
FILENAME[0:30]; !
END; !
INT(32) FIELDMASK; ![33] bit mask to indicate
!which fields were supplied
!(1=supplied):
! .<0> = TANDEM-FILENAME
! .<1> = PRI-EXT-SIZE
! .<2> = SEC-EXT-SIZE
! .<3> = FILE-CODE
! .<4> = EXCLUSION-SIZE
! .<5> = ACCESS-SPEC
! .<6> = RECORD-SIZE
! .<7> = BLOCK-SIZE
!
STRUCT TANDEMFILENAME; ![35] TANDEM-FILENAME
BEGIN !
INT VOLUME[0:3], !
SUBVOL[0:3], !
DFILE [0:3]; !
END; !
!
!CREATESPEC !
!
INT PRIMARYEXTENT, ![47]
SECONDARYEXTENT, !
FILECODE, !
EXCLUSIONSPEC, ![50] %00 if shared
! %20 if exclusive
! %60 if protected
!
ACCESSSPEC, ![51] %0000 if I/O
! %2000 if input
! %4000 if output
![50-51] correspond to flag PARAM
!of OPEN
!
RECORDSIZE, ![52]
BLOCKSIZE; ![53]
END; ! ci ASSIGN ! !msg size = 108 bytes

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 5
Examples Pathsend Requester Example

?PAGE "STRUCTURE NEEDED TO PROCESS PARAM MSGS"

?SECTION PARAM

! The following is used to process PARAM messages

STRUCT CI^PARAM (*); !PARAM msg

BEGIN !
INT MSG^CODE, ![0] -3
NUM^PARAMS; ![1] number of PARAMS in this msg
STR PARAMETERS[0:1023]; ![2] PARAMS
END; !

?PAGE "STRUCTURE OF THE INPUT FILE"

?SECTION BREQ^INPUT

! The following is the record format of the edit file, which is the
! input to the BREQ program.
!

STRUCT BREQ^INPUT^REC^TEMPLATE (*);

BEGIN
STRUCT SERVER^REQUEST [0:1];
BEGIN
STR PATHMON^ASSIGN^NAME[0:30];
STR PATHMON^SYSTEM^AND^PROCESS^NAME[0:14];
STR SERVER^CLASS[0:14];
END;
END;

?PAGE "STRUCTURE OF THE BREQ OUTPUT RECORD"

?SECTION BREQ^OUTPUT

! The following is the structure of the output record that BREQ

! writes to the OUT file specified in the run line.

STRUCT BREQ^OUTPUT^REC^TEMPLATE (*);

BEGIN
STRUCT SERVER^REPLY [0:1];
BEGIN
STR SYSTEM^NAME[0:7];
STR PROCESS^NAME[0:7];
STRUCT ERROR^MSG;
BEGIN
STR PATHSEND^ERROR[0:77];
STR FILE^SYSTEM^ERROR[0:77];
END;
END;

STR NON^SEND^ERROR^MSG[0:77];
END; ! BREQ output rec template

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 6
Examples Pathsend Requester Example

?PAGE "STRUCTURE OF A REQUEST TO PATHSRV"

?SECTION PATHSRV^REQUEST

! The following is the format of the msg to PATHSRV. It is

! used in BREQ.

STRUCT PATHSRV^REQUEST^TEMPLATE (*);

BEGIN
STR PATHMON^ASSIGN^NAME[0:30];
STR PATHMON^SYSTEM^AND^PROCESS^NAME[0:14];
STR SERVER^CLASS[0:14];
END;

?PAGE "STRUCTURE OF A REPLY FROM PATHSRV"

?SECTION PATHSRV^REPLY

! The following is the format of the reply from PATHSRV. It is

! used in BREQ.

STRUCT PATHSRV^REPLY^TEMPLATE (*);

BEGIN
INT REPLY^CODE;
STRUCT THIS^SERVER;
BEGIN
STR SYSTEM^NAME[0:7];
STR PROCESS^NAME[0:7];
END;

STRUCT SUBSIDIARY^SERVER;
BEGIN
STR SYSTEM^NAME[0:7];
STR PROCESS^NAME[0:7];
END;

STR TMF^ABORT^REQUIRED;

STRUCT ERROR^MSG;
BEGIN
STR PATHSEND^ERROR[0:77];
STR FILE^SYSTEM^ERROR[0:77];
END;

STR NON^SEND^ERROR^MSG[0:77];
END; ! PATHSRV reply template

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 7
Examples Pathsend Requester Example

?PAGE "STRUCTURE OF THE CONTROL BLOCK"

?SECTION CONTROL^BLOCK

! This structure is used to keep data about a Pathsend send or

! message log write. Control blocks (cb's) are allocated from the
! upper 32K using GETPOOL.

! For each transaction, which is driven by data from one record in

! the input file, 3 cb's are allocated (one each for the Pathsend
! sends, and one for the message log write) and linked together in a
! linked list. At transaction completion, the cb memory is returned
! using PUTPOOL.

STRUCT control^block^template (*);

BEGIN
INT type; ! Pathsend cb or msg log cb (see types below)
INT fnum; ! file # of msg log file if cb type is msg log
INT scsend^opnum = fnum; ! sc send op # if cb type is Pathsend
INT record^number; ! corresponding input file rec number

INT io^buf [0: $MAX ($LEN (pathsrv^request^template), ! I/O buf

$LEN (pathsrv^reply^template)) - 1]; ! for sends
! and writes.
STRUCT pathsend^req^buf (pathsrv^request^template) = io^buf;
! for sends
STRUCT pathsend^reply^buf (pathsrv^reply^template) = io^buf;
! for sends

INT .input^data^buf (breq^input^rec^template ) = io^buf;

! for writes, a ptr

STR .pathmon^system^and^process^name; ! pointer into input buf

STR .serverclass^name; ! pointer into input buf

INT pathsend^error; ! if error on send

INT file^system^error; ! if error on send or write

UNSIGNED (1) io^posted; ! if send or write outstanding

UNSIGNED (1) error^is^retryable; ! if we can retry i/o
UNSIGNED (14) not^used; ! filler

STR pathsend^error^msg [0:77]; ! text for pathsend error

STR file^system^error^msg [0:77]; ! text for filesys error

INT .EXT next^cb (control^block^template); ! ptr to next cb in list,

END; ! must be last field in struct

! Types of control blocks. A cb can be used for a PATHSEND send or

! used for a message log write.

LIT cb^type^pathsend = 1;
LIT cb^type^msg^log = 2;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 8
Examples Pathsend Requester Example

?PAGE "GLOBAL DECLARATIONS FOR BREQ"

! The following are GLOBAL literals:

LIT nil^addr = -1D;

LIT true = -1;

LIT false = 0;

LIT e^eof = -1; ! end of file

LIT e^security^violation= 48; ! an access denied

LIT open^read^only = %2000;

LIT open^write^only = %4000;

LIT open^shared = %0;

LIT open^nowait^disk = %1;

LIT retry^path^failures = 1; ! sync depth param in the open

! The following are TRACE file literals:

LIT trace^read^input = 0;
LIT trace^pathsend = 1;
LIT trace^write^to^msg^log = 2;
LIT trace^begin^transaction = 3;
LIT trace^end^transaction = 4;
LIT trace^abort^transaction = 5;
LIT trace^cancel^request = 6;
LIT trace^write^blank^line = 7;
LIT trace^completed^io = 8;
LIT trace^io^timed^out = 9;
LIT trace^io^failed = 10;
LIT trace^retry^transaction = 11;
LIT trace^max^retries^exceeded = 12;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B- 9
Examples Pathsend Requester Example

?PAGE "T9153 PATHWAY PATHSEND SC Errors"

!
! The following is a list of all the Server Class Errors. PATHSEND has
! error numbers 900 - 950 reserved.
!

LITERAL
FEScFirstError = 900, ! First avail. PATHSEND Error.
FEScInvalidServerClassName = 900, ! Invalid server class name.
FEScInvalidPathmonName = 901, ! Invalid Pathmon process name.
FEScPathmonConnect = 902, ! Error with Pathmon
! connection (eg. Open, I/O,
! etc.).
FEScPathmonMessage = 903, ! Unknown message received from
! PATHMON.
FEScServerLinkConnect = 904, ! Error with Server Link
! connection
! (eg. Open, I/O, etc. ).
FEScNoServerAvailable = 905, ! No Server Available.
FEScNoSendEverCalled = 906, ! the user called SC_SEND_INFO
! before ever calling SC_SEND_.
FEScInvalidSegmentID = 907, ! The caller uses an extended
! segment id that is out of range.
FEScNoSegmentInUse = 908, ! The caller supplied a ref.
! parameter that is an extended
! address, but doesn't have an
! extended segment in use.
FEScInvalidFlagsValue = 909, ! The caller set bits in flags
! parameter that are reserved
! and must be 0.
FEScMissingParameter = 910, ! A required parameter was not
! supplied.
FEScInvalidBufferLength = 911, ! One of the buffer length
! parameters is invalid.
FEScParameterBoundsError = 912, ! A reference parameter is out
! of bounds.
FEScServerClassFrozen = 913, ! The Server Class is Frozen.
FEScUnknownServerClass = 914, ! PATHMON does not recognize
! Server Class name.
FEScPathmonShutDown = 915, ! Send denied because Pathmon
! is shutting down.
FEScServerCreationFailure = 916, ! Send denied by PATHMON
! because of Server creation
! failure.
FEScServerClassTmfViolation = 917, ! The Tmf Transaction mode of
! the Send does not match that
! of the ServerClass (eg.
! Requester Send has a TransId
! and the ServerClass is
! configured with TMF OFF).
FEScOperationAborted = 918, ! Send operation aborted. See
! accompanying Guardian error
! for more information.
FEScInvalidTimeoutValue = 919, ! The caller supplied an
! invalid timeout value.
FEScPFSUseError = 920, ! The caller's PFS segment
! could not be accessed.
FEScTooManyPathmons = 921, ! The max. number of Pathmons
! allowed has been exceeded.
FEScTooManyServerClasses = 922, ! The maximum number of server
! classes has been exceeded.
FEScTooManyServerLinks = 923, ! The maximum number of server
! links has been exceeded.
FEScTooManySendRequests = 924, ! The maximum number of send
! requests has been exceeded.
FEScTooManyRequesters = 925, ! The maximum number of allowed

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -10
Examples Pathsend Requester Example

! requesters has been exceeded.

FEScLinkMonConnect = 947, ! Error with LINKMON connection
! (eg. Open, I/O, etc. ).
FEScLastError = 950; ! Last avail. PATHSEND error.
! This is for checking ranges
! not currently returned.

! Miscellaneous file system errors

LIT FEok = 0;
LIT FEInvalidOp = 2;
LIT FEBoundsErr = 22;
LIT FEMissparam = 29;
LIT FESCError = 233;

! Various file numbers used globally

INT term^fnum := -1;

INT in^fnum := -1;
INT out^fnum := -1;
INT error^log^fnum := -1;
INT msg^log^fnum := -1;
INT trace^fnum := -1;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -11
Examples Pathsend Requester Example

INT .my^processid[0:3]; ! used to preface error msgs with my pid

! Define a global array to hold error msgs that aren't PATHSEND send
! errors. For example, reference parameter errors, AWAITIOX timeout
! errors, and validation errors.

STR .global^non^pathsend^error^msg [0:77] := [78 * [" "]];

LIT max^assigns = 56;

STRUCT .assign^table (ci^assign) [0:max^assigns - 1];

INT assign^count := 0; ! count of assigns in table

INT max^retries := -1; ! #times to retry Pathsend failures

INT .edit^control^block[0: (40 + (1024 / 2)) - 1];

! We use GETPOOL and PUTPOOL to manage memory in the upper 32K.

! Here are some vars used by these procedures.
!
LIT HeadSize = 38D; ! 19 words
LIT PoolAddr = %200000D; ! upper 32K

INT .EXT PoolHead := PoolAddr; ! room for the header

INT .EXT pool := PoolAddr + HeadSize; ! start of pool

LIT PoolSize = 32D*1024D; ! 32K bytes

?LIST,PAGE

?PAGE "BREQ PROGRAM EXTDECS"

?PAGE "PROCESS THE ASSIGN MSGS"

PROC assign^proc (rucb, passthru, assign^msg, msg^len, match) VARIABLE;

INT .rucb,
.passthru,
.assign^msg,
msg^len,
match;
EXTERNAL;

?PAGE "RETURN AN ENTRY FROM THE ASSIGN TABLE"

INT PROC get^assign (logical^name, len);
STR .logical^name;
INT len;
EXTERNAL;

?PAGE "READ THE STARTUP AND ASSIGN MSGS, AND OPEN THE LOG FILE"
PROC initialize;
EXTERNAL;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -12
Examples Pathsend Requester Example

?PAGE "START ONE I/O IN THE TRANSACTION"

INT PROC initiate^IO (cb);

INT .EXT cb (control^block^template);
EXTERNAL;

?PAGE "START A WRITE OF THE INPUT REQUEST RECORD TO THE MSG LOG"
INT PROC initiate^write^to^message^log (cb);
INT .EXT cb (control^block^template);
EXTERNAL;

?PAGE "REPORT AN IO ERROR"

PROC IO^error (fnum);
INT fnum;
EXTERNAL;

?PAGE "SEARCH THE CONTROL BLOCKS FOR OUTSTANDING I/O"

INT PROC io^outstanding (cb);
INT .EXT cb (control^block^template);
EXTERNAL;

?PAGE "PRINT A MESSAGE TO THE TERMINAL AND ABEND"

PROC abend^with^my^abend^msg;
EXTERNAL;

?PAGE "PROCESS THE PARAM MSGS"

PROC param^proc (rucb, buf, param^msg, msg^len, match) VARIABLE;

INT .rucb,
.buf,
.param^msg (ci^param),
msg^len,
match;
EXTERNAL;

?PAGE "READ FROM AN EDIT FILE"

INT PROC read^ (fnum, buf, read^count, error);
INT fnum;
STR .buf;
INT read^count;
INT .error;
EXTERNAL;

?PAGE "LOOK AT WHY SERVERCLASS_SEND_INFO_ FAILED"

PROC ServerClass^Send^Info^error (error);
INT error;
EXTERNAL;

?PAGE "SET UP THE CONTROL BLOCKS FOR ONE TRANSACTION"

DBL PROC setup^control^blocks (input^buf, record^number);

INT .input^buf (breq^input^rec^template);
INT record^number;
EXTERNAL;

?PAGE "START A TMF TRANSACTION"

PROC start^the^tmf^transaction;
EXTERNAL;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -13
Examples Pathsend Requester Example

?PAGE "PROCESS THE STARTUP MSG"

PROC startup^proc (rucb, buf, msg, msg^len, match) VARIABLE;
INT .rucb,
.buf,
.msg (ci^startup),
msg^len,
match;
EXTERNAL;

?PAGE "STORE DATA IN THE CONTROL BLOCK"

PROC store^control^block^info (cb, data^buf, cb^type, record^number);

INT .EXT cb (control^block^template);
STR .data^buf (pathsrv^request^template);
INT cb^type;
INT record^number;
EXTERNAL;

?PAGE "BE SURE AN INCOMING REQUEST IS VALID"

INT PROC validate^breq^request (input^rec);
INT .input^rec (breq^input^rec^template);
EXTERNAL;

?PAGE "GET MY NAME OR PROCESS ID"

PROC who^am^i;
EXTERNAL;

?PAGE "WRITE TO THE TRACE FILE"

PROC write^trace^file (function, record^number) VARIABLE;
INT function;
INT record^number;
EXTERNAL;

?LIST

?NOLIST,SOURCE $SYSTEM.SYSTEM.EXTDECS0 (ABEND, ABORTTRANSACTION,

? AWAITIOX, BEGINTRANSACTION,
? CANCELREQ, CLOSE, DEBUG,
? DEVICEINFO, DNUMOUT, EDITREAD,
? EDITREADINIT, ENDTRANSACTION,
? FILEINFO, FNAMECOLLAPSE,
? FNAMEEXPAND, GETCRTPID,
? GETTRANSID, INITIALIZER,
? MYPID, NEWPROCESS, NUMIN,
? NUMOUT, OPEN, PROCESSORSTATUS,
? READ, READUPDATE, REPLY,
? SETMODE, STOP, WRITE,
? WRITEREAD, WRITEX,
? DEFINEPOOL, GETPOOL, PUTPOOL,
? MYTERM,
? SERVERCLASS_SEND_ ,
? SERVERCLASS_SEND_INFO_ )

?LIST

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -14
Examples Pathsend Requester Example

! The following dummy procedure is for example program version control

PS^EXAMPLE^VERSION^PROC;

?PAGE "PRINT A MESSAGE TO THE TERMINAL AND ABEND"

! Many procedures check for error conditions that, under normal

! circumstances, should not happen. In these cases, before
! ABENDing, a procedure will call this procedure to write a message
! to the home terminal. This proc calls ABEND.

PROC abend^with^my^abend^msg;

BEGIN
INT .filename [0:11];
INT term^fnum;
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

CALL MYTERM (filename);

CALL OPEN (filename, term^fnum);

IF <>
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

sbuf ':=' "INTERNAL ERROR DETECTED, PROGRAM ABENDING" -> @sp;

CALL WRITE (term^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL CLOSE (term^fnum);

CALL ABEND;

END; ! PROC abend^with^my^abend^msg

?PAGE "ABORT THE TRANSACTION"

!
! Abort the transaction I started
!
PROC abort^tmf^transaction;

BEGIN
INT tmf^error;
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -15
Examples Pathsend Requester Example

IF (tmf^error := ABORTTRANSACTION)
THEN ! failed to abort, file sys error in tmf^error
BEGIN
sbuf ':=' "ERROR WITH ABORTTRANSACTION: " -> @sp;
CALL NUMOUT (sp, tmf^error, 10, 3);
@sp := @sp[3];

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;
END;

! Successfully aborted the transaction

CALL write^trace^file (trace^abort^transaction);

sbuf ':=' "TRANSACTION ABORTED NORMALLY" -> @sp;

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (msg^log^fnum);

END; ! PROC abort^tmf^transaction

?PAGE "FIND OUT WHY SERVERCLASS_SEND_ FAILED"

! The following procedure is called from PROCs complete^io and

! initiate^IO. That is, SERVERCLASS_SEND_ and AWAITIOX can both
! return file system error 233 (FESCERROR).

! This proc gets detailed information about error 233 by calling

! SERVERCLASS_SEND_INFO_ , which returns the pathsend error and
! the file system error.

PROC analyze^send^error^233 (cb);

INT .EXT cb (control^block^template);

BEGIN
STR .EXT sp;
STR .EXT st;
INT error; ! used in call to serverclass_send_info_
INT pathsend^error; ! used in call to serverclass_send_info_
INT filesystem^error; ! used in call to serverclass_send_info_

! Default to a non-retryable error

cb.error^is^retryable := false;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -16
Examples Pathsend Requester Example

IF (error := SERVERCLASS_SEND_INFO_ (pathsend^error, filesystem^error))

THEN ! problem making the call to SERVERCLASS_SEND_INFO_

BEGIN
CALL serverclass^send^info^error (error);
RETURN;
END;

! Save off the errors and error msg text

cb.pathsend^error := pathsend^error;
cb.file^system^error := filesystem^error;

! This application treats Pathsend errors 916, 918, and 924 as

! retryable. Other errors may be retryable, depending on your
! application. They generally reflect PATHWAY configuration
! problems, and would be retryable after a delay. For example,

! 904 (FEScServerLinkConnect): error with server link connection

! (Open, I/O, etc.).

! 905 (FEScNoServerAvailable): no server in the requested server

! class has a free link. the maximun number of links is specified
! in PATHCOM by the server class attribute MAXLINKS. A link could
! become available if a TCP was shutdown.

! 913 (FEScServerClassFrozen): the server class is frozen.

! 916 (FEScServerCreationFailure): send denied by PATHMON because

! of a server creation failure.

! 918 (FEScSendOperationAborted): the send operation was

! terminated. See file system error for more information. For
! example, corresponding file system error could be error 40,
! time out.

! 924 (FEScTooManySendRequests): the maximum number of allowed

! send requests has been exceeded.

IF (pathsend^error = FEScServerCreationFailure !916!

OR pathsend^error = FEScOperationAborted !918!
OR pathsend^error = FEScTooManySendRequests !924!)

THEN ! we'll retry the SERVERCLASS_SEND_

cb.error^is^retryable := true

ELSE ! store error msg text, trap parameter errors

BEGIN
cb.pathsend^error^msg ':=' "PATHSEND ERROR: " -> @sp;
CALL DNUMOUT (sp, $UDBL (pathsend^error), 10, 3);
@sp := @sp[3];

cb.file^system^error^msg ':=' "FILE SYSTEM ERROR: " -> @st;

CALL DNUMOUT (st, $UDBL (filesystem^error), 10, 3);
@st := @st[3];

! Use the pathsend error to check for parameter errors, and flag
! them as non-retryable. Alternatively, we could use the value
! returned in the file system error to trap a bad param (error 2:
! FEInvalidOp, error 21: FEBadCount, error error 22: FEBadParam,
! error 29: FEMissParam).
!
! The pathsend error does provide more specific information about
! about the problem. For example, pathsend errors 907, 908, 909,
! and 919 all return the same file system error (error 2,
! FEInvalOp).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -17
Examples Pathsend Requester Example

IF pathsend^error = FEScInvalidSegmentId !907!

OR pathsend^error = FEScNoSegmentInUse !908!
OR pathsend^error = FEScInvalidFlagsValue !909!
OR pathsend^error = FEScMissingParameter !910!
OR pathsend^error = FEScInvalidBufferLength !911!
OR pathsend^error = FEScParameterBoundsError !912!
OR pathsend^error = FEScInvalidTimeoutValue !919!

THEN ! the error returned from serverclass_send_ was a param error

sp ':=' "(BAD PARAMETER PASSED TO SERVERCLASS_SEND_)" -> @sp;

! Include a msg if the file system error is a security violation

IF filesystem^error = e^security^violation !48!

THEN ! add error text for the user
st ':=' "(SECURITY VIOLATION)";

END; ! else trap parameter errors

END; ! PROC analyze^send^error^233

?PAGE "PROCESS THE ASSIGN MESSAGES"

! This procedure is called by the GUARDIAN 90 PROC INITIALIZER from

! PROC initialize. It saves ASSIGN msgs, passed from our ancestor
! TACL, in our ASSIGN table, and is called once for each ASSIGN msg
! passed.

! A count of the number of ASSIGNs in the table is kept in global

! var assign^count, and is used in table lookups.

PROC assign^proc (rucb, passthru, assign^msg, msg^len, match) VARIABLE;

INT .rucb,
.passthru,
.assign^msg,
msg^len,
match;

BEGIN
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -18
Examples Pathsend Requester Example

IF assign^count >= max^assigns

THEN ! will exceed the size of our assign table
BEGIN
sbuf ':=' "NUMBER OF ASSIGNS EXCEEDS MAX^ASSIGNS" -> @sp;

CALL WRITE (term^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL ABEND;
END;

assign^table [assign^count] ':=' assign^msg FOR (msg^len + 1) / 2;

assign^count := assign^count + 1;

END; ! PROC assign^proc

?PAGE "CANCEL ALL OUR OUTSTANDING I/O"

! This proc is called to cancel I/O posted but not completed. We

! use the io^posted flag in the control block to get the state of
! the I/O.

! This proc is called when one I/O in a transaction failed, either

! at initiation time or completion time.

PROC cancel^outstanding^io (cb^list^head);

INT .EXT cb^list^head (control^block^template); ! ptr to head of cb list

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;

WHILE @cb <> nil^addr DO

BEGIN

IF cb.io^posted
THEN ! this cb has I/O that hasn't completed
BEGIN
CALL CANCELREQ (cb.fnum, @cb);
IF <
THEN ! print an error msg and abend
CALL IO^error (cb.fnum);

cb.io^posted := false;
CALL write^trace^file (trace^cancel^request, cb.record^number);
END;

@cb := @cb.next^cb;

END; ! while

END; ! PROC cancel^outstanding^io

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -19
Examples Pathsend Requester Example

?PAGE "COMPLETE THE OUTSTANDING I/O"

! This PROC waits for I/O to complete and returns the address of the
! control block associated with the completion. It returns nil^addr
! if I/O completed with an error, or nil^addr if there was no
! completion (in the case of a timeout, error 40).

DBL PROC complete^io;

BEGIN
INT .EXT cb (control^block^template);
DBL tag := -1D;
DBL timelimit := 300D; ! wait 3 seconds (0.01 sec units)
LIT anyfnum = -1; ! wait on any file
INT fnum; ! used in call to awaitiox
STR .EXT sp;
INT error;
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .st;

! When waiting for any I/O to complete (fnum = -1) with an AWAITIO
! time limit <> 0, I/O is considered complete and no longer
! outstanding (whether an error is returned or not) in every case
! except when a timeout (error 40) occurs.
!
! When a completion occurs, the file number of the completed call
! is returned in the <fnum> param, and tag is returned in <tag>.

fnum := anyfnum;

CALL AWAITIOX (fnum,,, tag, timelimit);

IF =
THEN ! one I/O successfully completed
BEGIN
@cb := tag;

! Do some basic checks to make sure the control block is what we

! expect. We expect the file number returned by AWAITIOX to be
! the same as the fle number we saved in the control block when
! the I/O was initiated. And we expect the control block to be
! a valid type.

