Streamlining the imports #3

fariza · 2013-12-19T01:14:04Z

This came after reading trhought the code trying to understand the flow of things.

Please correct me if I'm wrong

Figuremanager and FigureCanvas

If I understood correctly Figuremanager and FigureCanvas should be reserved for ready to use classes

I removed Figuremanager and FigureCanvas from

_backend_gtk3.py
backend_gtk3.py
_backend_qt4.py
backend_qt4.py

Those classes are not usable by as they are they have to be used with Agg or Cairo,.

Callbacks setting

There is no need to duplicate the setting of callbacks, it should remain in backend_gtk3.py and backend_qt4.py.
The Agg or Cairo variants don't need to redefine this.

This commit refactors backend_bases into two parts, the Gcf independent parts, which moved to backends/_backend_bases.py, and the Gcf dependent stuff which stayed in backend_bases.py. The contents of backends/_backend_bases.py are bulk imported into backend_bases so this change should be backwards compatible. Added an attribute to FigureManagerBase for the _key_press_handler function. This might be a stop-gap if the logic in `Gcf.destroy` should be moved into the FigureManager or canvas classes.

added a static method to FigureManagerQt to gracefully patch in the Gcf.destroy call back. As with the _key_press_handler call back this may be a stop-gap if that logic gets re-factored.

One part (_backend_qt4Agg.py) has no dependencies on Gcf or pyplot, which are all collected in backend_qt4agg.py. Should be fully back-wards compatible.

get pop-up graph windows as part of a larger gui application.

Two line-length violations on long urls in comments are left.

changed imports to _backend_bases to avoid implicit Gcf import

one which includes Gcf and one that does not.

_backend_gtkagg.py has no dependence on `Gcf`, backend_gtkagg.py depends on `Gcf` and maintains reverse compatibility.

Same long url as backend_gkt.py is the lone violation

_backend_gtk3.py does not depend on `Gcf`. backend_gkt3.py does depend on `Gcf` and maintains reverse compatibility

parts.

gcf independent parts. Same poor naming scheme as everything else in this branch.

Gcf

and one not. Same naming scheme as rest.

updated example to reflect that the class the class `FigureManagerQTAgg` has been removed.

Added `FigureCanvas` and `FigureManager` aliases to `_backend_qt4agg.py` and `_backend_qt4.py`.

adjusting imports

tacaswell · 2013-12-19T05:50:50Z

The problem with this is that we need to maintain reverse compatibility. I am pretty sure that the existing top level modules export FigureManager so, for now, we need to keep exporting them.

You have a fair point about the call backs design-wise, however one of the goals here is to break nothing. The callbacks used to be in the baseclases in backend_bases so anything that derives from them and ends up public-facing needs to also include them. We don't use those classes anywhere else, but that does not mean no one else is.

I may be taking no reverse incompatible changes a little too seriously. I would like feed back from @mdboom and company on this.

mdboom · 2013-12-19T13:46:51Z

-1 on removing FigureManager from the namespaces

+1 on not redefining the callbacks in *Agg and *Cairo, which should already be there from the base classes.

tacaswell · 2013-12-19T20:38:00Z

Right, I didn't follow through the inheritance/had them siloed wrong in my head last night.

The same thing should be done for backend_gtk*.

@fariza Let me know if you want to make the changes. I want to get this merged before I do the rebase (which I am not looking forward to).

fariza · 2013-12-19T23:59:46Z

The FigureManager and FigureCanvas were added only a month ago matplotlib@8159c81

@mdboom Why do you want to expose these two "names" in classes that can not be used directly?
It makes more sense to reserve those for "ready-to-use" classes doesn't it?

@tacaswell I think the best would be just to keep modifing two backends qt4 and Gtk3 until we agree on the way to do it (on the fly assemby by pyplot, or two files per backend, or ....) then we can modify all the rest.