IF cb.fnum <> fnum OR

(cb.type <> cb^type^pathsend AND cb.type <> cb^type^msg^log)

THEN ! assertion failed

CALL abend^with^my^abend^msg;

cb.io^posted := false;

CALL write^trace^file (trace^completed^io, cb.record^number);

RETURN @cb;
END;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -20
Examples Pathsend Requester Example

! An error or warning occurred

CALL FILEINFO (fnum, error);

IF <>
THEN ! couldn't access the ACB
CALL abend^with^my^abend^msg;

! Check for a timeout error

IF error = 40
THEN ! AWAITIO timed out and no completion has occurred
BEGIN
sbuf ':=' "AWAITIO TIMED OUT (ERROR 40)" -> @st;

CALL WRITE (error^log^fnum, buf, @st '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL write^trace^file (trace^io^timed^out);

! Save the error msg so we can include it in the output record

global^non^pathsend^error^msg ':=' sbuf FOR @st '-' @sbuf;

RETURN nil^addr;
END;

! I/O completed with some error, find out what it is

@cb := tag;

! We expect the file number returned by AWAITIOX to be the same as

! the fle number we saved in the control block when the I/O was
! initiated.

IF cb.fnum <> fnum

THEN ! assertion failed
CALL abend^with^my^abend^msg;

cb.io^posted := false;

CALL write^trace^file (trace^io^failed, cb.record^number);

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -21
Examples Pathsend Requester Example

! Check for errors on the disk operation

CASE cb.type OF
BEGIN

cb^type^msg^log ->

BEGIN

! I/O to the message log file failed

cb.file^system^error := error;
cb.file^system^error^msg ':=' "ERROR: " -> @sp;
CALL DNUMOUT (sp, $UDBL (error), 10,3);
@sp := @sp[3];
sp ':=' "RETURNED FROM AWAITIO ON I/O TO THE MESSAGE LOG FILE";

RETURN nil^addr;
END;

cb^type^pathsend ->

BEGIN

! A nowaited ServerClass_Send_ failed

IF error = FeSCError !file system error 233!

THEN ! call ServerClass_Send_Info_ for Pathsend & filesys errors
CALL analyze^send^error^233 (cb)

ELSE ! no other errors should be returned by AWAITIOX for

CALL abend^with^my^abend^msg; ! Pathsend sends

END;

OTHERWISE ->

! There are no other types of control blocks, so

CALL abend^with^my^abend^msg;

END; ! case

RETURN nil^addr;

END; ! PROC complete^io

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -22
Examples Pathsend Requester Example

?PAGE "CREATE THE CONTROL BLOCK MEMORY POOL"

! This procedure uses the standard memory management routines to

! declare a memory pool starting at the upper 32K, with a size of
! 32K bytes.

! Control blocks (cb's) are managed as a linked list, and new cb's
! are allocated in get^control^block.

PROC create^the^memory^pool;

BEGIN
INT status;
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

IF (status := DEFINEPOOL (PoolHead, pool, PoolSize))

THEN ! some memory allocation problem
BEGIN
sbuf ':=' "DEFINEPOOL FAILED TO CREATE THE MEMORY POOL" -> @sp;

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;
END;

END; ! PROC create^the^memory^pool

?PAGE "END THE TMF TRANSACTION"

! End the transaction I started, and log the outcome to the error
! log file. If successful, write a record to the trace file.

PROC end^the^tmf^transaction;

BEGIN
INT tmf^error;
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

IF (tmf^error := ENDTRANSACTION)
THEN ! file sys error in tmf^error
BEGIN
sbuf ':=' "ERROR WITH ENDTRANSACTION: " -> @sp;
CALL NUMOUT (sp, tmf^error, 10, 3);
@sp := @sp[3];

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;
END;

! Successfully ended the transaction

CALL write^trace^file (trace^end^transaction);

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -23
Examples Pathsend Requester Example

sbuf ':=' "TRANSACTION ENDED NORMALLY" -> @sp;

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);
IF <
THEN ! print an error msg and abend
CALL IO^error (msg^log^fnum);

END; ! PROC end^the^tmf^transaction

?PAGE "RETURN AN ENTRY FROM THE ASSIGN TABLE"

! This procedure is passed a logical ASSIGN name and searchs the

! ASSIGN table for the ASSIGN name. If found it returns the address
! of the table entry, otherwise false.

INT PROC get^assign (logical^name, len);

STR .logical^name;
INT len;

BEGIN
INT i := 0;
INT .assign^ (ci^assign);

WHILE i < assign^count DO

BEGIN
@assign^ := @assign^table[i];

IF assign^.logicalunit.filename = logical^name FOR len

AND assign^.logicalunit.filenamelen = len

THEN ! found matching ASSIGN

RETURN @assign^;

i := i + 1;
END;

RETURN false;
END; ! INT PROC get^assign

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -24
Examples Pathsend Requester Example

?PAGE "ALLOCATE A NEW CONTROL BLOCK FROM THE MEMORY POOL"

! This proc gets memory from the pool for a new control block (cb).
! The caller passes in the head of a list of control blocks, and it
! links the new cb to the end of that list.

! If the list head has a nil address, then the new cb becomes the
! first element of the list.

! This proc returns the address of the new control block. Before
! returning, the new cb is zero'd.

DBL PROC get^control^block (cb^list^head);

INT .EXT cb^list^head (control^block^template);

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;
INT .EXT new^cb (control^block^template);
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

@new^cb := GETPOOL (PoolHead, $UDBL ($LEN (cb)));

IF @new^cb = -1D
THEN ! couldn't get memory
BEGIN
sbuf ':=' "GETPOOL FAILED TO GET MEMORY IN GET^CONTROL^BLOCK" -> @sp;

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;
END;

! If the caller passed in a list head with nil address, then make
! the new cb the first element in the list.

IF @cb = nil^addr
THEN ! the user passed an empty list
@cb := @new^cb

ELSE ! link the new cb to the end of the list

BEGIN

WHILE @cb.next^cb <> nil^addr DO

@cb := @cb.next^cb;

@cb.next^cb := @new^cb;
@cb := @cb.next^cb;

END;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -25
Examples Pathsend Requester Example

! Clean the new control block, and set the link to nil address

cb ':=' 0 & cb FOR (($LEN (cb) + 1) / 2) - 1;

@cb.next^cb := nil^addr;

RETURN @cb;

END; ! DBL PROC get^control^block

?PAGE "READ THE STARTUP AND ASSIGN MSGS, AND OPEN VARIOUS LOG FILES"

! This procedure calls the GUARDIAN 90 PROC INITIALIZER, which opens

! $RECEIVE and handles the protocol with our ancestor, TACL.
! INITIALIZER calls procs local to this program to read the startup
! msg, save the ASSIGN msgs, and save the param msg.

! This proc then opens the error log file, message log file, and the
! trace file.

PROC initialize;

BEGIN
STRUCT .startup^msg (ci^startup);
INT .buf [0:79];
STR .sp := @buf '<<' 1;
STR .st;
INT .filename [0:11];
INT .error^log^file[0:11];
INT .msg^log^file[0:11];
INT .trace^file[0:11];
INT .assign^ (ci^assign);
STR .assign^name[0:17] := [18*[" "]];
STR .logical^name[0:30];
INT setmode^param1;
INT devtype, phys^reclen;

! Open the home terminal to report initialization errors

CALL MYTERM (filename);

CALL OPEN (filename, term^fnum);
IF <>
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

! Get my process id and read the startup msg, ASSIGNs and PARAMs

CALL who^am^i;

CALL INITIALIZER (!rucb!, startup^msg, startup^proc, param^proc,

assign^proc);

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -26
Examples Pathsend Requester Example

! Verify that the ASSIGNs for msg log file, error log file, and
! trace file are present in our ASSIGN table.

! The msg log file contains each record read from the input file

logical^name ':=' "MESSAGE-LOG-FILE ";

IF NOT (@assign^ := get^assign (logical^name, 16))

THEN ! user forgot assign
BEGIN
sp ':=' "MISSING ASSIGN 'MESSAGE-LOG-FILE' " ->@st;

CALL WRITE (term^fnum, buf, @st '-' @sp);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL ABEND;
END;

! The logical ASSIGN name is in assign.logicalunit.filename and the

! physical file name is in internal format in assign.tandemfilename.
! When a user adds an ASSIGN and omits the $VOL and/or SUBVOL, these
! fields are also omitted from the ASSIGN STRUCT.

! By collapsing the file name to it's external format

! ($VOL.SUBVOL.FN) and then expanding it to it's internal format
! ("\246ANCHORSUBVOL FILE "), we let the GUARDIAN 90 PROCs do
! the work of adding default values.

CALL FNAMECOLLAPSE (assign^.tandemfilename.volume, msg^log^file);

CALL FNAMEEXPAND (msg^log^file, msg^log^file, startup^msg.default);

! The error log file is used to log errors generated by BREQ

logical^name ':=' "ERROR-LOG-FILE ";

IF NOT (@assign^ := get^assign (logical^name, 14))

THEN ! user forgot assign
BEGIN
sp ':=' "MISSING ASSIGN 'ERROR-LOG-FILE' " ->@st;

CALL WRITE (term^fnum, buf, @st '-' @sp);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL ABEND;
END;

CALL FNAMECOLLAPSE (assign^.tandemfilename.volume, error^log^file);

CALL FNAMEEXPAND (error^log^file, error^log^file, startup^msg.default);

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -27
Examples Pathsend Requester Example

! The trace file contains a log of specific events that occurred in

! BREQ, e.g., a PATHSEND send or ENDTRANSACTION.

logical^name ':=' "TRACE-FILE ";

IF (@assign^ := get^assign (logical^name, 10))

THEN ! optional assign, use if included
BEGIN
CALL FNAMECOLLAPSE (assign^.tandemfilename.volume, trace^file);
CALL FNAMEEXPAND (trace^file, trace^file, startup^msg.default);

CALL OPEN (trace^file, trace^fnum, open^write^only LOR

open^shared, retry^path^failures);
IF <>
THEN ! open failed
BEGIN
sp ':=' "FAILED TO OPEN TRACE FILE" ->@st;

CALL WRITE (term^fnum, buf, @st '-' @sp);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL IO^error (trace^fnum);

END;

! By sharing the same terminal as the CI, which uses BREAK, we

! gain BREAK access to the term with SETMODE 12 (set file
! type access, param 2 = 1). File access is BREAK access.

CALL DEVICEINFO (trace^file, devtype, phys^reclen);

IF devtype.<4:9> = 6 ! Type TERM!

THEN ! Do a setmode so we can write to it.
CALL SETMODE (trace^fnum, 12, 0, 1);

END
ELSE ! trace assign not present, so don't trace
trace^fnum := -1;

CALL OPEN (error^log^file, error^log^fnum, open^write^only LOR

open^shared, retry^path^failures);
IF <>
THEN ! open failed
BEGIN
sp ':=' "FAILED TO OPEN ERROR LOG FILE" ->@st;

CALL WRITE (term^fnum, buf, @st '-' @sp);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL IO^error (error^log^fnum);

END;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -28
Examples Pathsend Requester Example

CALL DEVICEINFO (error^log^file, devtype, phys^reclen);

IF devtype.<4:9> = 6 !type term!

THEN ! Do a setmode so we can write to an un-paused TERM.
CALL SETMODE (error^log^fnum, 12, 0, 1);

CALL OPEN (msg^log^file, msg^log^fnum, open^write^only LOR open^shared

LOR open^nowait^disk, retry^path^failures);
IF <>
THEN ! open failed
BEGIN
sp ':=' "FAILED TO OPEN MESSAGE LOG FILE" -> @st;

CALL WRITE (term^fnum, buf, @st '-' @sp);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL IO^error (msg^log^fnum);

END;

CALL CLOSE (term^fnum);

END; ! PROC initialize

?PAGE "START ONE I/O IN THE TRANSACTION"

! This procedure uses data in the control block (cb) to start a

! Pathsend send or message log WRITE. The I/O is nowaited, and the
! I/O tag is the cb address. If we fail to initiate the
! SERVERCLASS_SEND_, then call analyze^send^error^233 to get the
! associated Pathsend error and file system error.

! This application treats initiation errors as non-retryable.

INT PROC initiate^IO (cb);

INT .EXT cb (control^block^template);

BEGIN
INT len;
INT ^ScSendOpNum;
INT pathmon^process^name^len;
INT serverclass^name^len;
INT flags := 0;
DBL timeout;
INT error;
STR .sp;

! If this is a msg log control block then start the write to the msg
! log file.

IF cb.type = cb^type^msg^log
THEN ! start the disc op
RETURN initiate^write^to^message^log (cb);

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -29
Examples Pathsend Requester Example

! Start the Pathsend send to the server class. Fill the send buf
! with blanks so there's not a cascading Pathsend send from server
! to server class.

cb.io^buf ':=' " " & cb.io^buf FOR $OCCURS (cb.io^buf) - 1;

! Set up the ServerClass_Send_ parameters: get the PATHMON name and

! server class name lengths, and set the values for <flags> and
! <timeout>.

len := $OCCURS (cb.pathsend^req^buf.pathmon^system^and^process^name) - 1;

WHILE cb.pathmon^system^and^process^name [len] = " " OR

cb.pathmon^system^and^process^name [len] = 0

DO ! skip blanks and nulls

len := len - 1;

pathmon^process^name^len := len + 1;

len := $OCCURS (cb.pathsend^req^buf.server^class) - 1;

WHILE cb.serverclass^name [len] = " " OR

cb.serverclass^name [len] = 0

DO ! skip blanks and nulls

len := len - 1;

serverclass^name^len := len + 1;

! Setting bit 15 in flags means send nowaited. Setting timeout to

! -1D means we're willing to wait forever for the send to complete,
! and LINKMON will not cause this send to time out.

! Remember, when a Pathsend send times out (AWAITIOX completes with

! error 233 and ServerClass_Send_Info returns file system error 40),
! the outstanding I/O to the server process is cancelled.

! Note that setting a timeout value here is independent of setting a

! timeout value in the call to AWAITIOX.

flags.<15> := 1;
timeout := -1D;

! Initiate the Pathsend I/O. If no error is returned from

! SERVERCLASS_SEND_, then the I/O is successfully initiated.

! If an error is returned, then the I/O was NOT successfully

! initiated and NO I/O to a server process is outstanding.

! NOTE: A value of -1 is returned in the <scsend-op-num> param for

! waited sends. A value of -1 is also returned for nowaited sends
! that are not successfully initiated.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -30
Examples Pathsend Requester Example

! Also note that the <actual-reply-len> param is an optional

! reference param that has a return value of 0 for nowaited I/O.

error := SERVERCLASS_SEND_ (cb.pathmon^system^and^process^name,

pathmon^process^name^len,
cb.serverclass^name,
serverclass^name^len,
cb.pathsend^req^buf,
$LEN (pathsrv^request^template),
$LEN (pathsrv^reply^template),
, ! actual reply len for waited I/O only !
timeout,
flags,
^ScSendOpNum,
@cb !tag! );

IF error = FEok !0!

THEN ! successfully initiated the Pathsend send
BEGIN
cb.ScSend^OpNum := ^ScSendOpNum;
cb.io^posted := true;
CALL write^trace^file (trace^pathsend, cb.record^number);
RETURN true;
END;

! The send failed with an initiation error. Find out

! what happened.

cb.io^posted := false;

IF error = FESCError !file system error 233!

THEN ! call SERVERCLASS_SEND_INFO to get Pathsend and file system errors
CALL analyze^send^error^233 (cb)

ELSE ! we should only see errors 0 or 233

CALL abend^with^my^abend^msg;

RETURN false;

END; ! PROC initiate^IO

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -31
Examples Pathsend Requester Example

?PAGE "START A WRITE OF THE INPUT REQUEST RECORD TO THE MSG LOG"

! This proc is called from initiate^IO, and starts the I/O to the
! message log file. The WRITE is tagged so we can identify this I/O
! when completing it with AWAITIOX, or cancelling it with CANCELREQ.

! If the WRITE fails, we will not return.

INT PROC initiate^write^to^message^log (cb);

INT .EXT cb (control^block^template);

BEGIN
INT error := 0;

CALL WRITEX (msg^log^fnum, cb.input^data^buf, $LEN(cb.input^data^buf),, @cb);

IF <
THEN ! error occurred on the write
BEGIN

CALL FILEINFO (msg^log^fnum, error);

IF <>
THEN ! couldn't access the ACB
CALL abend^with^my^abend^msg;

! Print an error message and abend

CALL IO^error (msg^log^fnum);

END;

! I/O successfuly initiated

cb.io^posted := true;
cb.fnum := msg^log^fnum;

CALL write^trace^file (trace^write^to^msg^log, cb.record^number);

RETURN true;

END; ! INT PROC initiate^write^to^message^log

?PAGE "WRITE AN ERROR MSG TO THE ERROR LOG"

! This procedure is passed a file number and gets the file system
! error and file name associated with that file number. It builds an
! error msg and writes it to either the error log file or the home
! terminal. This PROC calls ABEND.

PROC IO^error (fnum);

INT fnum;

BEGIN
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;
INT .filename [0:11];
INT len;
INT error;
INT OutFnum;

! If an OPEN failed, fnum is -1 and FILEINFO will return the error

! associated with the failed OPEN. The file name param, in this
! case, will be invalid.

CALL FILEINFO (fnum, error, filename);

IF <>
THEN ! couldn't access the ACB

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -32
Examples Pathsend Requester Example

CALL abend^with^my^abend^msg;

sbuf ':=' "FILE SYSTEM ERROR " -> @sp;

CALL NUMOUT (sp, error, 10, 3);
@sp := @sp[3];

! If the error is on a file already opened, include the file name

! in the error msg. If the error occurred while trying to open
! a file, we can't include the file name.

IF fnum <> -1
THEN ! the error occurred on a file already opened
BEGIN
sp ':=' " FILE " -> @sp;
@sp := @sp [len := FNAMECOLLAPSE (filename, sp)];
END;

! Write the msg to the terminal if the error log isn't open

IF error^log^fnum <= 0
THEN ! send output to the home term
BEGIN
CALL MYTERM (filename);
CALL OPEN (filename, OutFnum);
IF <> THEN CALL ABEND;
END

ELSE ! error log file is open

OutFnum := error^log^fnum;

CALL WRITE (OutFnum, buf, @sp '-' @sbuf);

CALL CLOSE (OutFnum);

CALL ABEND;

END; ! PROC IO^error

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -33
Examples Pathsend Requester Example

?PAGE "SEARCH THE CONTROL BLOCKS FOR I/O POSTED BUT NOT COMPLETED"

! I/O's were initiated out of control blocks. Here we step through

! the list of cb's and check for I/O that was initiated but not
! completed. If there are any outstanding I/O's, return true,
! otherwise return false.

INT PROC io^outstanding (cb^list^head);

INT .EXT cb^list^head (control^block^template); ! first element in cb list

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;

WHILE @cb <> nil^addr

DO
BEGIN
IF cb.io^posted
THEN ! yes, there is outstanding I/O
RETURN true;

@cb := @cb.next^cb;
END;

RETURN false;

END; ! PROC io^outstanding

?PAGE "OUTPUT THE RESULTS OF ONE TRANSACTION"

! This procedure moves data from the two Pathsend control blocks
! into the output buffer, and then writes the output buffer to the
! out file.

PROC output^the^results (cb^list^head);

INT .EXT cb^list^head (control^block^template);

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;
STRUCT .out^buf (breq^output^rec^template);
INT i := 0;
STR .EXT sp;

! Clear the output buffer

out^buf ':=' " " & out^buf FOR ($LEN (out^buf) + 1) / 2;

! Store the outcome of the two Pathsend sends into the output buffer

WHILE @cb <> nil^addr DO

BEGIN

IF cb.type = cb^type^pathsend
THEN ! this cb has data about the outcome of a Pathsend send
BEGIN

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -34
Examples Pathsend Requester Example

! Store the server's reply: system name and process name

out^buf.server^reply[i].system^name ':='
cb.pathsend^reply^buf.this^server.system^name
FOR $OCCURS (cb.pathsend^reply^buf.this^server.system^name);

out^buf.server^reply[i].process^name ':='
cb.pathsend^reply^buf.this^server.process^name
FOR $OCCURS (cb.pathsend^reply^buf.this^server.process^name);

! Store Pathsend error msg text and file system error msg text
! generated by this program.

! When an error is retryable we retry the transaction and

! don't include error msg text. If the retries also failed,
! we need to add msg text now.

IF (cb.pathsend^error AND NOT cb.pathsend^error^msg) OR

(cb.file^system^error AND NOT cb.file^system^error^msg)

THEN ! include error msg text

BEGIN
cb.pathsend^error^msg ':=' "PATHSEND ERROR: " -> @sp;
CALL DNUMOUT (sp, $UDBL (cb.pathsend^error), 10, 3);

cb.file^system^error^msg ':=' "FILE SYSTEM ERROR: " -> @sp;

CALL DNUMOUT (sp, $UDBL (cb.file^system^error), 10, 3);

END;

IF cb.pathsend^error^msg
THEN ! include msg text

out^buf.server^reply[i].error^msg.pathsend^error ':='
cb.pathsend^error^msg FOR $OCCURS (cb.pathsend^error^msg);

IF cb.file^system^error^msg
THEN ! include msg text

out^buf.server^reply[i].error^msg.file^system^error ':='
cb.file^system^error^msg FOR $OCCURS (cb.file^system^error^msg);

! Index to next server reply element in output buffer

i := i + 1;

END; ! if

@cb := @cb.next^cb;

END; ! while

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -35
Examples Pathsend Requester Example

! Store non-Pathsend error messages generated by this program

out^buf.non^send^error^msg ':=' global^non^pathsend^error^msg

FOR $OCCURS (out^buf.non^send^error^msg);

! Write the record to the output file

CALL WRITE (out^fnum, out^buf, $LEN (out^buf));

IF <>
THEN ! print an error msg and abend
CALL IO^error (out^fnum);

END; ! PROC output^the^results

?PAGE "PROCESS THE PARAM MSGS"

! This procedure is called by GUARDIAN 90 PROC INITIALIZER from PROC

! initialize to save the PARAM MAX-RETRIES and store into variable
! MAX-RETRIES.

PROC param^proc (rucb, passthru, param^msg, msg^len, match) VARIABLE;

INT .rucb;
INT .passthru;
INT .param^msg (ci^param);
INT msg^len;
INT match;

BEGIN
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
INT i := 0, found := false;
STR .sp, .st;
INT status;
INT early^exit := false;

! Unlike ASSIGNs, we receive all the PARAMs in a single

! PARAM msg.

@sp := @param^msg.parameters;

! The early^exit flag is used to exit the loop if the parameter has
! an invalid value specified. The found flag is used to exit the
! loop if the param is found and has a valid value.

WHILE (i := i + 1) <= param^msg.num^params AND NOT found AND NOT early^exit

DO
BEGIN

IF sp[1] = "MAX-RETRIES"
THEN ! found the param
BEGIN
sbuf ':=' sp[sp + 2] FOR sp[sp + 1] -> @st;
st := 0;

CALL NUMIN (sbuf, max^retries, 10, status);

IF status
THEN ! illegal value
BEGIN
max^retries := -1;
early^exit := true;
END

ELSE
found := true;

END;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -36
Examples Pathsend Requester Example

IF NOT found AND NOT early^exit

THEN ! look at the next param
@sp := @sp[sp + sp[sp + 1] + 2];

END; ! while i < num params and not found

IF NOT found
THEN ! missing or bad param
BEGIN
sbuf ':=' "MISSING OR ILLEGAL PARAM 'MAX-RETRIES' " -> @sp;

CALL WRITE (term^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;

END; ! if not found

END; ! PROC param^proc

?PAGE "PROCESS ONE TRANSACTION"

! This PROC executes one transaction. It initiates the two nowaited

! ServerClass_Sends_ and one message log write, and calls the
! routine complete^IO to finish the I/O.

! If the I/O completes successfully, this proc ends the transaction,

! otherwise it aborts the transaction and cancels any outstanding
! I/O.

INT PROC process^transaction (cb^list^head);

INT .EXT cb^list^head (control^block^template);

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;

CALL start^the^tmf^transaction;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -37
Examples Pathsend Requester Example

! Step through the list of control blocks (cb's), initiating the I/O
! associated with each cb (a send or a write).

WHILE @cb <> nil^addr DO

BEGIN

IF NOT initiate^IO (cb)

THEN ! failed to start the operation, treat error as non-retryable
BEGIN
CALL cancel^outstanding^io (cb^list^head);
CALL abort^tmf^transaction;
RETURN false;
END;

! Successfully started the I/O for this cb, goto next cb

@cb := @cb.next^cb;

END; ! while

! IO^Outstanding is a proc that steps through the list of cb's

! checking for I/O that has been posted but not yet completed.

WHILE io^outstanding (cb^list^head) DO

BEGIN

! Complete^IO returns @cb if the I/O completed successfully,

! otherwise it returns the nil address.

@cb := complete^io;

IF @cb = nil^addr
THEN ! an I/O failed
BEGIN
CALL cancel^outstanding^io (cb^list^head);
CALL abort^tmf^transaction;
RETURN false;
END;

! I/O completed successfully, check the reply buffer

IF cb.type = cb^type^pathsend !pathsend control block!

AND cb.pathsend^reply^buf.reply^code <> 0 !reply from server!

THEN ! we expect the reply code to be 0

CALL abend^with^my^abend^msg;

END; ! while

! Transaction completed successfully

CALL end^the^tmf^transaction;

RETURN true;

END; ! INT PROC process^transaction

?PAGE "READ FROM AN EDIT FILE"

! This procedure reads one line from the IN file.

! It returns the outcome of the read to the caller, with the
! variable error containing the error associated with the read.

INT PROC read^ (fnum, buf, read^count, error);

INT fnum;
STR .buf;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -38
Examples Pathsend Requester Example

INT read^count;
INT .error;

BEGIN
INT status;
DBL LineNum;

! If status >= 0, the read was successful and status contains the
! count of chars in the text line. Actual bytes transferred <=
! read^count. If status < 0, an unrecoverable error occurred.

error := 0;

IF (status := EDITREAD (edit^control^block, buf, read^count, LineNum)) >= 0

THEN ! read was successful
RETURN true;

IF (error := status) = -1
THEN ! end of file
RETURN false;

! Unexpected error reading the IN file, print an error msg and abend

CALL IO^error (fnum);

END; ! INT PROC read^

?PAGE "SEARCH THE CONTROL BLOCKS FOR RETRYABLE ERRORS"

! Search the list of control blocks for errors that can be retried.
! All Pathsend I/O that failed must be retryable (not just one send,
! but both) for the transaction to be considered retryable.

! When a transaction is retried, we re-use the existing list of

! control blocks.

INT PROC retryable^transaction (cb^list^head);

INT .EXT cb^list^head (control^block^template); ! ptr to hd of cb list

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;
INT transaction^is^retryable := false;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -39
Examples Pathsend Requester Example

WHILE @cb <> nil^addr

DO
BEGIN
IF cb.type = cb^type^pathsend AND cb.pathsend^error
THEN ! an error occurred on this Pathsend send

BEGIN
IF NOT cb.error^is^retryable
THEN ! the transaction isn't retryable

RETURN false;

transaction^is^retryable := true;
END;

@cb := @cb.next^cb;
END;

RETURN transaction^is^retryable;

END; ! INT PROC retryable^transaction

?PAGE "RETURN THE CONTROL BLOCK MEMORY TO THE POOL"

! After we are done using all the control blocks in a list, return
! the memory allocated for each control block.

PROC return^control^block^memory (cb^list^head);

INT .EXT cb^list^head (control^block^template);

BEGIN
INT .EXT cb (control^block^template) := @cb^list^head;
INT .EXT next^cb (control^block^template);
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

WHILE @cb <> nil^addr DO

BEGIN

! Save the address of the next element in the list before we

! return the current list element.

@next^cb := @cb.next^cb;

CALL PUTPOOL (PoolHead, cb);

IF <
THEN ! couldn't return the memory
BEGIN
sbuf ':=' "PUTPOOL FAILED TO RETURN MEMORY" -> @sp;

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;
END; ! if

@cb := @next^cb;
END; ! while

END; ! PROC return^control^block^memory

?PAGE "LOOK AT WHY SERVERCLASS_SEND_INFO_ FAILED"

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -40
Examples Pathsend Requester Example

! The following procedure is called from PROC analyze^send^error^233

! when SERVERCLASS_SEND_INFO_ fails. All these errors are
! non-retryable, and all are programming errors.

PROC ServerClass^Send^Info^error (error);

INT error;

BEGIN

CASE error OF
BEGIN

FEInvalidOp !2! ->

! Invalid segment in use or no segment in use and a param has xaddr

global^non^pathsend^error^msg ':='
"SERVERCLASS_SEND_INFO_ EXTENDED SEGMENT USAGE ERROR";

FEBoundsErr !22! ->

! Param out of bounds

global^non^pathsend^error^msg ':='
"SERVERCLASS_SEND_INFO_ PARAMETER OUT OF BOUNDS";

FEMissParam !29! ->

! A required param is missing

global^non^pathsend^error^msg ':='
"SERVERCLASS_SEND_INFO_ MISSING REQUIRED PARAMETER";

OTHERWISE ->

! No other errors should be returned from ServerClass_Send_Info_

global^non^pathsend^error^msg ':='
"SERVERCLASS_SEND_INFO_ UNEXPECTED ERROR";

END; ! case

END; ! PROC ServerClass^Send^Info^error

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -41
Examples Pathsend Requester Example

?PAGE "SET UP THE CONTROL BLOCKS FOR ONE TRANSACTION"

! Here we get memory for each control block (cb) used in the
! transaction, and we store data from the input record into the cb.
! The data we store is data necessary to execute that part of the
! transaction, a Pathsend send or msg log WRITE.

! This proc returns the address of the first element in the list of
! control blocks.

DBL PROC setup^control^blocks (input^buf, record^number);

INT .input^buf (breq^input^rec^template); ! input buffer
INT record^number; ! input file rec num

BEGIN
INT .EXT cb^list^head (control^block^template) := nil^addr;
INT .EXT cb (control^block^template);

! We pass into get^control^block the head of a list, and it creates

! a new cb and links it onto the end of the list. If we pass in a
! list head with nil address, then list head becomes the first
! element in the list.

@cb^list^head := get^control^block (cb^list^head);

! Store data into this control block, used for the first Pathsend
! send.

CALL store^control^block^info (cb^list^head, input^buf.server^request[0],

cb^type^pathsend, record^number);

! Get the second element in the list, used for the WRITE to the msg
! log file, and link it to the end of the list.

@cb := get^control^block (cb^list^head);

CALL store^control^block^info (cb, input^buf, cb^type^msg^log,

record^number);

! Get the third element in the list, used for the second Pathsend
! send, and link it to the end of the list.

@cb := get^control^block (cb^list^head);

CALL store^control^block^info (cb, input^buf.server^request[1],

cb^type^pathsend, record^number);

RETURN @cb^list^head;

END; ! DBL PROC setup^control^blocks

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -42
Examples Pathsend Requester Example

?PAGE "START THE TMF TRANSACTION"

! The following procedure is called by the MAIN proc to start the

! TMF transaction.

PROC start^the^tmf^transaction;

BEGIN
DBL trans^tag;
INT tmf^error;
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;

IF (tmf^error := BEGINTRANSACTION (trans^tag))

THEN ! file system error returned in tmf^error
BEGIN
sbuf ':=' "ERROR WITH BEGINTRANSACTION: " -> @sp;
CALL NUMOUT (sp, tmf^error, 10, 3);
@sp := @sp[3];

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

CALL ABEND;
END;

CALL write^trace^file (trace^begin^transaction);

END; ! PROC start^the^tmf^transaction

?PAGE "PROCESS THE STARTUP MSG"

! This procedure is called from the GUARDIAN 90 PROC INITIALIZER

! to save the startup msg, and OPEN the IN and OUT files.

PROC startup^proc (rucb, passthru, msg, msg^len, match) VARIABLE;

INT .rucb;
INT .passthru;
INT .msg (ci^startup);
INT msg^len;
INT match;

BEGIN
INT .buf [0:79];
STR .sbuf := @buf '<<' 1;
STR .sp;
INT FileCode;
INT status;

! Pass the startup msg back because PROC initialize needs access
! to the default volume and subvolume.

passthru ':=' msg for msg^len;

CALL OPEN (msg.infile, in^fnum, open^read^only);

IF <>
THEN ! error opening the input file
BEGIN
sbuf ':=' "FAILED TO OPEN IN FILE" ->@sp;

CALL WRITE (term^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -43
Examples Pathsend Requester Example

CALL IO^error (term^fnum);

CALL IO^error (in^fnum);

END;

! The IN file must be an EDIT file, so verify that now

CALL FILEINFO (in^fnum,,,,,,,,, FileCode);

IF FileCode <> 101

THEN ! illegal file type
BEGIN
sbuf ':=' "INFILE MUST BE TYPE EDIT (101)" -> @sp;

CALL WRITE (term^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL ABEND;
END;

! Setup the edit control block (a GLOBAL array) for calls

! to EDITREAD.

IF (status := EDITREADINIT (edit^control^block, in^fnum, 1024))

THEN ! print an error msg and abend
CALL IO^error (in^fnum);

CALL OPEN (msg.outfile, out^fnum);

IF <>
THEN ! error opening the output file
BEGIN
sbuf ':=' "FAILED TO OPEN OUT FILE" -> @sp;

CALL WRITE (term^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (term^fnum);

CALL IO^error (out^fnum);

END;

END; ! PROC startup^proc

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -44
Examples Pathsend Requester Example

?PAGE "STORE DATA IN THE CONTROL BLOCK"

! This proc stores data to make one Pathsend send or one write to
! the message log file. The data is from one input record, and is
! stored in the control block that's passed into this procedure.

PROC store^control^block^info (cb, data^buf, cb^type, record^number);

INT .EXT cb (control^block^template);
STR .data^buf (pathsrv^request^template);
INT cb^type;
INT record^number;

BEGIN
IF cb^type <> cb^type^pathsend AND cb^type <> cb^type^msg^log
THEN ! assertion failed
CALL abend^with^my^abend^msg;

cb.type := cb^type;
cb.record^number := record^number;

IF cb.type = cb^type^pathsend
THEN ! store data necessary to do a Pathsend send
BEGIN
@cb.pathmon^system^and^process^name :=
@data^buf.pathmon^system^and^process^name;

@cb.serverclass^name := @data^buf.server^class;
END

ELSE ! the cb is for a write to msg log

BEGIN
@cb.input^data^buf := @data^buf '>>' 1;
END;

END; ! PROC store^control^block^info

?PAGE "VALIDATE A REQUEST RECORD"

! A valid request is in the format of breq^input^rec^template, and

! contains data needed to make 2 Pathsend sends. The data is a
! PATHMON name (either explicit or an ASSIGN) and a server class
! name.

! If an ASSIGN is specified, then this proc looks up the ASSIGN name

! in our ASSIGN table and puts the corresponding file name into the
! PATHMON system and process name field of the input record.

! This proc returns the outcome of the validation check to the

! caller.

INT PROC validate^breq^input (input^rec);

INT .input^rec (breq^input^rec^template);

BEGIN
INT i;
INT .buf [0:79] := [80*[" "]];
STR .sbuf := @buf '<<' 1;
STR .sp;
STR .logical^name[0:30];
INT .assign^ (ci^assign);

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -45
Examples Pathsend Requester Example

?PAGE "SUBPROC TO WRITE A VALIDATION ERROR MSG"

!
! If a record is invalid, write an error msg to the error log file
!
SUBPROC output^msg;

BEGIN

CALL WRITE (error^log^fnum, buf, @sp '-' @sbuf);

IF <
THEN ! print an error msg and abend
CALL IO^error (error^log^fnum);

! Save the error msg so we can include it in the output record

global^non^pathsend^error^msg ':=' sbuf FOR @sp '-' @sbuf;

END; ! SUBPROC output^msg

?PAGE "BEGIN BODY OF VALIDATE BREQ INPUT"

FOR i := 0 to 1 DO
BEGIN

IF input^rec.server^request[i].server^class = [15*[" "]]

THEN ! Invalid request
BEGIN
sbuf ':=' "INVALID REQUEST: SERVER CLASSES MUST BE SPECIFIED" ->
@sp;
CALL output^msg;
RETURN false;
END;

IF (input^rec.server^request[i].pathmon^assign^name = [31*[" "]] AND

input^rec.server^request[i].pathmon^system^and^process^name =
[15*[" "]])
THEN ! invalid request
BEGIN
sbuf ':=' "INVALID REQUEST: ASSIGN NAME " -> @sp;
sp ':=' "OR SYSTEM AND PROCESS NAME MUST BE SPECIFIED" -> @sp;

CALL output^msg;
RETURN false;
END;

IF (input^rec.server^request[i].pathmon^assign^name <> [31*[" "]] AND

input^rec.server^request[i].pathmon^system^and^process^name <>
[15*[" "]])

THEN ! invalid request

BEGIN
sbuf ':=' "INVALID REQUEST: BOTH ASSIGN AND SYSTEM " -> @sp;
sp ':=' "AND PROCESS NAMES CAN'T BE SPECIFIED" -> @sp;

CALL output^msg;
RETURN false;
END;

! No format errors detected with server class names and PATHMON

! ASSIGNs and process names.

IF input^rec.server^request[i].pathmon^assign^name <> [31*[" "]]

THEN ! try to get the ASSIGN name
BEGIN
logical^name ':=' input^rec.server^request[i].pathmon^assign^name

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -46
Examples Pathsend Requester Example

FOR $OCCURS (input^rec.server^request.pathmon^assign^name);

RSCAN logical^name[30] WHILE " " -> @sp;

IF NOT (@assign^ := get^assign (logical^name, @sp[1] '-' @logical^name))

THEN ! the ASSIGN isn't in our ASSIGN table
BEGIN
sbuf ':=' "ASSIGN '" -> @sp;
sp ':=' logical^name FOR 31 -> @sp;
RSCAN sp[-1] WHILE " " -> @sp;
sp[1] ':=' "' NOT FOUND" -> @sp;

CALL output^msg;
RETURN false;
END;

! Found the ASSIGN, save the PATHMON name

CALL FNAMECOLLAPSE (assign^.tandemfilename.volume,

input^rec.server^request[i].pathmon^system^and^process^name);

END; ! look up the ASSIGN

END; ! for loop

RETURN true;

END; ! INT PROC validate^breq^input

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -47
Examples Pathsend Requester Example

?PAGE "GET MY NAME OR PROCESS ID"

!
! The following procedure gets my name or process id so that
! error msg text can be preceded by my process name or cpu,pin.
!
PROC who^am^i;

BEGIN
STR .spid := @my^processid '<<' 1;

! MYPID returns my cpu,pin and is input to GETCRTPID, which returns

! my^processid in this form:
! [0:2] = process name or creation timestamp
! [3].<4:7> = cpu
! [3].<8:15> = pin

CALL GETCRTPID (MYPID, my^processid);

IF spid <> "$"

THEN ! I am not a named process, convert cpu,pin for output
BEGIN
CALL NUMOUT (spid, my^processid[3].<4:7>, 10, 2);
spid[2] := ",";
CALL NUMOUT (spid[3], my^processid[3].<8:15>, 10, 3);
END;

END; ! PROC who^am^i

?PAGE "WRITE TO THE TRACE FILE"

! This procedure prefaces each trace msg with this process's

! process id, and does a waited WRITE to the trace file. The trace
! msg written depends on the function passed in.

PROC write^trace^file (function, record^number) VARIABLE;

INT function; ! what action we are tracing
INT record^number; ! input file record number (optional param)

BEGIN
INT .write^buf[0:79] := [80*[" "]];
STR .swrite^buf := @write^buf '<<' 1;
STR .sp;
STR .spid := @my^processid '<<' 1;

IF NOT $PARAM (function)

THEN ! missing required param
CALL abend^with^my^abend^msg;

IF trace^fnum <= 0
THEN ! user doesn't want to trace
RETURN;

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -48
Examples Pathsend Requester Example

! Let the user know who this msg is from

swrite^buf ':=' spid FOR 6 BYTES -> @sp;

RSCAN sp[-1] WHILE " " -> @sp;
@sp := @sp[1];
sp ':=' ": " -> @sp;

! Include the record number

IF $PARAM (record^number)
THEN ! include it in the msg
BEGIN
sp ':=' "RECORD #" -> @sp;
CALL NUMOUT (sp, record^number, 10, 3);
@sp := @sp[4];
END;

CASE function OF
BEGIN
!0! sp ':=' "READ ONE REQUEST MSG FROM INPUT FILE" -> @sp;
!1! sp ':=' "INITIATED ONE PATHSEND SEND" -> @sp;
!2! sp ':=' "INITIATED WRITE TO MESSAGE-LOG-FILE" -> @sp;
!3! sp ':=' "BEGIN TRANSACTION" -> @sp;
!4! sp ':=' "END TRANSACTION" -> @sp;
!5! sp ':=' "ABORT TRANSACTION" -> @sp;
!6! sp ':=' "CANCELREQ" -> @sp;
!7! @sp := @swrite^buf; ! write a blank line
!8! sp ':=' "I/O COMPLETED SUCCESSFULLY" -> @sp;
!9! sp ':=' "AWAITIO TIMED OUT" -> @sp;
!10! sp ':=' "I/O COMPLETED WITH AN ERROR" -> @sp;
!11! sp ':=' "RETRYING THE TRANSACTION" -> @sp;
!12! sp ':=' "MAX-RETRIES EXCEEDED, TRANSACTION ABORTED" -> @sp;

!*! OTHERWISE call abend^with^my^abend^msg;

END; ! Case

CALL WRITE (trace^fnum, write^buf, @sp '-' @swrite^buf);

IF <
THEN ! print an error msg and abend
CALL IO^error (trace^fnum);

END; ! PROC write^trace^file

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -49
Examples Pathsend Requester Example

?PAGE "THE BREQ PROGRAM'S MAIN PROCEDURE"

! This procedure calls the initialize routine to start up, sets up a

! memory pool, and loops reading the IN file.

! Input records are validated, control blocks are allocated, the

! transaction is initiated and then completed, and control blocks
! are de-allocated.

! If we fail to execute the transaction and the error is retryable,

! we retry it MAX-RETRIES times, where MAX-RETRIES is a param msg
! defined at the TACL prompt.

PROC breq^program MAIN;

BEGIN
STRUCT .input^buf (breq^input^rec^template); ! buf to hold input record
INT .EXT cb^list^head (control^block^template) := nil^addr;
!first cb in linked list.
INT error := 0; ! used with read of input file
INT record^number := 0; ! input file record number we're processing
INT retry^count; ! # times we retried transaction a failure

LIT global^msg^len = $OCCURS(breq^output^rec^template.non^send^error^msg);

! len of non-Pathsend error msg buffer
CALL initialize;

! Define the memory pool that we use to store the control blocks
! (cb's), maintained as a linked list in the upper 32K. Cb's are
! allocated when we process one transaction, and are deallocated
! when we complete one transaction. A transaction is driven by data
! in one record of the input file.

CALL create^the^memory^pool;

! Loop until end of IN file, reading one record and doing the
! transaction associated with that record. A transaction is two
! Pathsend sends and one WRITE to the message log file.

WHILE NOT error DO

BEGIN
input^buf ':=' " " & input^buf FOR (($LEN (input^buf) + 1) / 2) - 1;

global^non^pathsend^error^msg ':=' global^msg^len * [" "];

IF read^ (in^fnum, input^buf, $LEN (input^buf), error)

THEN ! successfully read 1 record
BEGIN

! Record^Number is the input file record number we're

! processing, and is used in the output to associate the
! outcome of a transaction with the record used to generate
! the transaction.

record^number := record^number + 1;

! Tracing is enabled by specifying a trace file ASSIGN. If

! enabled, a trace record is written to the trace file when
! this program performs certain actions. For example, reading
! an input record, and starting or aborting a transaction.

CALL write^trace^file (trace^write^blank^line);

CALL write^trace^file (trace^read^input, record^number);

! Besides validating the input record, the validate routine

! looks up ASSIGN names for PATHMON system and process names

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -50
Examples Pathsend Requester Example

! specified in the input record, and puts the process name

! associated with the ASSIGN into the PATHMON system and
! process name field of the input record.

IF validate^breq^input (input^buf)
THEN ! input record is valid
BEGIN

! A valid request record is in the format of

! breq^input^rec^template, and contains data needed to
! make 2 Pathsend sends. The data is a PATHMON name
! (either explicit or an ASSIGN) and a server class name.

! Allocate one control block (cb) each for the two

! Pathsend sends, and one cb for the message log write.
! The cb's are link listed together, and the head of the
! list is returned. Data stored in the cb's is from the
! input record.

@cb^list^head := setup^control^blocks (input^buf, record^number);

! Process^transaction starts the tmf transaction,

! initiates the I/O, completes the I/O, and ends the tmf
! transaction.

! If an I/O fails, then process^transaction aborts the tmf

! transaction, cancels the outstanding I/O, and returns
! false. In this case we will retry the transaction
! MAX-RETRIES times if the transaction is retryable.

retry^count := 0;

IF NOT process^transaction (cb^list^head)

THEN ! the transaction failed
BEGIN

! Retryable^transaction is a proc that returns true

! if all I/O that failed is retryable.

WHILE retry^count < max^retries AND

retryable^transaction (cb^list^head)
DO
BEGIN
CALL write^trace^file (trace^retry^transaction);
CALL process^transaction (cb^list^head);

retry^count := retry^count + 1;
END;

If retry^count >= max^retries

THEN
! the transaction failed after retrying max^retries times
CALL write^trace^file (trace^max^retries^exceeded);

END; ! if

END; ! input record is valid

! Write the output to the OUT file

CALL output^the^results (cb^list^head);

! Return all control block memory to the pool, and set the
! the first element in the list to nil address.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -51
Examples Pathsend Requester Example

CALL return^control^block^memory (cb^list^head);

@cb^list^head := nil^addr;

END; ! if read^

END; ! while not error

IF error <> e^eof

THEN ! print an error msg and abend
CALL IO^error (in^fnum);

! end of file input file

CALL CLOSE (in^fnum);

CALL CLOSE (out^fnum);
CALL CLOSE (msg^log^fnum);
CALL CLOSE (error^log^fnum);

IF trace^fnum > 0 THEN

CALL CLOSE (trace^fnum);

CALL STOP;

END; ! PROC breq^program

?NOMAP

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -52
Examples Nested Server Example

Nested Server Example

Example B-2, PATHSRV, is a context-free nested server coded in COBOL85.
PATHSRV uses Pathsend calls within a server to access another server.

Example B-2. Context-Free Server Program

* @@@ START COPYRIGHT @@@
* Tandem Confidential: Need to Know only
* Copyright (c) 1980-1985, 1987-1995, Tandem Computers Incorporated
* Protected as an unpublished work.
* All Rights Reserved.
*
* The computer program listings, specifications, and documentation
* herein are the property of Tandem Computers Incorporated and shall
* not be reproduced, copied, disclosed, or used in whole or in part
* for any reason without the prior express written permission of
* Tandem Computers Incorporated.
* @@@ END COPYRIGHT @@@
?SYMBOLS

?SEARCH $SYSTEM.SYSTEM.COBOLLIB
?CONSULT $SYSTEM.SYSTEM.COBOLEX0

?SAVE ASSIGNS 50, PARAM

IDENTIFICATION DIVISION.
PROGRAM-ID. PATHSRV.

* THE COBOL85 SERVER PATHSRV READS A REQUEST FROM $RECEIVE AND

* RETURNS ITS GUARDIAN PROCESS-ID AND THAT OF A SUBSIDIARY SERVER
* CLASS (IF ANY) IDENTIFIED ON THE REQUEST MESSAGE. THE PATHMON OF
* THE SUBSIDIARY SERVER CLASS IS IDENTIFIED IN THE REQUEST MESSAGE IN
* ONE OF THREE WAYS AS FOLLOWS:
*
* - BY AN ASSIGN NAME WHICH REFERENCES AN ASSIGN WITH THE
* PATHMON SYSTEM AND PROCESS NAME.
*
* - WITH AN EXPLICIT SYSTEM AND PROCESS NAME OF PATHMON.
*
* - IF BOTH THE ASSIGN NAME AND THE EXPLICIT SYSTEM AND PROCESS
* NAME IN THE REQUEST MESSAGE ARE BLANK THEN THE PATHMON FOR THE
* SUBSIDIARY SERVER CLASS DEFAULTS TO BE THE SAME PATHMON AS
* THAT WHICH OWNS THE SERVER MAKING THE PATHSEND SEND. THIS IS
* POSSIBLE AS LONG AS THE SENDING SERVER IS NOT ASSOCIATIVE (IN
* WHICH CASE ITS ANCESTOR MAY NOT BE A PATHMON).
*
* THE FIRST AND THIRD OPTIONS DEMONSTRATE WAYS TO AVOID HARD
* CODING A PATHMON SYSTEM AND PROCESS IN A PROGRAM THAT DOES A
* PATHSEND SEND.
*
* IN THE EVENT OF A PROBLEM AN ERROR MESSAGE AND A FLAG TO INDICATE
* THAT A TMF ABORT IS REQUIRED ARE ALSO RETURNED TO THE REQUESTER.
*

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -53
Examples Nested Server Example

* ANY ERRORS THAT CAN NOT BE REPORTED BY PATHSRV BACK TO ITS

* REQUESTER ARE WRITTEN TO AN ERROR LOG FILE. THIS ENTRY SEQUENCED
* FILE WITH 132 BYTES RECORDS MUST EXIST PRIOR TO THE EXECUTION OF
* PATHSRV. ITS ASSIGN NAME IS ERROR-LOG-FILE.

ENVIRONMENT DIVISION.
CONFIGURATION SECTION.
SOURCE-COMPUTER. TANDEM.
OBJECT-COMPUTER. TANDEM.

INPUT-OUTPUT SECTION.

FILE-CONTROL.

SELECT MESSAGE-IN-FILE
ASSIGN TO $RECEIVE
FILE STATUS IS WS-FILE-STATUS.

SELECT MESSAGE-OUT-FILE
ASSIGN TO $RECEIVE
FILE STATUS IS WS-FILE-STATUS.

SELECT ERROR-LOG-FILE
ASSIGN TO "ERRORLOG"
ORGANIZATION IS SEQUENTIAL
ACCESS IS SEQUENTIAL
FILE STATUS IS WS-FILE-STATUS.

RECEIVE-CONTROL.

TABLE OCCURS 10 TIMES

SYNCDEPTH 1.

/
DATA DIVISION.

FILE SECTION.

FD MESSAGE-IN-FILE
DATA RECORD IS PATHSRV-REQUEST.

01 PATHSRV-REQUEST.
03 SUBSIDIARY-SERVER.
05 PATHMON-ASSIGN-NAME PIC X(31).
05 PATHMON-SYSTEM-AND-PROCESS PIC X(15).
05 SERVER-CLASS PIC X(15).

FD MESSAGE-OUT-FILE
DATA RECORD IS PATHSRV-REPLY.

01 PATHSRV-REPLY.
03 REPLY-CODE PIC S9(4) COMP.
03 THIS-SERVER.
05 SYSTEM-NAME PIC X(8).
05 PROCESS-NAME PIC X(8).
03 SUBSIDIARY-SERVER.
05 SYSTEM-NAME PIC X(8).
05 PROCESS-NAME PIC X(8).
03 TMF-ABORT-REQUIRED PIC X.
03 ERROR-MESSAGE.
05 PATHSEND-ERROR PIC X(78).
05 FILE-SYSTEM-ERROR PIC X(78).
03 NON-SEND-ERROR-MESSAGE PIC X(78).

FD ERROR-LOG-FILE

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -54
Examples Nested Server Example

DATA RECORD IS ERROR-LOG-REC.

01 ERROR-LOG-REC PIC X(132).

WORKING-STORAGE SECTION.

01 WS-FILE-STATUS PIC X(2) VALUE ZERO.

88 WS-CLOSE-FROM-REQUESTER VALUE "10".

01 WS-FILE-SYSTEM-ERROR-MESSAGE.
03 FILLER PIC X(25)
VALUE "FILE SYSTEM ERROR. FILE: ".
03 WS-LOGICAL-FILE-NAME PIC X(31).
03 FILLER PIC X(13)
VALUE "FILE STATUS: ".
03 WS-FILE-STATUS-ERROR-MESSAGE PIC 99.

01 WS-NUMERIC-DISPLAY PIC S9(9).

* YOU CAN USE THE FOLLOWING WS-ASSIGN TABLE TO STORE UP TO FIFTY

* ASSIGNS. THESE ASSIGNS ARE FOR THE PATHMON SYSTEM AND PROCESS
* NAMES OF SUBSIDIARY SERVERS. IF YOU USE PATHMON-ASSIGN-NAME IN THE
* REQUEST MESSAGE (PATHSRV-REQUEST), THEN THE ASSOCIATED PATHMON
* SHOULD BE IN THIS TABLE. THIS IS ONE WAY OF AVOIDING HARD-CODING
* PATHMON SYSTEM AND PROCESS NAMES.

01 WS-ASSIGN-TABLE.
03 WS-ASSIGN-NAME PIC X(31) OCCURS 50.
03 WS-SYSTEM-AND-PROCESS PIC X(15) OCCURS 50.
03 WS-NUMBER-OF-ENTRIES PIC S9(4) COMP VALUE ZERO.
03 WS-INDEX PIC S9(4) COMP VALUE ZERO.

* THE FOLLOWING PARAMETERS ARE USED WHEN CALLING GETASSIGNTEXT

01 WS-GETASSIGNTEXT-PARAM.
03 WS-PORTION PIC X(30).
03 WS-TEXT PIC X(32).
03 WS-MESSAGE-NUMBER PIC S9(4) COMP VALUE 1.
03 WS-RESULT PIC S9(4) COMP VALUE ZERO.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -55
Examples Nested Server Example

* PROCESSINFO IS CALLED TO SEE IF THE MOM OF THE SERVER PROCESS IS A

* PATHMON. (THIS IDENTIFIES WHETHER THE SERVER IS ASSOCIATIVE.) AN
* ERROR NUMBER (WS-ERROR) MY BE RETURNED FOR A CALL TO PROCESSINFO.

01 WS-PROCESSINFO-PARAM.
03 WS-ERROR PIC S9(4) COMP.

* THE FOLLOWING PARAMETERS ARE USED IN A CALL TO LOOKUPPROCESSNAME,

* WHICH LOCATES THE MOM OF THE CURRENT SERVER PROCESS (TO SEE IF IT'S
* A PATHMON).

01 WS-LOOKUPPROCESSNAME-PARAM.
03 WS-PROCESS-NAME PIC X(6).
03 FILLER PIC X(4).
03 WS-ANCESTOR-PROCESS-ID PIC X(8).

* THE FOLLOWING PARAMETERS ARE USED IN A CALL TO SERVERCLASS_SEND_.

01 WS-SERVERCLASS-SEND-PARAM.
03 WS-ERROR PIC S9(4) COMP VALUE ZERO.
03 WS-PATHMON-PROCESS-NAME PIC X(15).
03 WS-PATHMON-PROCESS-NAME-LEN PIC S9(4) COMP VALUE 15.
03 WS-SERVER-CLASS-NAME PIC X(15).
03 WS-SERVER-CLASS-NAME-LEN PIC S9(4) COMP VALUE 15.
03 WS-MESSAGE-BUFFER PIC X(300).
03 WS-REQUEST-LEN PIC S9(4) COMP VALUE 63.
03 WS-MAXIMUM-REPLY-LEN PIC S9(4) COMP VALUE 300.
03 WS-ACTUAL-REPLY-LEN PIC S9(4) COMP.
03 WS-TIMEOUT PIC S9(9) COMP VALUE 12000.

* THE FOLLOWING PARAMETERS ARE USED IN A CALL TO SERVERCLASS_SEND_INFO_

01 WS-SERVERCLASS-SEND-INFO-PARAM.
03 WS-ERROR PIC S9(4) COMP VALUE ZERO.
03 WS-PATHSEND-ERROR PIC S9(4) COMP.
03 WS-FILE-SYSTEM-ERROR PIC S9(4) COMP.

* THE FOLLOWING PARAMETERS ARE USED IN A CALL TO MYPID AND MYCRTPID,

* WHICH ARE USED TO IDENTIFY THE CURRENT PROCESS. THE PROCESS NAME IS
* RETURNED TO THE REQUESTER IN THE PATHSRV-REPLY MESSAGE.

01 WS-MY-PROCESS.
03 WS-CPU-PIN PIC S9(4) COMP.
03 WS-PROCESS-ID.
05 WS-PROCESS-NAME PIC X(6).
05 FILLER PIC X(2).
03 WS-SYSTEM-NUMBER PIC 9(4) COMP.
03 WS-SYSTEM-NAME PIC X(8).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -56
Examples Nested Server Example

* THE FOLLOWING PARAMETERS ARE USED IN CALLS TO PROCESSINFO AND

* MYSYSTEMNUMBER, WHICH ARE USED TO IDENTIFY THE MOM OF THE SERVER
* PROCESS.

01 WS-MOM-PROCESS.
03 WS-PROCESS-ID-GENERIC.
05 WS-BYTE-1 PIC X.
88 WS-LOCAL VALUE "$".
88 WS-NETWORK VALUE "\".
05 FILLER PIC X(5).
05 WS-CPU-PIN PIC S9(4) COMP.
03 WS-PROCESS-ID-LOCAL REDEFINES WS-PROCESS-ID-GENERIC.
05 WS-PROCESS-ID PIC X(6).
05 WS-CPU-PIN PIC S9(4) COMP.
03 WS-PROCESS-ID-NETWORK REDEFINES WS-PROCESS-ID-GENERIC.
05 WS-BACKSLASH PIC X.
05 WS-SYSTEM-NUMBER-1-BYTE PIC X.
05 WS-PROCESS-ID PIC X(4).
05 WS-CPU-PIN PIC S9(4) COMP.
03 WS-SYSTEM-NUMBER PIC 9(4) COMP.
03 WS-SYSTEM-NUMBER-2-BYTES REDEFINES WS-SYSTEM-NUMBER.
05 FILLER PIC X.
05 WS-BYTE-2 PIC X.
03 WS-PROGRAM-FILENAME.
05 FILLER PIC X(16).
05 WS-FILE PIC X(8).
03 WS-SYSTEM-NAME PIC X(8).
03 WS-PROCESS-NAME PIC X(6).
03 WS-SYSTEM-AND-PROCESS PIC X(15).

* THIS FLAG INDICATES WHETHER THE MOM OF THE SERVER PROCESS IS A

* PATHMON. THE VALUE "N" MEANS THAT MOM IS A PATHMON, THE VALUE "Y"
* MEANS THAT MON IS NOT A PATHMON, IN WHICH CASE THE PATHSRV-REQUEST
* MESSAGE MUST IDENTIFY THE PATHMON WITH AN EXPLICIT NAME OR AN
* ASSIGN. (THE DEFAULT TO THE SERVER PROCESS MOM CANNOT BE TAKEN.)

01 WS-SERVER-IS-ASSOCIATIVE PIC X.

* THE PATHSRV-REQUEST MESSAGE FROM THE REQUESTER IS VALIDATED IN THIS

* PROGRAM. IF IT IS NOT VALID, THIS FLAG IS SET TO "N"

01 WS-VALID-PATHSRV-REQUEST PIC X.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -57
Examples Nested Server Example

* IF PATHSRV-REQUEST SPECIFIES A SERVER CLASS, THEN A PATHSEND

* REQUEST IS MADE TO A SUBSIDIARY SERVER. THE FOLLOWING GROUP DATA
* ITEMS ARE THE REQUEST AND REPLY MESSAGE LAYOUTS. THE REQUEST IS
* ALWAYS SPACE FILLED BECAUSE THE SUBSIDIARY SERVER IS NEVER ASKED TO
* MAKE A REQUEST OF ANOTHER SERVER.

01 SUBSIDIARY-REQUEST.
03 SUBSIDIARY-SERVER.
05 PATHMON-ASSIGN-NAME PIC X(31).
05 PATHMON-SYSTEM-AND-PROCESS PIC X(15).
05 SERVER-CLASS PIC X(15).

01 SUBSIDIARY-REPLY.
03 REPLY-CODE PIC S9(4) COMP.
03 THIS-SERVER.
05 SYSTEM-NAME PIC X(8).
05 PROCESS-NAME PIC X(8).
03 SUBSIDIARY-SERVER.
05 SYSTEM-NAME PIC X(8).
05 PROCESS-NAME PIC X(8).
03 TMF-ABORT-REQUIRED PIC X.
03 ERROR-MESSAGE.
05 PATHSEND-ERROR PIC X(78).
05 FILE-SYSTEM-ERROR PIC X(78).
03 NON-SEND-ERROR-MESSAGE PIC X(78).

* THE FOLLOWING FLAG INDICATES THE SUCCESS OR FAILURE OF A CALL TO

* SERVERCLASS_SEND_. "Y" MEANS SUCCESS. "N" MEANS FAILURE.

01 WS-SERVERCLASS-SEND-OKAY PIC X.

/
PROCEDURE DIVISION.

DECLARATIVES.

DECL-MESSAGE-IN-FILE SECTION.
USE AFTER ERROR PROCEDURE ON MESSAGE-IN-FILE.
MOVE "MESSAGE-IN-FILE" TO WS-LOGICAL-FILE-NAME.
MOVE WS-FILE-STATUS TO WS-FILE-STATUS-ERROR-MESSAGE.

DECL-MESSAGE-OUT-FILE SECTION.
USE AFTER ERROR PROCEDURE ON MESSAGE-OUT-FILE.
MOVE "MESSAGE-OUT-FILE" TO WS-LOGICAL-FILE-NAME.
MOVE WS-FILE-STATUS TO WS-FILE-STATUS-ERROR-MESSAGE.

DECL-ERROR-LOG-FILE SECTION.
USE AFTER ERROR PROCEDURE ON ERROR-LOG-FILE.
MOVE "ERROR-LOG-FILE" TO WS-LOGICAL-FILE-NAME.
MOVE WS-FILE-STATUS TO WS-FILE-STATUS-ERROR-MESSAGE.

END DECLARATIVES.

0000-MAINLINE.

* THIS PARAGRAPH CONTAINS THE OVERALL LOGIC OF THE PROGRAM. AFTER

* INITIALIZATION, $RECEIVE IS READ. IF THE READ IS SUCCESSFUL, THE
* REQUEST MESSAGE IS PROCESSED IN 0200-PROCESS-TRANSACTION AND THE
* REPLY (PATHSRV-REPLY) IS WRITTEN TO $RECEIVE. $RECEIVE IS THEN READ
* AGAIN. THE LOOP CONTINUES UNTIL A CLOSE IS RECEIVED FROM THE

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -58
Examples Nested Server Example

* REQUESTER ON $RECEIVE.
*
* IF AN ERROR IS FOUND WHILE READING $RECEIVE, A MESSAGE IS WRITTEN
* TO THE ERROR-LOG-FILE AND THE PROGRAM STOPS.

PERFORM 0100-INITIALIZE.
PERFORM UNTIL WS-CLOSE-FROM-REQUESTER
READ MESSAGE-IN-FILE
IF WS-FILE-STATUS = ZERO
PERFORM 0200-PROCESS-TRANSACTION
WRITE PATHSRV-REPLY
ELSE
IF NOT WS-CLOSE-FROM-REQUESTER
MOVE WS-FILE-SYSTEM-ERROR-MESSAGE TO ERROR-LOG-REC
PERFORM 9000-WRITE-ERROR-LOG-REC
PERFORM 9900-STOP-RUN
END-IF
END-IF
END-PERFORM
PERFORM 0300-CLOSEDOWN
PERFORM 9900-STOP-RUN.

0100-INITIALIZE.
OPEN EXTEND ERROR-LOG-FILE SHARED.
OPEN INPUT MESSAGE-IN-FILE.
OPEN OUTPUT MESSAGE-OUT-FILE.
PERFORM 0400-BUILD-ASSIGN-TABLE.
PERFORM 0410-GET-MY-SYSTEM-PROCESS.
PERFORM 0420-GET-MOM-SYSTEM-PROCESS.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -59
Examples Nested Server Example

0200-PROCESS-TRANSACTION.

* THIS PARAGRAPH PROCESSES ONE MESSAGE FROM $RECEIVE. THE FORMAT OF

* THE MESSAGE IS PATHSRV-REQUEST. FIRST THE PATHSRV-REPLY MESSAGE IS
* INITIALIZED. THE SYSTEM AND PROCESS NAME OF THE CURRENT SERVER
* PROCESS ARE ALWAYS RETURNED TO THE REQUESTER IN THE PATHSRV-REPLY
* MESSAGE.
*
* IF THE SUBSIDIARY SERVER CLASS IS SPECIFIED IN THE REQUEST, THEN A
* PATHSEND SEND IS DONE TO THAT SUBSIDIARY SERVER CLASS.

MOVE SPACES TO PATHSRV-REPLY.

MOVE ZERO TO REPLY-CODE OF PATHSRV-REPLY.
MOVE "N" TO TMF-ABORT-REQUIRED OF PATHSRV-REPLY.
MOVE WS-PROCESS-NAME OF WS-MY-PROCESS
TO PROCESS-NAME OF THIS-SERVER OF PATHSRV-REPLY.
MOVE WS-SYSTEM-NAME OF WS-MY-PROCESS
TO SYSTEM-NAME OF THIS-SERVER OF PATHSRV-REPLY.

* IF SERVER-CLASS IN SPECIFIED IN THE REQUEST MESSAGE THEN PROCESSING

* IS DONE TO PREPARE FOR A PATHSEND SEND TO A SUBSIDIARY SERVER.

IF SERVER-CLASS OF PATHSRV-REQUEST NOT = SPACES

MOVE "Y" TO WS-VALID-PATHSRV-REQUEST
PERFORM 0440-VALIDATE-PATHSRV-REQUEST
IF WS-VALID-PATHSRV-REQUEST = "Y"

* IF THE PATHMON-ASSIGN NAME IS SPECIFIED, THEN THE PATHMON SYSTEM

* AND PROCESS NAME IS LOOKED UP IN THE TABLE OF ASSIGNs. THIS PATHMON
* SYSTEM AND PROCESS NAME IS USED FOR THE PATHSEND SEND.

IF PATHMON-ASSIGN-NAME OF PATHSRV-REQUEST NOT = SPACES

PERFORM 0450-LOOKUP-PATHMON-ASSIGN
ELSE

* IF THE PROGRAM LOGIC ARRIVES HERE, THEN AN ASSIGN TO IDENTIFY THE

* PATHMON IS NOT SPECIFIED. (IF AN EXPLICIT PATHMON SYSTEM AND
* PROCESS NAME ARE SPECIFIED THEN THEY WILL BE USED FOR THE PATHSEND
* SEND.)

IF PATHMON-SYSTEM-AND-PROCESS OF PATHSRV-REQUEST NOT = SPACES

MOVE PATHMON-SYSTEM-AND-PROCESS OF PATHSRV-REQUEST
TO WS-PATHMON-PROCESS-NAME OF WS-SERVERCLASS-SEND-PARAM
ELSE

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -60
Examples Nested Server Example

* IF THE PROGRAM LOGIC ARRIVES HERE, THEN NEITHER AN ASSIGN NOR AN

* EXPLICIT SYSTEM AND PROCESS NAME ARE USED TO IDENTIFY THE PATHMON
* TO WHICH THE PATHSEND SEND SHOULD BE DONE. IF THE MOM OF THE
* CURRENT SERVER PROCESS IS A PATHMON (I.E. WS-SERVER-IS-ASSOCIATIVE
* = "N") THEN ITS SYSTEM AND PROCESS NAME IS USED FOR THE PATHSEND
* SEND.

IF WS-SERVER-IS-ASSOCIATIVE = "N"
MOVE WS-SYSTEM-AND-PROCESS OF WS-MOM-PROCESS
TO WS-PATHMON-PROCESS-NAME
OF WS-SERVERCLASS-SEND-PARAM
ELSE
MOVE "PATHMON NOT KNOWN - SERVER IS ASSOCIATIVE" TO
NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
MOVE "N" TO WS-VALID-PATHSRV-REQUEST
END-IF
END-IF
END-IF
END-IF

IF WS-VALID-PATHSRV-REQUEST = "Y"

* THE REQUEST MESSAGE TO THE SUBSIDIARY SERVER IS ALWAYS SPACES. THE

* SUBSIDIARY SERVER IS NOT ASKED TO MAKE A REQUEST OF ANOTHER (THIRD
* LEVEL) SERVER. RATHER IT IS JUST BEING ASKED FOR ITS OWN GUARDIAN
* PROCESS-ID.

MOVE SPACES TO SUBSIDIARY-REQUEST

PERFORM 0460-SEND-TO-SUBSIDIARY-SERVER
IF WS-SERVERCLASS-SEND-OKAY = "Y"
MOVE SYSTEM-NAME OF THIS-SERVER OF SUBSIDIARY-REPLY
TO SYSTEM-NAME OF SUBSIDIARY-SERVER OF PATHSRV-REPLY
MOVE PROCESS-NAME OF THIS-SERVER OF SUBSIDIARY-REPLY
TO PROCESS-NAME OF SUBSIDIARY-SERVER OF PATHSRV-REPLY
ELSE
MOVE "Y" TO TMF-ABORT-REQUIRED OF PATHSRV-REPLY
END-IF
ELSE
MOVE "Y" TO TMF-ABORT-REQUIRED OF PATHSRV-REPLY
END-IF
END-IF.

0300-CLOSEDOWN.
CLOSE ERROR-LOG-FILE.
CLOSE MESSAGE-IN-FILE.
CLOSE MESSAGE-OUT-FILE.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -61
Examples Nested Server Example

0400-BUILD-ASSIGN-TABLE.

* THIS PARAGRAPH LOADS WS-ASSIGN-TABLE FROM THE STARTUP ASSIGN

* MESSAGES.
*
* WS-ASSIGN-TABLE IS USED TO STORE UP TO FIFTY ASSIGN's. THESE
* ASSIGN's ARE FOR PATHMON SYSTEM AND PROCESS NAMES OF SUBSIDIARY
* SERVERS. IF PATHMON-ASSIGN-NAME IS USED IN THE REQUEST MESSAGE
* (PATHSRV-REQUEST) THEN THE ASSOCIATED PATHMON SHOULD BE IN THIS
* TABLE. THIS IS ONE WAY OF AVOIDING HARD-CODING PATHMON SYSTEM AND
* PROCESS NAMES.

PERFORM VARYING WS-MESSAGE-NUMBER OF WS-GETASSIGNTEXT-PARAM

FROM 1 BY 1
UNTIL WS-RESULT OF WS-GETASSIGNTEXT-PARAM = -1 OR
WS-MESSAGE-NUMBER OF WS-GETASSIGNTEXT-PARAM = 51
MOVE "LOGICALNAME" TO WS-PORTION OF WS-GETASSIGNTEXT-PARAM
ENTER "GETASSIGNTEXT"
USING
WS-PORTION OF WS-GETASSIGNTEXT-PARAM
WS-TEXT OF WS-GETASSIGNTEXT-PARAM
WS-MESSAGE-NUMBER OF WS-GETASSIGNTEXT-PARAM
GIVING
WS-RESULT OF WS-GETASSIGNTEXT-PARAM
IF WS-RESULT OF WS-GETASSIGNTEXT-PARAM NOT = -1
ADD 1 TO WS-NUMBER-OF-ENTRIES OF WS-ASSIGN-TABLE
MOVE WS-TEXT OF WS-GETASSIGNTEXT-PARAM TO
WS-ASSIGN-NAME OF WS-ASSIGN-TABLE
(WS-NUMBER-OF-ENTRIES OF WS-ASSIGN-TABLE)
END-IF

MOVE "TANDEMNAME" TO WS-PORTION OF WS-GETASSIGNTEXT-PARAM

ENTER "GETASSIGNTEXT"
USING
WS-PORTION OF WS-GETASSIGNTEXT-PARAM
WS-TEXT OF WS-GETASSIGNTEXT-PARAM
WS-MESSAGE-NUMBER OF WS-GETASSIGNTEXT-PARAM
GIVING
WS-RESULT OF WS-GETASSIGNTEXT-PARAM
IF WS-RESULT OF WS-GETASSIGNTEXT-PARAM NOT = -1
MOVE WS-TEXT OF WS-GETASSIGNTEXT-PARAM TO
WS-SYSTEM-AND-PROCESS OF WS-ASSIGN-TABLE
(WS-NUMBER-OF-ENTRIES OF WS-ASSIGN-TABLE)
END-IF
END-PERFORM.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -62
Examples Nested Server Example

0410-GET-MY-SYSTEM-PROCESS.

* THIS PARAGRAPH GETS THE PROCESS-ID FOR THE CURRENT SERVER PROCESS.
* THIS PROCESS-ID IS RETURNED TO THE REQUESTER IN PATHSRV-REPLY.

ENTER TAL "MYPID" GIVING WS-CPU-PIN OF WS-MY-PROCESS.

ENTER TAL "GETCRTPID" USING WS-CPU-PIN OF WS-MY-PROCESS
WS-PROCESS-ID OF WS-MY-PROCESS.
ENTER TAL "MYSYSTEMNUMBER" GIVING WS-SYSTEM-NUMBER OF WS-MY-PROCESS.
ENTER TAL "GETSYSTEMNAME" USING WS-SYSTEM-NUMBER OF WS-MY-PROCESS
WS-SYSTEM-NAME OF WS-MY-PROCESS.
0420-GET-MOM-SYSTEM-PROCESS.

* THIS PARAGRAPH GETS INFORMATION ABOUT THE MOM OF THE CURRENT SERVER
* PROCESS. THIS INFORMATION IS USED TO DETERMINE IF THE MOM IS A
* PATHMON. THE MOM PROCESS IS USED IN PATHSEND SENDS IF THE REQUEST
* (IN PATHSRV-REQUEST) DOES NOT SPECIFY A PATHMON (EITHER EXPLICITLY
* OR WITH AN ASSIGN).

MOVE WS-PROCESS-NAME OF WS-MY-PROCESS

TO WS-PROCESS-NAME OF WS-LOOKUPPROCESSNAME-PARAM.
ENTER TAL "LOOKUPPROCESSNAME" USING WS-LOOKUPPROCESSNAME-PARAM.
MOVE WS-ANCESTOR-PROCESS-ID OF WS-LOOKUPPROCESSNAME-PARAM
TO WS-PROCESS-ID-GENERIC OF WS-MOM-PROCESS.
IF NOT WS-LOCAL OF WS-MOM-PROCESS
MOVE ZERO TO WS-SYSTEM-NUMBER OF WS-MOM-PROCESS
MOVE WS-SYSTEM-NUMBER-1-BYTE OF WS-MOM-PROCESS TO
WS-BYTE-2 OF WS-SYSTEM-NUMBER-2-BYTES OF WS-MOM-PROCESS
ELSE
ENTER TAL "MYSYSTEMNUMBER" GIVING WS-SYSTEM-NUMBER OF WS-MOM-PROCESS
END-IF.
ENTER TAL "PROCESSINFO" USING WS-CPU-PIN OF WS-PROCESS-ID-GENERIC
OF WS-MOM-PROCESS
WS-PROCESS-ID-GENERIC OF WS-MOM-PROCESS
OMITTED
OMITTED
OMITTED
WS-PROGRAM-FILENAME OF WS-MOM-PROCESS
OMITTED
WS-SYSTEM-NUMBER OF WS-MOM-PROCESS
OMITTED
OMITTED
OMITTED
OMITTED
OMITTED
OMITTED
OMITTED
GIVING WS-ERROR OF WS-PROCESSINFO-PARAM.
IF WS-ERROR OF WS-PROCESSINFO-PARAM NOT = ZERO
MOVE "Y" TO WS-SERVER-IS-ASSOCIATIVE
ELSE
IF WS-FILE OF WS-PROGRAM-FILENAME OF WS-MOM-PROCESS = "PATHMON"
MOVE "N" TO WS-SERVER-IS-ASSOCIATIVE
ELSE
MOVE "Y" TO WS-SERVER-IS-ASSOCIATIVE
END-IF
END-IF.

IF WS-SERVER-IS-ASSOCIATIVE = "N"
IF WS-LOCAL OF WS-MOM-PROCESS
MOVE WS-PROCESS-ID OF WS-PROCESS-ID-LOCAL OF WS-MOM-PROCESS
TO WS-PROCESS-NAME OF WS-MOM-PROCESS
ELSE

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -63
Examples Nested Server Example

STRING "$"
WS-PROCESS-ID OF WS-PROCESS-ID-NETWORK OF WS-MOM-PROCESS
DELIMITED BY " "
INTO WS-PROCESS-NAME OF WS-MOM-PROCESS
END-IF
ENTER TAL "GETSYSTEMNAME" USING WS-SYSTEM-NUMBER OF WS-MOM-PROCESS
WS-SYSTEM-NAME OF WS-MOM-PROCESS
STRING WS-SYSTEM-NAME OF WS-MOM-PROCESS DELIMITED BY " "
"." DELIMITED BY SIZE
WS-PROCESS-NAME OF WS-MOM-PROCESS DELIMITED BY " "
INTO WS-SYSTEM-AND-PROCESS OF WS-MOM-PROCESS
END-IF.

0440-VALIDATE-PATHSRV-REQUEST.
IF PATHMON-ASSIGN-NAME OF PATHSRV-REQUEST NOT = SPACES AND
PATHMON-SYSTEM-AND-PROCESS OF PATHSRV-REQUEST NOT = SPACES
MOVE "BOTH ASSIGN NAME AND PROCESS NAME SHOULD NOT BE NON-BLANK"
TO NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
MOVE "N" TO WS-VALID-PATHSRV-REQUEST
END-IF.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -64
Examples Nested Server Example

0450-LOOKUP-PATHMON-ASSIGN.

* THIS PARAGRAPH LOOKS UP AN ASSIGN FROM THE REQUEST MESSAGE TO SEE

* WHETHER A CORRESPONDING PATHMON SYSTEM AND PROCESS NAME CAN BE
* FOUND IN THE TABLE OF ASSIGNS BUILT WHEN THE SERVER PROCESS
* STARTED. THE RESULTING PATHMON SYSTEM AND PROCESS NAME IS PUT INTO
* THE PARAMETER VARIABLE THAT IS USED IN THE PATHSEND SEND.

PERFORM VARYING WS-INDEX OF WS-ASSIGN-TABLE

FROM 1 BY 1 UNTIL WS-INDEX OF WS-ASSIGN-TABLE >
WS-NUMBER-OF-ENTRIES OF WS-ASSIGN-TABLE
IF WS-ASSIGN-NAME OF WS-ASSIGN-TABLE (WS-INDEX OF WS-ASSIGN-TABLE) =
PATHMON-ASSIGN-NAME OF PATHSRV-REQUEST
MOVE WS-SYSTEM-AND-PROCESS OF WS-ASSIGN-TABLE
(WS-INDEX OF WS-ASSIGN-TABLE)
TO WS-PATHMON-PROCESS-NAME OF WS-SERVERCLASS-SEND-PARAM
MOVE 99 TO WS-INDEX OF WS-ASSIGN-TABLE
END-IF
END-PERFORM
IF WS-INDEX OF WS-ASSIGN-TABLE LESS 99
MOVE "N" TO WS-VALID-PATHSRV-REQUEST
STRING "ASSIGN NAME MISSING : "
PATHMON-ASSIGN-NAME OF PATHSRV-REQUEST
DELIMITED BY SIZE
INTO NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
END-IF.

0460-SEND-TO-SUBSIDIARY-SERVER.

* THIS PARAGRAPH PERFORMS THE PATHSEND SEND TO THE SUBSIDIARY SERVER

MOVE "N" TO WS-SERVERCLASS-SEND-OKAY.

MOVE SERVER-CLASS OF PATHSRV-REQUEST
TO WS-SERVER-CLASS-NAME OF WS-SERVERCLASS-SEND-PARAM.
MOVE SUBSIDIARY-REQUEST TO WS-MESSAGE-BUFFER
OF WS-SERVERCLASS-SEND-PARAM.

ENTER TAL "SERVERCLASS_SEND_"

USING
WS-PATHMON-PROCESS-NAME OF WS-SERVERCLASS-SEND-PARAM
WS-PATHMON-PROCESS-NAME-LEN OF WS-SERVERCLASS-SEND-PARAM
WS-SERVER-CLASS-NAME OF WS-SERVERCLASS-SEND-PARAM
WS-SERVER-CLASS-NAME-LEN OF WS-SERVERCLASS-SEND-PARAM
WS-MESSAGE-BUFFER OF WS-SERVERCLASS-SEND-PARAM
WS-REQUEST-LEN OF WS-SERVERCLASS-SEND-PARAM
WS-MAXIMUM-REPLY-LEN OF WS-SERVERCLASS-SEND-PARAM
WS-ACTUAL-REPLY-LEN OF WS-SERVERCLASS-SEND-PARAM
WS-TIMEOUT OF WS-SERVERCLASS-SEND-PARAM
GIVING
WS-ERROR OF WS-SERVERCLASS-SEND-PARAM.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -65
Examples Nested Server Example

* THE SERVER CLASS OPERATION NUMBER <SCSEND-OP-NUM> PARAMETER IS

* OMITTED BECAUSE THE SEND IS BEING DONE IN WAIT MODE. IN NOWAIT
* MODE THE <SCSEND-OP-NUM> IS USED IN CALLS TO AWAITIOX.
*
* THE <TAG> AND <FLAGS> PARAMETER ARE OMITTED BECAUSE THE SEND IS
* BEING DONE IN WAIT MODE. THEY ARE ONLY NEEDED IN NOWAIT MODE.

IF WS-ERROR OF WS-SERVERCLASS-SEND-PARAM = 0
MOVE "SUCCESSFUL" TO PATHSEND-ERROR OF PATHSRV-REPLY
MOVE WS-MESSAGE-BUFFER OF WS-SERVERCLASS-SEND-PARAM
TO SUBSIDIARY-REPLY
MOVE "Y" TO WS-SERVERCLASS-SEND-OKAY
ELSE
IF WS-ERROR OF WS-SERVERCLASS-SEND-PARAM = 233

* A "233" (SERVER CLASS ERROR) ERROR MUST BE FURTHER ANALYZED WITH

* SERVERCLASS_SEND_INFO_ TO GET THE PATHSEND ERROR NUMBER AND FILE
* SYSTEM ERROR NUMBER.

PERFORM 0500-ANALYZE-SEND-ERROR-233
ELSE
MOVE "UNEXPECTED ERROR FOUND AFTER SERVERCLASS_SEND_ CALL"
TO PATHSEND-ERROR OF PATHSRV-REPLY
END-IF
END-IF.

0500-ANALYZE-SEND-ERROR-233.

* THIS PARAGRAPH CALLS SERVERCLASS_SEND_INFO_ TO GET THE PATHSEND

* ERROR NUMBER AND TO GET THE FILE SYSTEM ERROR NUMBER.

ENTER TAL "SERVERCLASS_SEND_INFO_"

USING
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM
WS-FILE-SYSTEM-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM
GIVING
WS-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM.

IF WS-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM NOT = ZERO

PERFORM 0700-SERVERCLASS-SEND-INFO-ERR
ELSE
IF WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 907 OR
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 908 OR
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 909 OR
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 910 OR
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 911 OR
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 912 OR
WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 919

* ERROR 907 : INVALID (EXTENDED) SEGMENT ID

* ERROR 908 : NO (EXTENDED) SEGMENT IN USE
* ERROR 909 : INVALID VALUE FOR FLAGS PARAMETER
* ERROR 910 : REQUIRED PARAMETER NOT SUPPLIED
* ERROR 911 : ONE OF THE BUFFER LENGTH PARAMETERS IS INVALID
* ERROR 912 : A REFERENCE PARAMETER IS OUT OF BOUNDS
* ERROR 919 : INVALID VALUE FOR TIMEOUT PARAMETER

MOVE WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM

TO WS-NUMERIC-DISPLAY
STRING "BAD PARAMETER PASSED TO SERVERCLASS_SEND_ (ERROR = "
WS-NUMERIC-DISPLAY
")"
DELIMITED BY SIZE

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -66
Examples Nested Server Example

INTO PATHSEND-ERROR OF PATHSRV-REPLY

ELSE
MOVE WS-PATHSEND-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM
TO WS-NUMERIC-DISPLAY
STRING "SERVERCLASS_SEND_ PATHSEND ERROR NUMBER : "
WS-NUMERIC-DISPLAY
DELIMITED BY SIZE
INTO PATHSEND-ERROR OF PATHSRV-REPLY
END-IF

IF WS-FILE-SYSTEM-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 48
MOVE "SERVERCLASS_SEND_ SECURITY VIOLATION (ERROR 48)"
TO FILE-SYSTEM-ERROR OF PATHSRV-REPLY
ELSE
IF WS-FILE-SYSTEM-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM = 40
MOVE "SERVERCLASS_SEND_ TIMED OUT (ERROR 40)"
TO FILE-SYSTEM-ERROR OF PATHSRV-REPLY
ELSE
MOVE WS-FILE-SYSTEM-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM
TO WS-NUMERIC-DISPLAY
STRING "SERVERCLASS_SEND_ FILE SYSTEM ERROR NUMBER : "
WS-NUMERIC-DISPLAY
DELIMITED BY SIZE
INTO FILE-SYSTEM-ERROR OF PATHSRV-REPLY
END-IF
END-IF
END-IF.

0700-SERVERCLASS-SEND-INFO-ERR.

* THIS PARAGRAPH IS CALLED WHEN AN ERROR IS ENCOUNTERED CALLING

* SERVERCLASS_SEND_INFO_.

EVALUATE WS-ERROR OF WS-SERVERCLASS-SEND-INFO-PARAM

WHEN 2
MOVE "SERVERCLASS_SEND_INFO_ EXTENDED SEGMENT USAGE ERROR"
TO NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
WHEN 22
MOVE "SERVERCLASS_SEND_INFO_ PARAMETER OUT OF BOUNDS"
TO NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
WHEN 29
MOVE "SERVERCLASS_SEND_INFO_ MISSING REQUIRED PARAMETER"
TO NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
WHEN OTHER
MOVE "UNEXPECTED ERROR FOUND IN CALL TO SERVERCLASS_SEND_INFO_"
TO NON-SEND-ERROR-MESSAGE OF PATHSRV-REPLY
END-EVALUATE.

9000-WRITE-ERROR-LOG-REC.
WRITE ERROR-LOG-REC.

9900-STOP-RUN.
STOP RUN.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -67
Examples Nested Server Example

NonStop TS/MP Pathsend and Server Programming Manual– 132500

B -68
 Glossary
Note. This glossary does not include terms for elements of the SCREEN COBOL language
that are also found in standard COBOL. For definitions of such terms, refer to standard COBOL
texts or to the text of the Pathway/TS SCREEN COBOL Reference Manual.

absolute pathname. An OSS pathname that begins with a slash (/) character and is resolved
beginning with the root directory. See also OSS pathname, relative pathname, and root
directory.

accept operation. An operation in which a screen program waits for a response from the
terminal and allows data to be input into the program data area from the terminal.

advisory message. A message displayed in the terminal advisory field to inform the terminal
operator of errors detected during input checking.

API. See application program interface (API).

application. A complete set of programs or routines that perform a function. See also
Pathway application and NonStop TUXEDO application.

application program interface (API). A set of services (such as programming language

functions or procedures) that are called by an application program to communicate with
other software components. For example, an API might consist of a set of procedure
calls that provide a workstation application with a standard interface for communicating
with a Tandem system. Other examples of APIs are the ATMI in BEA TUXEDO
systems and NonStop TUXEDO systems and the Pathsend procedures.

application terminal. A terminal on which a Pathway application runs. See also command
terminal.

Application-Transaction Monitor Interface (ATMI). The application programming

interface to the System/T transaction monitor in a NonStop TUXEDO system. This
interface includes transaction routines, message handling routines, service interface
routines, and buffer-management routines.

assignment. The use of an ASSIGN command to make logical file assignments for programs
in the Guardian environment. A logical assignment equates a Tandem file name with a
logical file of a program and, optionally, attributes characteristics to that file.

associative server. A process within a server class that can be started outside the Pathway
environment by a process other than the PATHMON process that controls the server
class.

ATMI. See Application-Transaction Monitor Interface (ATMI).

attributes. Those characteristics of an object that influence the operation of that object and
establish its capabilities.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -1
Glossary audited file

audited file. A database file that is flagged for auditing by the TMF subsystem; auditing is
the monitoring of transactions in preparation for recovery efforts.

audit trail. A record of database changes that can be used by the TMF subsystem to rebuild a
database in the event of a hardware or software failure. An audit trail is also known in
the industry as a transaction log.

availability. The amount of time an application running on a Tandem system can be used
effectively by a user of that application.

backup process. The member of a process pair that takes over the application work when the
primary process fails. See also primary process, process pair, and checkpoint message.

base screen. In SCREEN COBOL, a screen that occupies the entire physical display area of a
terminal and can be displayed independently of other screens. This type of screen can
contain areas on which overlay screens are displayed. See also screen and overlay
screen.

batch processing. A method of transaction processing in which transactions are first grouped
together and then processed at regular intervals. See also online transaction processing
(OLTP).

block mode. A terminal operating mode in which data is read from the terminal and
displayed on the terminal one screen at a time. See also conversational mode.

cache. A temporary storage buffer.

cascading server. A term formerly used for a nested server. See nested server.

checkpoint message. In the Guardian environment, a message sent by a primary process to

its backup process that keeps the backup process up to date on the state of the
application. A checkpoint message provides a snapshot of process activity that can be
used in the event of a takeover by a backup process to allow the backup process to
maintain fault-tolerant operation.

CISC. See complex instruction-set computing (CISC).

client. An application program that requests services to be performed. In discussions of the

Pathway environment, this term is used to refer to the part of an application that runs on
some other vendor’s hardware, such as a personal computer, Macintosh computer, UNIX
workstation, or mainframe computer system, and makes requests of a server process.
See also requester, server, and client/server model.

client/server model. A model for distributing applications. In general, but not always, in this
model the client process resides on a workstation and the server process resides on a
second workstation, minicomputer, or mainframe system. Communication takes the
form of request and reply pairs, which are initiated by the client and serviced by the
server. (A server can make requests of another server, thus acting as a client.)
Client/server computing is often used to connect different types of workstations or

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -2
Glossary client/transaction server model

personal computers to a host computer system by means of supported communications

protocols. See also requester/server model.

client/transaction server model. A model for client/server applications. The

client/transaction server model is the model of choice for high-volume OLTP
applications in which transaction volume is great and the processing requirements
change infrequently.
In the Tandem environment, an application following this model divides processing
between a client running on a workstation and servers running on a Tandem system. The
client handles the user interface and business logic and processing. The servers store
information for use by the client and handle database input and output functions.
Interprocess communication (IPC) messages transfer data between client and server.

COBOL85. The Tandem compiler and run-time support for the American National Standards
Institute (ANSI) programming language COBOL, X.3.23-1985. Pathway server
processes are often written in this language.

cold start. The operation that starts a PATHMON environment for the first time. This
operation either creates a new PATHMON configuration file (PATHCTL file) that
defines the PATHMON environment and its objects or overwrites an existing
PATHMON configuration file (which effectively creates a new PATHMON
environment). See also cool start.

command file. A file that serves as a source for command input. For example, users can
prepare a command file containing PATHCOM or SCREEN COBOL Utility Program
(SCUP) commands. They can then cause the commands in the file to be executed by
issuing the PATHCOM or SCUP OBEY command and specifying the name of the file.
Alternatively, they can specify this file as the input file when they execute PATHCOM or
SCUP.

command interpreter. An interactive program used to run programs, check system status,
create and delete disk files, and alter hardware states.

command terminal. A terminal at which a system manager or operator enters commands for
configuration and management, such as the PATHCOM commands that configure and
manage a PATHMON environment. See also application terminal.

complex instruction-set computing (CISC). A processor architecture based on a large

instruction set, characterized by numerous addressing modes, multicycle machine
instructions, and many special-purpose instructions. See also reduced instruction-set
computing (RISC).

configuration. The definition or alteration of characteristics of an object. See also object.

configured TERM object. A TERM object that is explicitly configured with an ADD TERM
command. Such a TERM object exists until it is explicitly deleted. Names of configured
TERM objects begin with a letter. See also temporary TERM object and TERM object.

consistency. See database consistency.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -3
Glossary context

context. Information required by a server to process the current request in an exchange of

multiple request and reply messages: for example, identification of the last item
processed. See also context-free server and terminal context.

context-free server. A server that does not retain any information about the processing of
previous requests. A context-free server accepts a single message from a requester,
performs the requested tasks, and issues a single reply to respond to the requester. After
the reply message is issued, the server retains no information, or context, that can be
used in subsequent requests. In general, context-free servers are relatively simple to
program and can be restarted quickly, but they require the requester to pass context
information to the server on each request. Servers handling requests from Pathsend
requesters can be either context-free or context-sensitive, but servers servicing requests
from SCREEN COBOL requesters must be context-free. A context-free server is
analogous to a NonStop TUXEDO request/response server. Tandem subsystems are
context-free servers; therefore, management applications using the Subsystem
Programmatic Interface (SPI) to communicate with Tandem subsystems must pass back
context information in continuation requests. See also context, context-sensitive server,
and Subsystem Programmatic Interface (SPI).

context-sensitive server. A server that retains information about the processing of previous
requests. A context-sensitive Pathway server can engage in a multiple-message
communication, or dialog, with a requester. Because context-sensitive servers must
maintain message context for the dialog, they are more complex to program than
context-free servers. They typically have longer restart times because they must recover
the requester context. See also context and context-free server.

context sensitivity. The ability of a requester to exchange a series of multiple request and
reply messages (that is, a dialog) with a particular server process. See also context-
sensitive server and dialog.

conversation. See dialog.

conversational mode. (1) A terminal operating mode in which data is read from the terminal
and displayed on the terminal screen one line at a time. See also block mode and
intelligent mode. (2) The mode of communication that enables an ongoing dialog
between a client (or requester) and a server. Data is sent and received in an iterating
fashion without return to the transaction monitor until the application dialog is
completed. Multiple messages can be exchanged between the client and server
participating in the communication. See also conversational server.

conversational model. A model for requester-server communication that enables an ongoing

dialog between a client (or requester) and a server. Multiple messages can be exchanged
between the client and server process before control is returned to the transaction
monitor. See also request/response model and conversational server.

conversational server. A server that offers conversational services and can participate in a
conversation, or dialog, with a client; that is, a context-sensitive server. See also
conversational mode (definition 2), request/response server, and context-sensitive server.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -4
Glossary cool start

cool start. The operation that restarts a PATHMON environment, using the information in an
existing PATHMON configuration file (PATHCTL file). The PATHMON environment
must have been previously started with a cold start operation. See also cold start.

Crossref cross-reference generator. A Tandem software tool that produces a cross-

referenced listing of selected identifiers—such as data variables, statement labels, or
subprograms—in an application program.

current working directory. The OSS directory from which relative pathnames are resolved.
See also OSS pathname and relative pathname.

database consistency. The state of a database in which items satisfy established criteria. For
example, an account balance must equal credits to the balance minus debits to the
balance. When the database satisfies these criteria, the database is considered to be
consistent. In general, a database is consistent when it is accurate and all changes
generated by transactions are complete. Database consistency is defined by the
application, which establishes the values and relationships of database fields and
records.

database management system (DBMS). A product, such as NonStop SQL/MP or Enscribe,

that serves as the interface between a user or program (for example, a Pathway server)
and the database. Among its many functions, the DBMS controls access to and
organization of data within the database.

Data Definition Language (DDL). (1) The set of data definition statements within the
Structured Query Language (SQL). (2) A Tandem product for defining data objects in
Enscribe files and translating object definitions into source code.

data integrity. The condition of a database when its data values are accurate, valid, and
consistent according to rules established for changing the database. See also database
consistency.

DBCS. See double-byte character set (DBCS).

DDL. See Data Definition Language (DDL).

deadlock. (1) A situation in which two processes cannot proceed because each is waiting for
a reply from the other. (2) A situation in which two transactions cannot proceed because
each is waiting for the other to release a lock.

dedicated device. A term formerly used for a terminal or other input/output device controlled
by a configured TERM object, so that a Pathway application always ran on that device
without having to be started from PATHCOM with a RUN PROGRAM command. (No
new term replaces this term; instead, the manual text now refers to such devices as those
associated with configured TERM objects.) See also nondedicated device and
configured TERM object.

default value. The value that the system uses for a particular attribute or parameter when a
value has not been supplied by the user.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -5
Glossary DEFINE

DEFINE. A named set of attributes and associated values. In a DEFINE (as with an ASSIGN
command), users can specify information to be communicated to processes they start.

definition files. A set of files containing data declarations for items related to SPI messages
and their processing. The core definitions required to use SPI are provided in a DDL file
and in several language-specific definition files, one for each programming language
that supports SPI. The Tandem DDL compiler generates the language-specific files from
the DDL file. Subsystems that support SPI provide additional definition files containing
subsystem-specific definitions.

delimiters. Characters that make it possible for a SCREEN COBOL requester and an external
device or front-end process to exchange compact variable-length messages efficiently;
delimiters can be message delimiters or field delimiters.

descriptor. For each elementary data item, the SCREEN COBOL compiler builds a data
structure that describes the size, type, usage, and dependencies of the item. All of the
information that pertains to a given item makes up the descriptor for that item. For
example, the PICTURE specification is included in the descriptor. The descriptors are
passed to the TCP in the pseudocode and provide a dictionary of information for
interpreting and handling incoming data. When the MAP or SMAP compiler option is
used, the descriptors appear in the compiler map at the end of the listing.

diagnostic screen. A screen of information that is displayed to inform the terminal operator
of error conditions and termination status.

dialog. A multiple-message communication between a requester and a context-sensitive

server. A dialog is also called a conversation. See also context sensitivity and context-
sensitive server.

disk process. In the Tandem NonStop Kernel operating environment, the portion of the
operating-system software that performs read, write, and lock operations on disk
volumes and creates TMF audit records. See also file system.

display attribute. A terminal display feature that is given a screen data name. The screen
data name can be associated with a predefined system name in the SPECIAL-NAMES
paragraph and thus be manipulated by a SCREEN COBOL program.

distributed data. Information (for example, customer names and addresses, inventory items,
and personnel records) that resides on more than one node in a network and can be
accessed by authorized users from any node in that network.

distributed processing. A type of processing environment in which resources are distributed

among CPUs within a single system or spread across a network of systems. A user on
any network node can, if properly authorized, access resources and database files
anywhere within the network.

Distributed Systems Management (DSM). A group of tools for managing a variety of

subsystems in a distributed processing environment.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -6
Glossary distributed transaction processing (DTP)

distributed transaction processing (DTP). The coordination of transactions among

application servers residing within an Expand network and possibly accessing different
database management systems (NonStop SQL/MP and Enscribe). DTP allows the
coordination of multiple, autonomous actions as a single logical unit of work.

double-byte character. A character represented in two bytes. See also double-byte character
set.

double-byte character set (DBCS). A character set, such as Tandem Kanji, that uses two
bytes of data to represent a single character.

DSM. See Distributed Systems Management (DSM).

dumb terminal. See fixed-function terminal.

dynamic server. A server process that the PATHMON process creates after a TCP or
LINKMON process has waited for a specified time period for a static server to become
available. A dynamic server process exists only as long as it is needed. See also static
server.

EDIT file. A source text file that can be augmented and modified by the user through a
Tandem text editor program such as TEDIT (PS Text Edit).

EMS. See Event Management Service (EMS).

EMS log file. See event log file.

Enable product. A tool provided by Tandem that allows users to develop simple data
management applications without using a conventional programming language. The
Enable product can generate SCREEN COBOL programs for use with Enscribe
databases.

Enscribe database record manager. Tandem database management software that provides a
record-at-a-time interface between servers and a distributed database. See also NonStop
SQL/MP.

event log file. A file maintained by the Event Management Service (EMS) for logging of
event messages.

Event Management Service (EMS). A part of DSM used to provide event collection, event
logging, and event distribution facilities. It provides for different descriptions of events
for people and for programs, lets an operator or application select specific event-
message data, and allows for flexible distribution of event messages within a system or
network. EMS has an SPI-based programmatic interface for reporting and retrieving
events. See also event message.

event message. A special kind of SPI message that describes an event occurring in the system
or network. Event messages are collected, logged, and distributed by EMS. See also
Event Management Service (EMS).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -7
Glossary expandability

expandability. See scalability.

Expand networking software. Tandem software that can connect up to 255 Tandem
NonStop systems into a single network.

Extended General Device Support (GDSX). A Tandem product that facilitates

communication between general I/O devices and a PATHMON environment by acting as
a front-end or a back-end process.

extensible structured token. In the Subsystem Programmatic Interface (SPI), a token with a
value that can be extended by appending new fields in later releases. The token is
accessed through reference to a token map containing field-version and null-value
information, allowing SPI to provide compatibility between different versions of the
structure. See also simple token and token (definition 2).

external PATHMON process. See external process.

external process. A process in a different PATHMON environment from the process with
which it is communicating. For example, suppose a TCP managed by PATHMON
process $PMB requests a link to a server process in a server class that is managed by
PATHMON process $PMA. Both the TCP and PATHMON process $PMB are external
processes with respect to PATHMON process $PMA and the server class managed by
$PMA.

external server. See external process.

external TCP. See external process.

fault tolerance. The ability of a Tandem NonStop system to continue processing despite the
failure of any single software or hardware component within the system.

field-characteristic clause. In SCREEN COBOL, an ordered set of characters that specify

the characteristics of a screen field.

file identifier (file ID). In the Guardian environment, the portion of a file name following the
subvolume name. In the OSS environment, a portion of the internal information used to
identify a file in the OSS file system. The two identifiers are not comparable.

file name. In the Guardian environment, the set of node name, volume name, subvolume
name, and and file identifier characters that uniquely identifies a file. This name is used
to open a file and thereby provide a connection between the opening process and the file.
See also fully qualified file name, partially qualified file name, OSS filename, and OSS
pathname.

file-name expansion. The expansion of a partially qualified Guardian file name for a disk file
to include the associated node, volume, and subvolume names.

file system. (1) In the Guardian environment, the application program interface for
communication between a process and a file. A file can be a disk file, a device other than
a disk, or another process. (2) In the OSS environment, a collection of files and file

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -8
Glossary File Utility Program (FUP)

attributes. A file system provides the namespace for the file serial numbers that uniquely
identify its files.

File Utility Program (FUP). A Tandem product that allows users to create, copy, purge, and
otherwise manipulate disk files interactively.

fixed-function terminal. A nonintelligent device (that is, a device without processing ability)
capable of sending and receiving information over communications lines. Fixed-function
terminals are often referred to as dumb terminals.

freeze condition. A condition in which communication between a terminal and a server class
is prohibited. See also thaw condition.

front-end process. A process, such as the process that accepts the TCP terminal-data stream,
that serves as the intermediary between one system, process, or device and another.

fully qualified file name. The complete name of a file in the Guardian environment. For a
permanent disk file, this consists of a node name (system name), volume name,
subvolume name, and file identifier (file ID). In interactive interfaces such as
PATHCOM and TACL, the parts of a file name are separated by periods. See also
partially qualified file name.

gateway process. A process, such as the Transaction Delivery Process (TDP) that is part of
RSC, that manages communications between dissimilar environments (for example, a
workstation and a Tandem system). A gateway process both transfers information and
converts it to a form compatible with the protocols used by the destination environment.

GDSX. See Extended General Device Support (GDSX).

graphical user interface (GUI). A type of screen interface that typically includes pull-down
menus, icons, dialog boxes, and online help.

Guardian. An environment available for interactive or programmatic use with the Tandem
NonStop Kernel. Processes that run in the Guardian environment use the Guardian
system procedure calls as their application program interface, and might also use related
APIs such as the Pathsend and TMF procedure calls. See also Open System Services
(OSS).

Guardian environment. The Guardian API, tools, and utilities. See also Guardian.

Guardian operating environment. See Guardian environment.

high PIN. A process identification number (PIN) in the range 256 through 65535. See also
process identification number (PIN) and low PIN.

IDS. See intelligent device support (IDS) facility.

Inspect. The Tandem debugging tool that can be used interactively to examine and modify
the execution of Guardian processes and SCREEN COBOL requester programs.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -9
Glossary Inspect command terminal

Inspect command terminal. The terminal on which programmers enter commands to

Inspect when debugging a SCREEN COBOL program or a Pathway server.

intelligent device. A device such as an automatic teller machine, a point-of-sale device, or a

communications line, or a process such as a Guardian process, that can communicate
with the Pathway environment through the intelligent device support (IDS) facility, the
Remote Server Call (RSC) product, or the Pathsend procedure calls.

intelligent device support (IDS) facility. A feature of the TCP that supports access to
Pathway server classes by intelligent devices. This facility allows SCREEN COBOL
requester programs to interact with external processes that, in turn, control devices such
as personal computers, automated teller machines, and point-of-sale devices.

intelligent mode. An operating mode in which data and messages are sent between an
intelligent device and the Pathway environment. See also conversational mode
(definition 1), intelligent device, and message-oriented requester.

interactive mode. An operating mode in which commands are entered from a terminal
keyboard.

interoperability. The ability to communicate, execute programs, or transfer data between

dissimilar environments.

interoperate. To communicate, execute programs, or transfer data between dissimilar

environments.

interprocess communication (IPC) message. The unit of communication between

requesters and servers. An IPC message consists of a request message and a reply
message.

I/O process. In the Guardian environment, a system process to manage input/output devices.
Applications use the Guardian file system to send requests to I/O processes. See also file
system.

IPC message. See interprocess communication (IPC) message.

keyword. A word in a command string or programming language that must be spelled and
positioned in a prescribed way, usually to indicate the meaning of an adjacent parameter.

library. A set of related files or common files that can be accessed by multiple programs or
processes.

linear expandability. See scalability.

link. (1) An open of a server process within a server class. When a link manager—that is, a
TCP or a LINKMON process—sends a request to a PATHMON process for a link to a
server in a specified server class, the PATHMON process selects a server process in that
server class (possibly starting a new server process if necessary) and then returns the
name of the server process to the requesting link manager. See also link granting and
link manager. (2) To examine, collect, associate together, and modify code and data

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -10
Glossary link access

blocks from one or more object files to produce a target object file. On Tandem NonStop
systems, linking for TNS/R native object files is performed by the nld utility.

link access. The actual transfer of data from a requester to a server process. In the Pathway
environment, link access is provided by TCPs and LINKMON processes. See also link
granting.

link granting. The process of selecting a particular server process in a server class to handle
a request from a link manager (TCP or LINKMON process) on behalf of a requester.
In the Pathway environment, link granting is done by the PATHMON process. See also
link and link access.

link management. The act of coordinating the sharing of links between requester or client
processes and server processes.

link manager. A process that requests links to server processes and provides link access after
the link is granted. TCPs and LINKMON processes are the link managers in the
Pathway environment. See also link access and link granting.

LINKMON process. A Guardian process that supports access to servers in the Pathway and
NonStop TUXEDO environments. LINKMON processes act as link managers for
requesters and clients that use the Pathsend procedure calls and other interfaces such as
RSC, POET, and the NonStop TUXEDO ATMI. See also link manager.

lock. A mechanism that coordinates access to the same data; locks are either shared or
exclusive.

log file. See PATHMON log file and event log file.

low PIN. A process identification number (PIN) in the range 0 through 255. (PIN 255 is
reserved by the system; it is never assigned to a running process, but is used by high-PIN
processes to communicate with low-PIN processes.) See also process identification
number (PIN) and high PIN.

manageability. The ability to easily and comprehensively manage a subsystem, system, or

network.

MAKEUL. A TACL macro used to perform pTAL compilation of user-written conversion

routines for use with the Pathway/TS TCP and SCREEN COBOL requesters and to
create the TNS/R native TCP user library containing these routines.

management application. An application program that automates configuration and

management tasks. Such a program can request from the PATHMON process the same
kinds of services that system managers can request through the PATHCOM interface. A
management application can also interact with subsystems other than the Pathway
subsystem. Management applications use the Subsystem Programmatic Interface (SPI)
to send commands to subsystems and the Event Management Service (EMS) to receive
notification of significant events.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -11
Glossary message-oriented requester

message-oriented requester. A SCREEN COBOL requester that sends data from working
storage to a device (or to a front-end process that controls a device) and receives data
from the device or process into working storage by way of Message Section templates.
SCREEN COBOL requesters that use the intelligent device support (IDS) facility are
message-oriented. These requesters use SEND MESSAGE statements and their REPLY
clauses in the Procedure Division to interact with the intelligent devices or front-end
processes. See also screen-oriented requester.

Message Section. A section in the Data Division of a SCREEN COBOL source program that
describes the data type, size, and relative position (sequence) of each field in a message
template. This section also defines the editing and conversion that must be performed on
each field. See also message-oriented requester.

mixed data item. A data item that contains both single-byte and double-byte characters; in a
COBOL or SCREEN COBOL program, these data items are declared as PIC X.

modified data tag (MDT). In SCREEN COBOL, a bit that is set or reset to indicate whether
data in an associated field is to be sent to the computer from the terminal.

multithreaded. A programming model that provides more than one thread of control within a
program. Multithreading allows multiple sequential processing tasks to be executed
concurrently within a process: for example, a terminal control process (TCP). See also
thread and single-threaded.

native System/T client. In the NonStop TUXEDO environment, a client program that
executes in the Open System Services (OSS) environment and communicates directly
with System/T. An example of this type of client is the Pathway translation server for the
NonStop TUXEDO system. Also called a NonStop TUXEDO native client.

nested server. A server that acts as a requester by sending requests to other servers. In the
Pathway environment, such requests are made by calls to Pathsend procedures.

nld utility. A utility that collects, links, and modifies code and data blocks from one or more
object files to produce a target TNS/R native object file.

node. A Tandem NonStop system that is part of an Expand network. The name of the node,
also called the system name, is the first of four parts of a file name in the Guardian
environment. See also Expand networking software.

no-early-reply rule. The rule that states that when a server process reads a request message,
it should completely process the request before it replies to it.

nondedicated device. A term formerly used for a terminal or other input/output device on
which a Pathway application could be started from PATHCOM with a RUN PROGRAM
command. The RUN PROGRAM command results in the creation of a temporary
TERM object to control the terminal. (No new term replaces this term; instead, the
manual text now refers to such devices as those associated with temporary TERM
objects.) See also dedicated device and temporary TERM object.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -12
Glossary noninteractive mode

noninteractive mode. An operating mode in which commands are entered through a

command file.

NonStop Kernel. See Tandem NonStop Kernel.

NonStop processing. On Tandem NonStop systems, processing characterized by continued

operation even when a component fails, when equipment is being repaired or replaced,
or while new processors or peripheral devices are being added to the system. In the
Guardian environment, NonStop processing is provided by means of fault tolerance and
process pairs.

NonStop SQL/MP. The Tandem relational database management system that promotes
efficient online access to large distributed databases. See also Structured Query
Language (SQL) and Enscribe database record manager.

NonStop Transaction Manager/MP (NonStop TM/MP). A Tandem software product that

provides transaction management, transaction protection, and database consistency in
online transaction processing (OLTP) environments. It gives full protection to
transactions that access NonStop SQL/MP and Enscribe databases, as well as recovery
capabilities for transactions, online disk volumes, and entire databases. The component
of NonStop TM/MP that provides these features is the TMF subsystem. See also
Transaction Management Facility (TMF) subsystem.

NonStop Transaction Services/MP (NonStop TS/MP). A Tandem product that provides

process management and link management functions for OLTP applications on Tandem
NonStop systems. NonStop TS/MP consists of the PATHMON process, the LINKMON
process, the PATHCOM process and interface, and the Pathsend procedures. Together
with NonStop Transaction Manager/MP (NonStop TM/MP), NonStop TS/MP forms the
foundation for Tandem’s open transaction processing services, including those provided
by the RSC and POET products and the NonStop TUXEDO system. See also
Pathway/TS and NonStop Transaction Manager/MP (NonStop TM/MP).

NonStop TUXEDO application. The collection of machines, servers and services, and
System/T components defined by a single configuration file in the NonStop TUXEDO
environment. See also NonStop TUXEDO transaction processing environment and
Pathway application.

NonStop TUXEDO environment. See NonStop TUXEDO transaction processing

environment.

NonStop TUXEDO native client. See native System/T client.

NonStop TUXEDO server. A server process or program managed by the NonStop

TUXEDO system administrative facilities. See also Pathway server.

NonStop TUXEDO system. Tandem’s implementation of the BEA Systems, Inc., TUXEDO
enterprise transaction processing system. Major features of the NonStop TUXEDO
implementation include the use of the Tandem core services (Tandem NonStop Kernel,
NonStop Transaction Services/MP, and NonStop Transaction Manager/MP) to provide
the Tandem fundamentals (scalability, availability, and manageability) for TUXEDO

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -13
Glossary NonStop TUXEDO transaction processing
environment

applications. The NonStop TUXEDO system runs in the OSS operating environment on
Tandem NonStop systems. See also OSS environment.

NonStop TUXEDO transaction processing environment. An environment that provides the

API and transaction monitor functions of the BEA Systems, Inc., TUXEDO transaction
processing system in addition to the benefits provided by the Tandem core services:
Tandem NonStop Kernel, NonStop Transaction Services/MP, and NonStop Transaction
Manager/MP. See also Pathway transaction processing environment.

OBEY command. A command in a Tandem interactive interface, such as PATHCOM or the

SCREEN COBOL Utility Program (SCUP), that allows users to execute the commands
in a command file. See also command file.

object. An entity that is subject to independent reference or control by one or more

subsystems. Examples of objects are devices, communications lines, processes, and
files. In the PATHMON environment, the types of objects referred to or controlled by
PATHCOM are PATHMON, PATHWAY, LINKMON, SERVER, TCP, TERM,
PROGRAM, and TELL. The PATHWAY object, used with the SET PATHWAY and
STATUS PATHWAY commands, refers to an entire PATHMON environment. The
LINKMON object is only referred to, not controlled, by PATHCOM: that is, users can
use PATHCOM to obtain information about LINKMON processes but cannot use it to
make any changes in the configuration or state of those processes. See also PATHMON
environment.

object attributes. See attributes.

object file. A file generated by a compiler, linker, or binder that contains machine instructions
and other information needed to construct the executable code spaces and initial data for
a process. The file can be a complete program that is ready for immediate execution, or
it can be incomplete and require linking with other object files before execution.
Compilers for languages such as COBOL85 produce object code. See also pseudocode
file.

object type. In the Subsystem Programmatic Interface (SPI) or in an interactive interface

such as PATHCOM, a category of objects to which a specific object belongs. Object
types for PATHCOM include PATHMON, PATHWAY, LINKMON, SERVER, TCP,
TERM, PROGRAM, and TELL. The SPI interface to the Pathway subsystem uses
different names for some of these object types and defines additional object types; for
example, PM, LM, PROGTERM, and TCPTERM are object types in the SPI interface.

OLTP. See online transaction processing (OLTP).

OLTP application. See online transaction processing (OLTP) application.

online transaction processing (OLTP). A method of processing transactions in which

entered transactions are immediately applied to the database. The information within the
database is readily available to all users through online screens and printed reports. The
transactions are processed while the requester waits, as opposed to queued or batched
transactions, which are processed at a later time. Online transaction processing can be

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -14
Glossary online transaction processing (OLTP) application

used for many different kinds of business tasks such as order processing, inventory
control, accounting functions, and banking operations. See also batch processing.

online transaction processing (OLTP) application. A set of programs that perform online
transaction processing (OLTP) tasks on behalf of the user. With an OLTP application,
many terminal users can update data simultaneously, recording the changes in the
database as they are entered. OLTP applications generally display, check, and accept
input data; manipulate the input data; and perform some type of data-output activity.

Open System Services (OSS). An open system environment available for interactive or
programmatic use with the Tandem NonStop Kernel. Processes that run in the OSS
environment use the OSS application program interface; interactive users of the OSS
environment use the OSS shell for their command interpreter. See also Guardian and
Guardian environment.

OSS. See Open System Services (OSS).

OSS environment. The NonStop Kernel Open System Services (OSS) API, tools, and
utilities. See also Open System Services (OSS).

OSS filename. A component of an OSS pathname containing any valid characters other than
a slash (/) or a null. See also file name, OSS pathname, and file system (definition 2).

OSS operating environment. See OSS environment.

OSS pathname. The string of characters that uniquely identifies a file within its file system
in the OSS environment. A pathname can be either absolute or relative. See also
absolute pathname, relative pathname, and file system (definition 2).

OSS username. A string that uniquely identifies a user within the user database for a node.

overlay area. In SCREEN COBOL, an area of a base screen within which an overlay screen
can be displayed.

overlay screen. In SCREEN COBOL, a screen that is displayed in an overlay area of a base
screen. A base screen can be used with various overlay screens. See also screen and base
screen.

partially qualified file name. A Guardian file name in which only the right-hand file-name
parts are specified. The remaining parts of the file name assume default values. See also
fully qualified file name.

PATHCOM. (1) The interactive interface to the PATHMON process, through which users
enter commands to configure and manage Pathway applications. (2) The process that
provides this interface.

PATHCOM command file. A file of PATHCOM commands that define and add the
PATHMON-controlled objects required to execute an application. This file can contain
all of the commands needed to start a PATHMON environment.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -15
Glossary PATHCOM command terminal

PATHCOM command terminal. See command terminal.

PATHCTL. See PATHMON configuration file.

Pathmaker product. A menu-driven application generator, provided by Tandem, that

increases the productivity of programmers developing Pathway applications. The
Pathmaker software generates requester programs in SCREEN COBOL and server
programs in C or COBOL85.

PATHMON configuration file. A disk file in which a PATHMON process maintains

configuration information for the objects under its control. The name of this file is
PATHCTL.

PATHMON-controlled object. An object defined and managed by a PATHMON process,

through PATHCOM or the Pathway management programming interface. In the
PATHCOM interface, such an object can be of type PATHWAY, PATHMON, SERVER,
TCP, TERM, PROGRAM, or TELL. See also object and Pathway object.

PATHMON environment. The servers, server classes, TCPs, terminals, SCREEN COBOL
programs, and tell messages that run together under the control of one PATHMON
process.

PATHMON log file. A file used by a PATHMON process for reporting errors and changes in
status.

PATHMON object. An object of type PATHMON; that is, a PATHMON process. See also
PATHMON process and PATHMON-controlled object.

PATHMON process. The central controlling process in the Pathway environment. The
PATHMON process maintains configuration-related data; grants links to server classes
in response to requests from TCPs and LINKMON processes; and performs all process
control (starting, monitoring, restarting, and stopping) of server processes and TCPs.

pathname. See OSS pathname.

Pathsend procedures. The set of Guardian procedure calls that provide general access to
Pathway server classes from any process on a Tandem system.

Pathsend process. A process, written as a Guardian program in C, C++, COBOL85, Pascal,

pTAL, or TAL, that makes calls to Pathsend procedures to request services from a
Pathway server. A Pathsend process can be either a standard requester, which initiates
application requests, or a nested server, which is configured as a server class but acts as
a requester by making requests to other servers. A Pathsend process is also known as a
Pathsend requester.

Pathsend program. A Guardian program, written in C, C++, COBOL85, Pascal, pTAL, or

TAL, that makes calls to Pathsend procedures to request services from a Pathway server.
A running Pathsend program is called a Pathsend process. See also Pathsend process.

Pathsend requester. See Pathsend process.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -16
Glossary PATHTCP2

PATHTCP2. The TCP object file, usually identified by the file name
$SYSTEM.SYSTEM.PATHTCP2.

PATHTCPL. The TCP user library object file.

Pathway application. A set of programs that perform online transaction processing tasks in
the Guardian environment, using interfaces defined by Tandem. A Pathway application
can include SCREEN COBOL requesters, Pathsend requesters, and Pathway servers
running on Tandem NonStop systems. It can also include GDSX front-end processes and
clients that use RSC or POET. See also NonStop TUXEDO application.

Pathway application development environment. A set of tools supporting the development

of applications for the Pathway transaction processing environment. Depending on the
customer’s needs and software configuration, this set of tools could include the
SCREEN COBOL compiler and the SCREEN COBOL Utility Program (SCUP), the
POET application development tools, the Pathmaker product, and various tools from
Tandem Alliance partners. See also Pathway transaction processing environment.

Pathway environment. See Pathway transaction processing environment.

Pathway management programming interface. A set of programmatic commands that

allow users to write management application programs that communicate directly with
the PATHMON process for configuration and management. This interface is based on
the Subsystem Programmatic Interface (SPI) within the Distributed Systems
Management (DSM) software. Programmatic commands communicating with the
PATHMON process use the Pathway subsystem ID. See also Pathway subsystem and
subsystem ID.

Pathway monitor process. See PATHMON process.

Pathway object. An object in the Pathway transaction processing environment. The set of
Pathway objects is a more inclusive set than the set of PATHMON-controlled objects:
for example, a LINKMON process is a Pathway object but not a PATHMON-controlled
object. See also object, PATHMON-controlled object, and Pathway transaction
processing environment.

Pathway Open Environment Toolkit (POET). A set of programs and utilities that helps
programmers create and run client/transaction server applications. In the POET
environment, the client is a program running on a workstation, and the server is a
Pathway server running on a Tandem NonStop system. POET uses the services of the
Remote Server Call (RSC) product. The programming tools provided by POET include
a simplified programming interface, name mapping, and conversion mapping.

Pathway server. A server process or program in the Pathway transaction processing

environment. See also NonStop TUXEDO server.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -17
Glossary Pathway subsystem

Pathway subsystem. The PATHMON environment components to which SPI commands are
sent under the Pathway subsystem ID and which generate EMS event messages with the
Pathway subsystem ID. All SPI commands for the Pathway subsystem are sent to the
PATHMON process, but the processsing for the command might involve other
processes, such as a TCP. Likewise, all EMS messages from the Pathway subsystem are
generated by the PATHMON process, but the information might originate from another
process. See also subsystem ID.

Pathway system. A term formerly used for the set of objects managed by a particular
PATHMON process; now called PATHMON environment. See PATHMON environment.

Pathway transaction processing environment. A run-time environment consisting of

Tandem’s transaction-processing products for the Guardian operating environment. This
term is often shortened to “Pathway environment.” Depending on the customer’s needs
and software configuration, the Pathway environment could include NonStop TS/MP,
the run-time portions of Pathway/TS (the TCP and the SCREEN COBOL run-time
environment), NonStop TM/MP, GDSX processes, the run-time portion of the RSC
product, the POET run-time environment, and the TRANSFER delivery system (when
used as a workflow aid in transaction processing). See also NonStop TUXEDO
transaction processing environment.

Pathway translation server for the NonStop TUXEDO system. A server process, provided
by Tandem as part of the NonStop TUXEDO product, that allows a Pathway (SCREEN
COBOL or Pathsend) requester to use the services of a NonStop TUXEDO server. The
translation server thus acts as a gateway process between the Pathway environment and
the NonStop TUXEDO environment. Requesters that use this translation server must
include special information in the header of each request message to identify the target
NonStop TUXEDO application and service.

Pathway/TS. A Tandem product that provides tools for developing and interpreting screen
programs to support OLTP applications in the Guardian environment on Tandem
NonStop systems. Pathway/TS screen programs communicate with terminals and
intelligent devices. Pathway/TS includes the TCP, the SCREEN COBOL compiler and
run-time environment, and the SCREEN COBOL Utility Program (SCUP). It requires
the services of the NonStop TS/MP product. See also NonStop Transaction Services/MP
(NonStop TS/MP).

PIN. See process identification number (PIN).

POBJ. The default prefix, or file-name root, used by the SCREEN COBOL compiler in
naming its output files. If no prefix is specified in the RUN command to run the
compiler, the compiler produces a code file named POBJCOD, a directory file named
POBJDIR, and (if the SYMBOLS option is enabled) a symbols file named POBJSYM.

POET. See Pathway Open Environment Toolkit (POET).

Portable Transaction Application Language (pTAL). A machine-independent systems

programming language based on TAL. The pTAL language excludes architecture-
specific TAL constructs and includes new constructs that replace the architecture-
specific constructs. See also Transaction Application Language (TAL).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -18
Glossary primary process

primary process. The currently active process of a process pair in the Guardian environment.
See also backup process and process pair.

process. (1) A unique execution of a program in the Guardian environment. (2) An entity in
the OSS environment consisting of an address space, a single thread of control that
executes within that address space, and the system resources required by that thread of
control. See also process type.

process identification number (PIN). An unsigned integer that identifies a process within a
processor module in a Tandem NonStop system.

process management. The act of configuring, creating, and initializing processes; the
monitoring and stopping of processes; and the recovery of failed processes. The
PATHMON process provides process management functions for OLTP applications on
Tandem NonStop systems.

process pair. A fault-tolerant arrangement of processes in the Guardian environment,

whereby two processes in separate processors share the same name and execute identical
code. One process functions as the primary process and the other functions as the
backup process. The two processes are kept in sync through checkpoint messages sent
from the primary to the backup process. If the primary process fails, the backup process
is notified that it is now the primary, and it resumes the application work from the last
valid checkpoint message.

process type. An attribute of a server class indicating whether the server processes in that
server class are Guardian processes or OSS processes. See also process.

program file. An executable object file. See also object file.

PROGRAM object. A template for creating and starting temporary TERM objects. See also
temporary TERM object.

pseudocode file. A file containing compiled code that is interpreted by software instead of
being executed by the hardware. The SCREEN COBOL compiler produces pseudocode
to be interpreted by the TCP. See also object file.

pseudo Pathsend procedure. In the Extended General Device Support (GDSX) software,
one of the TSCODE-supported procedures that have Pathsend procedure counterparts.

pTAL. See Portable Transaction Application Language (pTAL).

reduced instruction-set computing (RISC). A processor architecture based on a relatively

small and simple instruction set, a large number of general-purpose registers, and an
optimized instruction pipeline that supports high-performance instruction execution. See
also complex instruction-set computing (CISC).

relative pathname. An OSS pathname that does not begin with a slash (/) character.
A relative pathname is resolved beginning with the current working directory. See also
OSS pathname, absolute pathname, and current working directory.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -19
Glossary Remote Duplicate Database Facility (RDF)

Remote Duplicate Database Facility (RDF). The Tandem software product that assists in
disaster recovery for OLTP production databases, monitors database updates audited by
the TMF subsystem on a primary system, and applies those updates to a copy of the
database on a remote system.

Remote Server Call (RSC). A Tandem software product that facilitates client/server
computing, allowing personal computer (PC) or workstation applications running under
Microsoft Windows software or the MS-DOS or OS/2 operating system to access
Pathway server classes and Guardian processes. Transactions are transmitted from the
PC or workstation application (the client) to a Pathway application running on a Tandem
NonStop system (the server) by means of a supported communications protocol, such as
NETBIOS, TCP/IP, or an asynchronous connection.

reply message. The part of an interprocess communication message that is formatted by a

server and returned to the original requester. The message contains the results of the
server’s work. See also request message.

reply translation header. A group of header fields that the Pathway translation server for the
NonStop TUXEDO system inserts at the beginning of each reply message it relays from
the application service to a SCREEN COBOL or Pathsend requester. This header
indicates whether the request was processed successfully and whether the reply contains
data returned by the application service. See also request translation header.

requester. A process or program that runs in the Guardian environment on a Tandem

NonStop system and requests services from a server process. For example, a SCREEN
COBOL program is a requester program that is interpreted by the terminal control
process (TCP), which provides link access to Pathway server classes. Another type of
requester program makes requests through Pathsend procedure calls; such a requester
uses the LINKMON process for link access to server classes. A third type of requester
communicates with server processes directly by calling the Guardian WRITEREAD
procedure; this kind of requester does not use server classes. A requester is a specific
type of client. See also client, server, and requester/server model.

requester/server model. A model for application design that divides the tasks of data input,
data manipulation, and data output between two basic types of process: requesters and
servers. A requester sends a request to a server. The server takes the requested action and
then replies to the requester. The requester and server may reside on the same processor
or on different processors. This model is used for interprocess communication in the
Guardian environment. See also requester and server.

request message. The part of an interprocess communication message that is formatted by a

requester and sent to a specific server. The message contains any data and instructions
needed by the server to perform its processing. See also reply message.

request/response model. A model for requester/server communication in which a requester

passes a single request to a server process and receives a single response. See also
conversational model and request/response server.

request/response server. A server that provides request/response services. A service of type

request/response is handled like a procedure and has the following properties: it is

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -20
Glossary request translation header

executed until completion, it does not have any dialog with the requester, and it sends
back a return value to the requester. A request/response server is analogous to a context-
free Pathway server.

request translation header. A group of header fields that must be included at the beginning
of each request message a SCREEN COBOL or Pathsend requester sends to the
Pathway translation server for the NonStop TUXEDO system. This header specifies the
NonStop TUXEDO application service for which the message is destined, the NonStop
TUXEDO buffer types of the request and reply messages as seen by the service, and
options that modify the invocation of the service. The translation server removes this
header from the request message before sending it to the application service. See also
reply translation header.

reserved word. A word that can be used only as a keyword.

resources. The components of a computer system that work together to process transactions.
Terminals, workstations, CPUs, memory, I/O controllers, disk drives, processes, files,
and applications are examples of resources.

response time. The amount of time it takes to receive a response from the system after
initiating a request message (for example, by pressing a function key).

retryable operation. An operation that can be interrupted and repeated an indefinite number
of times without affecting the consistency of the database; for example, all read
operations are retryable.

RISC. See reduced instruction-set computing (RISC).

root directory. An OSS directory associated with a process that the system uses for pathname
resolution when a pathname begins with a slash (/) character. See also OSS pathname.

RSC. See Remote Server Call (RSC).

scalability. The ability to increase the size and processing power of an online transaction
processing system by adding processors and devices to a system, systems to a network,
and so on, and to do so easily and transparently without bringing systems down.
Scalability is also sometimes called expandability.

SCOBOLX. The object file for the SCREEN COBOL compiler program. This name is given
in a TACL command to invoke the compiler. See also SCREEN COBOL.

screen. A group of data fields that represent formatted data to be displayed on a terminal.
A screen is defined by a screen description entry in the Screen Section of a SCREEN
COBOL program. There are two types of screen: base screens and overlay screens. See
also base screen, overlay screen, and screen description entry.

SCREEN COBOL. A procedural language developed by Tandem and based on COBOL that
is used to define and control screen displays on terminals and other input/output devices.
SCREEN COBOL allows programmers to write requester programs that communicate
with operator terminals and intelligent input/output devices, and that send data to server

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -21
Glossary SCREEN COBOL Utility Program (SCUP)

processes that manage application databases. SCREEN COBOL programs are compiled
into pseudocode form by the SCREEN COBOL compiler and then interpreted by the
TCP. See also terminal control process (TCP).

SCREEN COBOL Utility Program (SCUP). A utility that provides control and
manipulation of SCREEN COBOL object files.

screen description entry. A declaration of a base screen, and, optionally, an overlay screen,
in the Screen Section of a SCREEN COBOL program. See also screen, base screen, and
overlay screen.

screen-oriented requester. A SCREEN COBOL requester that sends data from working
storage to the display screen of a terminal by way of screen templates defined in the
Screen Section of the Data Division. Similarly, such a requester receives data from the
terminal into working storage by way of Screen Section templates. It uses ACCEPT and
DISPLAY statements in the Procedure Division to interact with the display terminals.
Standard SCREEN COBOL requesters are screen-oriented. See also message-oriented
requester.

screen overlay area. See overlay area.

screen program. A SCREEN COBOL requester program. See also SCREEN COBOL.

Screen Section. A section in the Data Division of a SCREEN COBOL source program that
describes the types and locations of fields in screens that can be displayed on a terminal.

SCUP. See SCREEN COBOL Utility Program (SCUP).

SEND operation. In SCREEN COBOL, an operation in which a transaction request message

is sent to a server process and a reply is received back from the server process. See also
server-class send operation.

server. (1) A process or program that provides services to a client or a requester. Servers are
designed to receive request messages from clients or requesters; perform the desired
operations, such as database inquiries or updates, security verifications, numerical
calculations, or data routing to other computer systems; and return reply messages to the
clients or requesters. A server process is a running instance of a server program. (2) A
combination of hardware and software designed to provide services in response to
requests received from clients across a network. For example, Tandem’s Himalaya
servers provide transaction processing, database access, and other services. (In the
NonStop TS/MP and Pathway/TS manual set, the word “server” is generally used only
when definition 1 is meant; for definition 2, “system” is usually used instead of
“server.”) See also client, requester, client/server model, and requester/server model.

server class. A group of duplicate copies of a single server process, all of which execute the
same object program. Server classes are configured through the PATHMON process.

server-class send operation. The sending of a message to a Pathway server class by making
a call to a Pathsend procedure. The SERVERCLASS_SEND_,

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -22
Glossary SERVER object

SERVERCLASS_DIALOG_BEGIN_, and SERVERCLASS_DIALOG_SEND_

procedures perform server-class send operations. See also SEND operation.

SERVER object. A definition of a server class within the configuration of a PATHMON

process.

service. A function performed by a server process or program on behalf of a requester or

client. A server can perform one or several services. The concept of a service is built
into the design of the BEA TUXEDO system and the NonStop TUXEDO system; for
these products, a service is a module of application code that carries out a service
request.

simple token. In the Subsystem Programmatic Interface (SPI), a token consisting of a token
code and a value that is either a single elementary field, such as an integer or a character
string, or a fixed (nonextensible) structure. See also extensible structured token and
token (definition 2).

single-threaded. A programming model that provides a single thread of control within a

program. For example, a single-threaded server handles only one request at a time and
must complete that request before accepting another. See also thread and multithreaded.

special register. A data item defined by the SCREEN COBOL compiler, rather than
explicitly in the program. Each special register has a particular purpose and should be
used only as defined. The SCREEN COBOL language defines a different set of special
registers from those defined by the standard COBOL language.

SPI. See Subsystem Programmatic Interface (SPI).

static server. A server process that the PATHMON process creates when a START SERVER
command is issued. The PATHMON process starts the number of static servers defined
by the NUMSTATIC attribute for the server class. See also dynamic server.

Structured Query Language (SQL). A relational database language used to define,

manipulate, and control databases.

subsystem. In the context of the Subsystem Programmatic Interface (SPI) and the Event
Management Service (EMS), a process or collection of processes that gives users access
to a set of related resources or services. A subsystem typically controls a cohesive set of
objects. The Pathway subsystem includes PATHMON processes, TCPs, and all other
components of the PATHMON environment.

subsystem ID. In management applications, a data structure that uniquely identifies a

subsystem. The subsystem ID is specified in SPI commands to identify the target
subsystem, and in EMS event messages to identify the source of the event message.
The Pathway subsystem ID applies to all components of the PATHMON environment,
including PATHMON processes and TCPs.

Subsystem Programmatic Interface (SPI). A set of Guardian procedures, message formats,

and definition files that allows management applications to communicate directly with

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -23
Glossary subtype 30 process

subsystem processes, such as the PATHMON process, for configuration and control of
objects and for event management.

subtype 30 process. A nonprivileged process that simulates terminals and communications

devices.

subvolume. A related set of files, as defined by the user, within the Guardian environment.
The name of a subvolume is the third of the four parts of a file name.

swap file. A temporary file created by a process for temporary storage: for example, by the
Kernel Managed Swap Facility (KMSF) on behalf of a TCP for temporary storage of
terminal context.

syncdepth. A parameter to Guardian procedure calls that sets the maximum number of
operations or messages that a process is allowed to queue before action must be taken or
a reply must be performed.

sync ID. A value used in the Guardian environment to determine whether an I/O operation
has finished. In active backup programming, a file’s sync ID is used to prevent the
backup process from repeating I/O operations that have already been completed by the
primary process.

system name. (1) The first of the four parts of a Guardian file name; also called a node name.
(2) A name that identifies the Tandem system on which a PATHMON process is
running. In SCREEN COBOL programs, this name is given in SEND statements. (3) A
SCREEN COBOL word that identifies part of the Tandem operating environment; a
name can be associated with function keys or terminal display attributes.

System/T. The transaction monitor for a NonStop TUXEDO system. See also NonStop
TUXEDO application.

TACL. See Tandem Advanced Command Language (TACL).

TAL. See Transaction Application Language (TAL).

Tandem Advanced Command Language (TACL). The user interface to the Tandem
NonStop Kernel in the Guardian environment. The TACL product is both a command
interpreter and a command language.

Tandem Alliance. Tandem’s marketing and technical support program for third-party
vendors, which encourages the development of application software for the Pathway
environment. The Alliance attracts software developers, value-added resellers, and other
vendors who can provide industry-specific and general business applications for Tandem
customers.

Tandem NonStop Kernel. The operating system for Tandem NonStop systems. The
operating system does not include any application program interfaces.

Tandem NonStop Series (TNS). Tandem computers that support the Tandem NonStop
Kernel and that are based on complex instruction-set computing (CISC) technology.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -24
Glossary Tandem NonStop Series/RISC (TNS/R)

TNS processors implement the TNS instruction set. See also complex instruction-set
computing (CISC) and Tandem NonStop Series/RISC (TNS/R).

Tandem NonStop Series/RISC (TNS/R). Tandem computers that support the Tandem
NonStop Kernel and that are based on reduced instruction-set computing (RISC)
technology. TNS/R processors implement the RISC instruction set and are upwardly
compatible with the TNS system-level architecture. See also reduced instruction-set
computing (RISC) and Tandem NonStop Series (TNS).

task. The sequence of SCREEN COBOL program units that are executed as a result of a
PATHCOM START TERM or RUN PROGRAM command or an SPI START TERM or
START PROG command.

TCLPROG file. A SCREEN COBOL object library file.

TCP. See terminal control process (TCP).

TDA. See terminal data area (TDA).

TEDIT. A Tandem text editor used to create or modify a source text file. Also called PS Text
Edit.

tell message. An informational message sent by PATHCOM or a management application to

one or more terminals controlled by a SCREEN COBOL program, to be displayed for
the terminal operators.

TELL object. A temporary object used in PATHCOM and SPI commands to define a tell
message.

temporary TERM object. A TERM object created by the PATHMON process when a
PATHCOM RUN PROGRAM command or an SPI START PROG command is issued.
Temporary TERM objects are deleted by the PATHMON process when application
processing is completed or when a STOP TERM or ABORT TERM command is issued.
Names of temporary TERM objects begin with a number. See also configured TERM
object and TERM object.

terminal. An I/O device capable of sending and receiving information over communications
lines.

terminal context. Data maintained by a TCP for each active terminal under its control.

terminal control process (TCP). A process used for terminal management and transaction
control, provided by Tandem as part of the Pathway/TS product. A TCP is a
multithreaded process that interprets compiled SCREEN COBOL requester programs
(screen programs) in the user’s application, executing the appropriate program
instructions for each I/O device or process the TCP is configured to handle. The TCP
coordinates communication between screen programs and their I/O devices or processes
and, with the help of the PATHMON process, establishes links between screen programs
and server processes. See also requester and SCREEN COBOL.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -25
Glossary terminal data area (TDA)

terminal data area (TDA). In SCREEN COBOL, the area that the TCP allocates for
terminal context data. The MAXTERMDATA parameter of the PATHCOM SET TCP
command defines the upper limit for this data area.

TERM object. A definition of a task that uses a SCREEN COBOL program to control an
input/output device such as a terminal or workstation, or an input/output process such as
a front-end process. A TERM object can be either explicitly configured with an ADD
command or created by the PATHMON process through a PATHCOM RUN PROGRAM
or SPI START PROG command. TERM objects created by the latter method are called
temporary TERM objects. See also configured TERM object and temporary TERM
object.

thaw condition. A condition in which prohibition of communication between a terminal and

a server class is lifted. See also freeze condition.

thread. A task that is separately dispatched and that represents a sequential flow of control
within a process (for example, a TCP).

throughput. The number of transactions a system can process in a given period, such as one
second.

TMF. See Transaction Management Facility (TMF) subsystem.

TMF level recovery. Recovery of the database to a consistent state through the use of the
TMF subsystem. When a failure occurs, the TMF subsystem allows the application to
back out the entire transaction, returning the contents of the database to the values it
held when the transaction was started. The application can then retry the transaction.

TNS. See Tandem NonStop Series (TNS).

TNS/R. See Tandem NonStop Series/RISC (TNS/R).

token. (1) An attribute control element in the CONTROLLED clause of a SCREEN COBOL
program, which allows run-time control of display attributes. This token consists of an
attribute identifier and an attribute value. (2) In the Subsystem Programmatic Interface
(SPI), a distinguishable unit in a message. An SPI token consists of an identifier (token
code or token map) and a token value. Programs place tokens in an SPI buffer by calling
the SSPUT procedure and retrieve them from the buffer by using the SSGET procedure.

transaction. An operation or a series of operations that retrieves and updates information to

reflect an exchange of goods or services. In the process of retrieving and updating
information, a transaction transforms a database from one consistent state to another.
The TMF subsystem treats a transaction as a single unit; either all of the changes made
by a transaction are made permanent (the transaction is committed) or none of the
changes are made permanent (the transaction is aborted).

Transaction Application Language (TAL). A systems programming language with many

features specific to stack-oriented Tandem NonStop systems. See also Portable
Transaction Application Language (pTAL).

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -26
Glossary transaction backout

transaction backout. A TMF subsystem activity in which the effects of a partially completed
transaction are canceled.

Transaction Delivery Process (TDP). A multithreaded gateway process, part of the Remote
Server Call (RSC) product, that runs on a Tandem NonStop system. The TDP can be
replicated and can manage many connections and workstations at one time, as well as
multiple sessions from each workstation.

transaction identifier. A unique name that the TMF subsystem assigns to a transaction.

transaction log. See audit trail.

transaction management. The ability to coordinate transaction control functions, such as

beginning and ending transactions, committing or aborting transactions, and recovering
transactions.

Transaction Management Facility (TMF) subsystem. The major component of the

NonStop TM/MP product, which protects databases in online transaction processing
environments. To furnish this service, the TMF subsystem manages database
transactions, keeps track of database activity through audit trails, and provides database
recovery methods. See also NonStop Transaction Manager/MP (NonStop TM/MP) and
transaction.

transaction mode. The operating mode of a requester program between the execution of a
BEGINTRANSACTION procedure call or statement and the execution of the associated
ENDTRANSACTION or ABORTTRANSACTION call or statement.

transaction processing. See online transaction processing (OLTP).

transaction workload. The number of transactions to be processed by an online transaction

processing application.

TRANSFER delivery system. An information delivery system that enables organizations to

move and manage information efficiently within a single Tandem system or across a
network of systems. The TRANSFER delivery system supports communications
between users, I/O devices, and processes in the Guardian environment.

TSCODE. The object code for the part of a GDSX process that is supplied by Tandem.
TSCODE includes generic routines and services that support the development of a
multithreaded, fault-tolerant front-end process. See also USCODE and Extended
General Device Support (GDSX).

tuning. The process by which a system manager allocates and balances resources for
optimum system performance.

UMP. See unsolicited-message processing (UMP).

unsolicited message. A message that is sent to a SCREEN COBOL program and includes
application-dependent information to be processed by the program. Although the
program does not do anything to initiate the message, the message must conform to the

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -27
Glossary unsolicited-message processing (UMP)

format defined by the program. The message is sent first to the TCP. It contains a header
with information that is used by the TCP and a body with information that the TCP
delivers to the SCREEN COBOL program.

unsolicited-message processing (UMP). The feature that allows terminals running SCREEN
COBOL requesters to accept and reply to unsolicited messages sent to them by Guardian
processes outside of the PATHMON environment.

USCODE. The object code for the part of a GDSX process that is developed by the user to
provide device or access-method specifics such as control operations or data-stream
translation. USCODE is bound with TSCODE to produce an operational GDSX process.
See also TSCODE and Extended General Device Support (GDSX).

user conversion procedure. A procedure that lets users make their own validation checks or
conversions of data passed between a SCREEN COBOL program and a terminal screen
or intelligent device.

ViewPoint application. An extensible interactive application for managing operations in the

Guardian environment. It provides tools for interacting with multiple Tandem
subsystems, including PATHCOM, allowing a system manager to easily control an
integrated Tandem system from one location.

volume. A disk drive, or a pair of disk drives that forms a mirrored disk, in the Guardian
environment. The name of a volume is the second of the four parts of a file name.

$RECEIVE. A special Guardian file name through which a process receives and optionally
replies to messages from other processes by using Guardian procedure calls. This file is
analogous to a request queue defined for a NonStop TUXEDO server.

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Glossary -28
 Index
A Applications, Pathway (continued)
transaction processing
Aborted dialog (error 929) 6-12 scenario 1-15/1-17
Aborted dialogs, detecting 4-16/4-17 Applications, programming
Aborted send operation (error 918) 6-9 examples
Aborted transaction (error 934) 6-14 nested server program B-53/B-67
Aborting transactions 2-8, 2-23, B-15 Pathsend requester
ABORTTRANSACTION procedure 4-6 program B-1/B-52
Access Control Server (ACS) 3-7 skeleton Pathsend
Application development tools 1-3 program 2-14/2-15
Application generator, Pathmaker 1-14 skeleton server program 2-25/2-27
Application Transaction Monitor Interface for TMF subsystem 4-5/4-10
(ATMI) 3-11 invoking Pathsend procedures
Applications, designing from C and C++ programs 5-2
batch processing 2-27 from COBOL85 programs 5-3
database 2-9/2-10 from Pascal programs 5-4
example, OLTP 2-1/2-8 from TAL and pTAL programs 5-5
requester programs 2-11/2-19 overview 1-1
server programs 2-19/2-27 Pathsend environment 1-10/1-11
transactions 2-1/2-8 Pathsend requesters
Applications, Pathway See Requesters, writing Pathsend
client/server capabilities 1-12 Pathsend usage
data integrity 1-4 considerations 5-23/5-28
development of 1-3/1-4, 1-14/1-15 Pathway servers
distributed processing 1-7 See Server programs, writing
expansion fundamentals 1-7 writing Pathsend requesters 3-1/3-11
fault tolerance 1-5/1-6 writing Pathway servers 4-1/4-17
introduction 1-1 Application-Transaction Monitor Interface
(ATMI) 1-13
managing 1-4
ASSIGNs 3-7, B-5, B-12, B-18, B-55
overview 1-7
ATM devices 2-16
performance of 1-6, 1-8
ATMI functions, NonStop TUXEDO 1-13,
requester programs 1-9/1-10 3-11
security fundamentals 1-6 Audit trails, TMF 1-4, 2-8, 4-5
server classes 1-8 Audited and nonaudited servers 2-24
server languages 1-8 Audited files 4-7
server processes 1-8 Automatic retry, Pathsend requesters 1-9,
support for NonStop TUXEDO 3-6
environment 1-13 Autonomy, node 1-8
NonStop TS/MP Pathsend and Server Programming Manual– 132500
Index- 1
Index B

AWAITIOX procedure 5-7, 5-13, 5-17, COBOL85

5-25, 5-28 example Pathsend requester
program B-53/B-67
B example server program B-53/B-67
Back-end process 2-16, 2-24 for Pathsend requesters 1-10
BASIC, Extended, for Pathway servers 1-8 for Pathway servers 1-8
Batch processing 1-7, 2-13, 2-27 invoking Pathsend procedures 5-3
Bounds error (error 912) 6-6 Code, standardizing and testing 1-3
Buffer length invalid (error 911) 6-6 Colon, in Pathsend procedure
parameters 5-2
Buffer limits A-1
Common Run-Time Environment
(CRE) 4-3
C Compilers provided 1-14
C and C++ languages Concurrent processing 2-7/2-8, 4-5
for Pathsend requesters 1-10 Condition code
for Pathway servers 1-8 considerations 5-23
invoking Pathsend procedures 5-2 register 5-2
Calls Context-free
See Procedure calls, Pathsend, servers, using with context-sensitive
individual procedures requesters 3-8, 4-4
CANCEL procedure 5-25/5-26 Context-sensitive programming, requesters
Canceling server-class send calls 5-25/5-26 canceling server-class sends 3-10
CANCELREQ procedure 5-25/5-26 failure recovery 3-9/3-10
Changed transid (error 930) 6-13 overview 3-8
Checkpointing resource utilization 3-8
and Pathsend 3-6 using with context-free servers 3-8, 4-4
explanation of 1-5 Context-sensitive programming, servers
Pathsend limitations 2-14 controlling dialogs 4-14
server considerations 4-3 correlating messages with dialogs 4-16
Classes of data in database 2-9 detecting aborted dialogs 4-16/4-17
Clients, NonStop TUXEDO detecting new dialogs 4-14
interoperating with 4-17 functions performed 4-13
Pathsend requesters acting as 3-11 handling dialog messages 4-14/4-15
Client/server computing 2-15 managing dialogs 4-16
capabilities 1-12 overview 4-13
development tools 1-15 Control block examples B-8, B-13, B-14,
B-23, B-25
Conversational mode 2-12
Conversational servers 4-1
CPU halts and LINKMON processes 4-3

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index- 2
Index D

CRE (Common Run-Time Design

Environment) 4-3 application example 2-1/2-8
Creation failure, servers (error 916) 6-8 batch processing applications 2-27
Creator default for PATHMON names 3-7 database 2-9/2-10
CRE_Receive_Read_ procedure 4-14 requester programs 2-11/2-19
Crossref product 1-14 server programs 2-19/2-27
transactions 2-1/2-8
D Detecting aborted dialogs 4-16/4-17
Data Development
analyzing flow of 2-2/2-3 considerations 1-3/1-4
classes of 2-9 tools 1-14/1-15
integrity 1-4 Dialog abort system message (message
Database -121) 4-16
concurrency 2-7/2-8 Dialogs
consistency 1-4, 2-7/2-8 aborted (error 929) 6-12
consistency and concurrency 4-5 continuing, aborting, terminating 4-16
fields in 2-9 controlling 4-14
files in, normalizing 2-9 correlating messages with 4-16
integrity 1-8 detecting aborted 4-16/4-17
logical design 2-9 detecting new 4-13, 4-14
management systems 2-10 ended (error 931) 6-13
physical design 2-10 handling messages 4-14/4-15
records in 2-9 invalid (error 926) 6-11
relational 2-10, 4-5 models, types of 3-9
DBMS (database management system) model, specifying 5-10, 5-15
See Database outstanding (error 933) 6-13
Deadlocks too many (error 927) 6-12
for transactions 4-10 using TMF subsystem 4-15
with nested servers 2-22 Distributed processing 1-6, 1-7
Debugging Distributed transaction processing
(DTP) 1-7
Pathway servers 4-11/4-12
Dynamic links 3-4
tools for 1-14
Declarations, global B-9
Declaratives, example B-58

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index- 3
Index E

E Extended General Device Support (GDSX)

processes 1-10, 2-16/2-18, 2-24
Early replies 2-23
Enscribe product 2-10, 4-5 F
Entry-sequenced files 2-10
Failure recovery
Errors
context-sensitive requesters 3-9/3-10
See also Errors, file-system; Failure
recovery; individual errors fault-tolerant programming
dialog abort system message (message requesters 3-5/3-6
-121) 4-16 servers 4-10
FESCErr (error 233) 5-6, 5-7, 5-12, LINKMON limit errors 3-4
5-13, 5-17 LINKMON process 4-3
handling Pathsend 6-1/6-14 overview 3-3/3-4
Pathsend server TIMEOUT 5-27 Pathsend and TMF subsystem 3-5
returned by servers in replies 4-14 Pathsend requester 4-2
server process 3-4 server process 3-4
Errors, file-system servers and TMF subsystem 4-6/4-10
associated with Pathsend Fault tolerance
errors 6-1/6-14
and servers using the TMF
FEBoundsErr (error 22) 5-21 subsystem 2-23
FEContinue (error 70) 4-14, 5-7, 5-13 overview 1-5, 1-6
FEEOF (error 1) 4-14 Pathsend requester
FEInUse (error 12) 6-3 programming 3-5/3-6
FEInvalOp (error 2) 5-21 Pathway server programming 4-10
FEMissParam (error 29) 5-21 FEBoundsErr (error 22) 5-21
FENoBufSpace (error 31) 6-9 FEContinue (error 70) 3-10, 4-14, 5-7, 5-13
FENoDiscSpace (error 43) 6-14 FEEOF (error 1) 3-9, 4-14
FENoSuchDev (error 14) 6-3, 6-4 FEInUse (error 12) 6-3
FEOK (error 0) 4-14 FEInvalOp (error 2) 5-21
FEPathDown (error 201) 6-3, 6-5 FEMissParam (error 29) 5-21
FESecViol (error 48) 6-3, 6-4 FENoBufSpace (error 31) 6-9
FETimeout (error 40) 6-3 FENoDiscSpace (error 43) 6-14
FETimeout (error 440) 6-4 FENoSuchDev (error 14) 6-3, 6-4
Requesting process has no… (error FEOK (error 0) 4-14
75) 4-6 FEPathDown (error 201) 6-3, 6-5
timeout error (error 40) 4-11/4-12 FEScChangedTransid (error 930) 6-13
Event Management Service (EMS) 1-4 FEScDialogAborted (error 929) 3-4, 3-9,
Examples 6-12
nested server program B-53/B-67 FEScDialogEnded (error 931) 6-13
Pathsend requester program B-1/B-52 FEScDialogInvalid (error 926) 6-11
Extended BASIC, for Pathway servers 1-8 FEScDialogOutstanding (error 933) 6-13
NonStop TS/MP Pathsend and Server Programming Manual– 132500
Index- 4
Index G

FESCErr (error 233) 5-6, 5-7, 5-12, 5-13, FEScTransactionAborted (error 934) 6-14
5-17 FEScUnknownServerClass (error 914) 6-7
FEScError (error 233) 3-9 FESecViol (error 48) 6-3, 6-4
FEScInvalidBufferLength (error 911) 6-6 FETimeout (error 40) 6-3, 6-4
FEScInvalidFlagsValue (error 909) 6-6 Fields, database 2-9
FEScInvalidPathmonName (error 901) 6-2 FILEINFO 5-26
FEScInvalidSegmentId (error 907) 6-5 Files
FEScInvalidServerClassName (error database 2-9/2-10
900) 6-2
entry-sequenced 2-10
FEScInvalidTimeoutValue (error 919) 6-9
I/O, in SCREEN COBOL
FEScLinkmonConnect (error 947) 6-14 requesters 2-12
FEScMissingParameter (error 910) 6-6 key-sequenced 2-10
FEScNoSegmentInUse (error 908) 6-5 relative 2-10
FEScNoSendEverCalled (error 906) 6-5 unstructured 2-10
FEScNoServerLinkAvailable (error File-number parameter 5-26
905) 4-3, 6-5
File-system errors
FEScOutstandingSend (error 928) 6-12
See Errors, file-system
FEScParameterBoundsError (error 912) 6-6
FILE_GETRECEIVEINFO_
FEScPathmonConnect (error procedure 4-14, 4-16
902) 4-11/4-12, 6-3
flag parameter 3-9
FEScPathmonMessage (error 903) 6-4
Flags, invalid value (error 909) 6-6
FEScPathmonShutDown (error
915) 4-11/4-12, 6-7 FORTRAN, for Pathway servers 1-8
FEScPFSUseError (error 920) 6-9 Front-end process 2-16/2-18
FEScSendOperationAborted (error 918) 6-9 Frozen server class (error 913) 6-7
FeScSendOperationAborted (error Fundamentals of Tandem NonStop
918) 5-27 systems 1-3/1-7
FEScServerClassFrozen (error 913) 6-7
FEScServerClassTmfViolation (error G
917) 3-5, 6-8 GDSX (Extended General Device Support)
FEScServerCreationFailure (error 916) 6-8 processes 1-10, 2-16/2-18, 2-24
FEScServerLinkConnect (error 904) 3-4, Guardian operating environment
6-4 distributed processing in 1-7
FEScTooManyDialogs (error 927) 6-12 processes in 1-5
FEScTooManyPathmons (error 921) 6-9 security features of 1-6
FEScTooManyRequesters (error 925) 6-11 servers in 1-8, 2-19
FEScTooManySendRequests (error
924) 6-11
FEScTooManyServerClasses (error
H
922) 6-10 Halts, CPU 4-3
FEScTooManyServerLinks (error 923) 4-3,
6-10
NonStop TS/MP Pathsend and Server Programming Manual– 132500
Index- 5
Index I

I L
IDS Languages supported
See Intelligent device support (IDS) Pathsend requesters 1-10
Industrial robots 2-16 Pathway servers 1-8
Input file structure example B-6 Limits, Pathsend environment A-1
Inspect product 1-14 link denied (error 4) 4-3
Intelligent device support (IDS) Link management, with LINKMON
description 2-12 process 1-11, 2-13
GDSX programming for 2-17 LINKMON process
overview 1-7 connect error (error 947) 6-14
RSC requesters 1-12, 2-15 description 1-11
Intelligent mode 2-12 failures 4-3
Interoperation GDSX, relationship to 2-17/2-18
of Pathsend requesters with NonStop limits errors 3-4
TUXEDO servers 1-13 RSC, relationship to 1-12, 2-15
of Pathway servers with NonStop servers, relationship to 4-2
TUXEDO requesters (clients) 1-13 TMF transaction identifiers 5-24
with NonStop TUXEDO requesters Links
(clients) 4-17
allocating space for 4-3
with NonStop TUXEDO servers 3-11
server connect error (error 904) 6-4
Interprocess communication,
Pathsend 1-11, 3-2 static and dynamic 3-4
Invalid buffer length (error 911) 6-6 too many server links (error 923) 6-10
Invalid dialog (error 926) 6-11 Locking of records 4-8, 4-10
Invalid flagsValue (error 909) 6-6
Invalid PATHMON name (error 901) 6-2 M
Invalid server class name (error 900) 6-2 Manageability
Invalid timeout value (error 919) 6-9 of Pathway applications 1-4
I/O provided by server classes 1-8
cancel outstanding, example B-19 Management interfaces 1-8, 2-19
complete outstanding, example B-20 See also PATHCOM interface, SPI
starting example B-13 (Subsystem Programmatic Interface)
Message buffer
K SERVERCLASS_DIALOG_BEGIN_
procedure 5-8
Key field, database 2-9 SERVERCLASS_DIALOG_SEND_
Key-sequenced files 2-10 procedure 5-14
SERVERCLASS_SEND_
procedure 5-18

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index- 6
Index N

Messages NONSTOP parameter 4-10

dialog abort system message (message NonStop SQL/MP (Structured Query
-121) 4-16 Language/MP) product 2-10, 4-5
error, checking for 2-14 NonStop systems 1-5/1-7
example print to terminal and NonStop Transaction Manager/MP
abend B-15 (NonStop TM/MP)
for checkpointing 1-5 advantages 1-3
handling in a dialog 4-14/4-15 application restrictions 4-7
processing example B-60 audited files 4-7
receiving by servers 4-2 context-free Pathsend requesters 3-5
reply size, specifying maximum 5-9, context-sensitive requesters 3-9/3-10
5-14, 5-19 dialogs, using with 4-15
unrecognizable (error 903) 6-4 fault-tolerant servers 4-10
Missing parameter (error 910) 6-6 grouping transaction operations 4-8/4-9
Modes, terminal 2-12 limits A-1
Modularity 1-8 Pathmaker software 1-15
Multiprocessing 1-6 Pathsend procedure calls 3-5
Multithreading record locking 4-8
GDSX feature 2-16 requester example B-1/B-52
in server design 2-20 retryable operations 4-7
Pathsend requesters 1-9 server application structure 4-5/4-6
server class violation (error 917) 6-8
N server process pairs 2-23
Native System /T clients, NonStop TMF OFF server parameter 3-5
TUXEDO, interoperating with 4-17 transaction deadlocks 4-10
Nested servers 4-4 usage considerations 5-24
deadlocks 2-22 writing servers to use 4-5/4-10
designing 2-22/2-23 NonStop Transaction Services/MP (NonStop
example program B-53/B-67 TS/MP) product xi, 1-1, 1-3
overview 2-13 NonStop TUXEDO
Network security, Pathsend requesters 3-6 applications, interoperation with 1-13,
No send ever called (error 906) 6-5 3-11, 4-17
No server link available (error 905) 6-5 requesters (clients), writing Pathway
servers for 1-13, 4-17
Node autonomy 1-8
servers, writing Pathway requesters
Nonaudited files 4-7 for 3-11
Nonretryable requests, Pathsend 3-6
NonStop Kernel Open System Services
(OSS) server processes
See Open System Services (OSS) server
processes

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index- 7
Index O

Nowait send operations Pascal

errors for 6-1 for Pathsend requesters 1-10
limit on number of A-1 for Pathway servers 1-8
procedure calls for 5-7, 5-13, 5-17 invoking Pathsend procedures 5-4
specifying 5-9, 5-15, 5-20 PATHCOM interface
usage considerations 5-23/5-24, 5-28 description 1-4
No-early-reply rule 2-23 use in managing servers 1-8, 2-19
Pathmaker product 1-14
O PATHMON process
OLTP avoiding coded names 3-7
application design example 2-1/2-8 connect error (error 902) 6-3
development considerations 1-3/1-4 description 1-4
expanding systems for 1-7 fault-tolerance role 1-6
importance of fault tolerance for 1-5 invalid name (error 901) 6-2
manageability 1-4 name, specifying 5-18
Pathway environment 1-3/1-7 relationship to LINKMON
processes 1-11
support for NonStop TUXEDO
environment 1-13 shutdown (error 915) 6-7
transaction processing too many (error 921) 6-9
scenario 1-15/1-17 unrecognizable message (error 903) 6-4
Online transaction processing (OLTP) Pathsend application program interface
See OLTP (API)
Open System Services (OSS) See also Requesters, writing Pathsend;
processes 3-11 individual procedures
Pathway servers 2-19 error handling 6-1/6-14
Open System Services (OSS) server example nested server
processes 1-8, 2-19, 3-11 program B-53/B-67
Operation number, server-class 5-24/5-26 example requester program B-1/B-52
Output record example B-6 failures, LINKMON 4-3
Outstanding send (error 928) 6-12 failures, requester 4-2
OWNER attribute, server 3-7 interprocess communication 1-11, 3-2
limits, programming environment A-1
P LINKMON processes, relationship
to 1-11
Parameter bounds error (error 912) 6-6 nonretryable requests 3-6
Parameter missing (error 910) 6-6 processes using 1-10
Parameter pairs, in Pathsend procedure programming languages supported 1-10
syntax 5-2
requesters, overview 1-9
PARAMs example B-6, B-13
timeout considerations 5-27/5-28

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index- 8
Index P

Pathsend application program interface Performance

(API) (continued) provided by Pathsend requesters 1-9
transaction processing provided by Pathway applications 1-6
scenario 1-15/1-17
provided by server processes 1-8
Pathsend procedure calls
Personal computer support 2-15
See also individual procedures
PFS use error (error 920) 6-9
errors returned 6-1/6-14
POET (Pathway Open Environment Toolkit)
invoking from C and C++ programs 5-2 product 1-3, 1-12, 2-16
invoking from COBOL85 programs 5-3 Portable Transaction Application Language
invoking from Pascal programs 5-4 See pTAL
invoking from TAL and pTAL POS (point-of-sale) devices 2-16
programs 5-5
Presentation services 1-7
overview 5-1
Procedure calls, Pathsend
return errors 6-1
See also individual procedures
usage considerations 5-23/5-28
errors returned 6-1/6-14
use by NonStop TUXEDO
requesters 4-17 invoking from C and C++ programs 5-2
Pathsend requesters invoking from COBOL85 programs 5-3
checkpointing limitations 2-14 invoking from Pascal programs 5-4
description 1-9 invoking from TAL and pTAL
programs 5-5
design considerations 2-13/2-14
overview 3-1, 5-1
program structure 2-14/2-15
return errors 6-1
Pathway applications
summary 5-1
See Applications, Pathway;
Applications, programming; usage considerations 5-23/5-28
Applications, designing Process pairs 1-5, 2-23
Pathway environment Processes
See also Applications, programming; description 1-5
Applications, designing distribution of 1-6
advantages of 1-3/1-7 primary and backup 1-5
products for 1-1, 2-11 replication of 1-6
Pathway servers starting up example B-4
See also Server classes; Server Programming
processes; Server programs;
Applications, programming; See Applications, programming;
Applications, designing Applications, designing
description 1-7 Programming languages
Pathway to TUXEDO translation Pathsend requesters 1-10
server 1-13, 3-11 Pathway servers 1-8
Pathway/TS product xi, 1-1, 1-3 PROGRAM-STATUS special register 4-10

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index- 9
Index Q

pTAL Remote Server Call

for Pathsend requesters 1-10 See RSC (Remote Server Call)
for Pathway servers 1-8 Repeatable requests, Pathsend 3-6
invoking Pathsend procedures 5-5 Reply format, server program 4-1
PWY2TUX translation server 1-13, 3-11 Requesters
See also Applications, programming;
Q Applications, designing; Requesters,
writing Pathsend
Queuing of incomplete transactions 2-23 clients using POET 2-16
clients using RSC 2-15
R debugging 1-14
RDBMS (relational database management description 1-7
system) designing 2-11/2-19
See Database management systems dividing functions with servers 2-19
RDF Pathmaker, using to develop 1-14
See Remote Duplicate Database Facility Pathsend
(RDF)
description 1-9
Read operations, repeatable 4-8
design considerations 2-13/2-14
RECEIVE messages
limits A-1
allocating space 4-3
nested servers 2-22/2-23
determining new dialogs 4-13
program structure 2-14/2-15
Record locking 4-8, 4-10
SCREEN COBOL
Records, database 2-9
description 1-10
Recovery
IDS 2-12
context-sensitive requesters 3-9/3-10
standard 2-12
fault-tolerant programming
too many(error 925) 6-11
requesters 3-5/3-6
transaction processing
servers 4-10 scenario 1-15/1-17
LINKMON limit errors 3-4 types of 1-9, 2-11
LINKMON process 4-3 using GDSX 2-16/2-18
Pathsend and TMF subsystem 3-5 Requesters, writing Pathsend
Pathsend requester failures 4-2 automatic retry 3-6
requester overview 3-3/3-4 avoiding coded PATHMON names 3-7
server process failures 3-4 context-sensitive programming 3-8/3-10
servers and TMF subsystem 4-5/4-10 example program B-1/B-52
Relational database management 2-10, 4-5 failure recovery overview 3-3/3-4
Relative files 2-10 fault-tolerant programming 3-5/3-6
Remote Duplicate Database Facility interprocess communication 3-2
(RDF) 2-11
LINKMON limit errors 3-4

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -10
Index S

Requesters, writing Pathsend (continued) SCREEN COBOL (continued)

NonStop TUXEDO, interoperating unsolicited message processing
with 3-11 (UMP) 2-12
overview 3-3 Screen programs 2-12
procedure calls,list of 3-1 scsend-op-num parameter 5-10, 5-15, 5-20,
retryable requests 3-6 5-24/5-26
security issues 3-6/3-7 Security
server process failures 3-4 for Pathsend requesters 2-14
sharing servers 4-1 Pathsend programming issues 3-6/3-7
TMF subsystem 3-5 system 1-6
using ASSIGNs 3-7 SECURITY attribute, server 3-7
Requesting process has no… (error 75) 4-6 Send operation aborted (error 918) 6-9
Request/response servers 4-1 Send operation outstanding (error 928) 6-12
Resource utilization 3-8 Send requests, too many (error 924) 6-11
Response time 1-6 Server class frozen (error 913) 6-7
Retry mechanism, Pathsend calls 3-5 Server class unknown (error 914) 6-7
See also Fault tolerance Server classes
Retryable requests, Pathsend 3-6 accessing
Retryable server operations, TMF 4-7 See Requesters
Return errors 6-1 description 1-8
Robots, industrial 2-16 fault tolerance role 1-5
RSC (Remote Server Call) limits A-1
role in Pathway environment 1-12, 2-15 names, specifying 5-8, 5-18
security issues 3-7 security for Pathsend requesters 3-7
server programming considerations 4-4 send operation number 5-10, 5-15,
5-20, 5-24/5-26
timeouts, specifying 5-27
S TMF violation (error 917) 6-8
SCF (Subsystem Control Facility), use in Server processes
managing GDSX processes 2-18
benefits 1-8
SCREEN COBOL
description 1-7
context sensitivity not supported 4-13
transaction integrity 1-5
debugging 1-14
transaction processing
devices supported 2-12 scenario 1-15/1-17
Pathmaker application generator 1-14 Server programs
requesters aborting transactions 2-23
description 1-10 audited and nonaudited 2-24
designing 2-12/2-13 debugging 1-14
GDSX alternative 2-18 description 1-7
server programming considerations 4-3 designing 2-19/2-27

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -11
Index S

Server programs (continued) SERVERCLASS_SEND_ procedure

dividing functions with requesters 2-19 canceling 5-25/5-26
example nested server B-53/B-67 example B-31, B-56, B-65
GDSX back-end process, using 2-24 nowait considerations 5-23/5-24, 5-28
languages for 1-8 overview 3-3
nested 2-13 syntax and usage 5-17/5-20
nested servers 2-22/2-23 TMF considerations 5-24
no-early-reply rule 2-23 waited considerations 5-23, 5-27
packaging individual SERVERCLASS_SEND_INFO_ procedure
functions 2-21/2-22 example B-17, B-56, B-66
Pathmaker, using to develop 1-14 overview 3-3
single-threaded and multithreaded 2-20 syntax and usage 5-21/5-22
structure of 2-25/2-27 Servers
TMF subsystem and fault See Server classes; Server processes;
tolerance 2-23 Server programs; Applications,
utilization of 2-20 programming; Applications, designing
Server reply code canceling sends to server classes 3-10
for RSC clients 4-4 context-sensitive
for SCREEN COBOL requesters 4-3 errors returned 4-14
SERVERCLASS_DIALOG_ABORT_ functions of 4-13
procedure context-sensitive requesters, using 3-8,
overview 3-8 4-4
syntax and usage 5-6 conversational (context-sensitive) 4-1
SERVERCLASS_DIALOG_BEGIN_ creation failure (error 916) 6-8
procedure debugging 4-11/4-12
canceling 5-25/5-26 failures and Pathsend 3-4, 4-2
flag parameter 3-9 interoperation with TUXEDO
overview 3-8 requesters 4-17
syntax and usage 5-7/5-11 link error (error 904) 6-4
TMF considerations 5-24 nested 4-4
SERVERCLASS_DIALOG_END_ no link available (error 905) 6-5
procedure NonStop TUXEDO 3-11
overview 3-8 PWY2TUX translation server 3-11
syntax and usage 5-12 request/response (context-free) 4-1
SERVERCLASS_DIALOG_SEND_ timeouts, specifying 5-27
procedure
too many links (error 923) 6-10
canceling 5-25/5-26
too many server classes (error 922) 6-10
syntax and usage 5-13/5-16
TMF considerations 5-24

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -12
Index T

Servers, writing Subsystem Control Facility (SCF), use in

context-free with context-sensitive managing GDSX processes 2-18
requesters 4-4 Subsystem Programmatic Interface (SPI)
context-sensitive See SPI (Subsystem Programmatic
programming 4-13/4-17 Interface)
debugging 4-11/4-12 Synchronization IDs 3-6
Guardian 4-2, 4-3 Syntax
linkage space considerations 4-3 See individual procedures
nested servers 4-4 System /T clients, NonStop TUXEDO,
overview 4-1 interoperating with 4-17
Pathway 4-2 Systems
reply formats 4-1 expanding 1-7
RSC requester considerations 4-4 NonStop 1-5/1-7
SCREEN COBOL requester security 1-6
considerations 4-3
sharing by different requesters 4-1 T
TMF subsystem Tables, NonStop SQL/MP 2-10
application structure 4-5/4-6 TAL
audited files 4-7 example Pathsend requester
fault-tolerant programming 4-10 program B-1/B-52
grouping transaction for Pathsend requesters 1-10
operations 4-8/4-9 for Pathway servers 1-8
overview 4-5 invoking Pathsend procedures 5-5
record locking 4-8 Tandem
restrictions 4-7 Alliance 1-4
transaction deadlocks 4-10 computing fundamentals 1-3/1-7
Server-class name invalid (error 900) 6-2 TCP (terminal control process)
Server-to-server communication fault-tolerance role 1-6
See Nested servers features provided by 1-10
Single-threading in server design 2-20, 2-23 IDS requesters 2-12
Software development tools 1-14/1-15 RSC requesters 2-15
Special registers, PROGRAM- TDP (Transaction Delivery Process ) 2-15
STATUS 4-10
TEDIT text editor 1-14
SPI (Subsystem Programmatic Interface)
Terminals 2-12
description 1-4
TERMINATION-STATUS and -
use in managing GDSX processes 2-18 SUBSTATUS 4-11/4-12
use in managing servers 1-8, 2-19 Testing and standardizing code 1-4
Standardizing and testing code 1-3 Third-party vendors 1-4
Static links 3-4 Throughput, system 1-6

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -13
Index U

timelimit parameter, AWAITIOX Transactions (continued)

procedure 5-28 concurrency control 2-7/2-8
TIMEOUT attribute 5-27 cost per 1-6
Timeout error (error 40) 4-11/4-12 database consistency and
Timeout errors 4-11/4-12 concurrency 4-5
Timeouts deadlocks 4-10
invalid value (error 919) 6-9 defining for the TMF subsystem 2-6
Pathsend usage considerations 2-14, designing an application with 2-1/2-8
5-27/5-28 example abort B-15
send completion, specifying grouping operations 4-8/4-9
maximum 5-9, 5-15, 5-19
identifier 2-6
servers, specifying 5-27
identifier changed (error 930) 6-13
server-classes, specifying 5-27
identifying components 2-4/2-5
TMF OFF server parameter 3-5, 5-24
integrity 1-5
TMF (Transaction Management Facility)
protecting 2-6/2-8
audit-trail files 2-8
start with TMF, example B-13
defining transactions 2-6
TSCODE, GDSX process 2-16
description 1-4
TUX2PWY translation server 1-13, 4-17
fault-tolerance role 1-6
TUXEDO
Too many dialogs (error 927) 6-12
See NonStop TUXEDO
Too many Pathmons (error 921) 6-9
TUXEDO requesters (clients)
Too many requesters (error 925) 6-11
writing Pathway servers for 1-13, 4-17
Too many send requests (error 924) 6-11
TUXEDO to Pathway translation
Too many server classes (error 922) 6-10 server 1-13, 4-17
Too many server links (error 923) 6-10
Tools, software development 1-14/1-15 U
tpinit() ATMI function 3-11
Unknown server class (error 914) 6-7
Transaction Application Language
Unsolicited message processing
See TAL (UMP) 2-12
Transaction identifiers and LINKMON Unstructured files 2-10
processes 5-24
USCODE, GDSX process 2-16
Transaction Management Facility
Utilities, software development 1-14/1-15
See TMF
Utilization of resources 3-8
Transaction processing
See OLTP
Transaction processing scenario 1-15/1-17
V
Transactions Vendors, third-party 1-4
aborted (error 934) 6-14
aborting 2-8, 2-23
backout 2-8

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -14
Index W

W
Waited send operations
errors for 6-1
procedure calls for 5-7, 5-13, 5-17
specifying 5-9, 5-15, 5-20
usage considerations 5-23, 5-27
Workstation clients, NonStop TUXEDO,
interoperating with 4-17

Special Character
$RECEIVE messages
allocating space 4-3
determining new dialogs 4-13

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -15
Index Special Character

NonStop TS/MP Pathsend and Server Programming Manual– 132500

Index -16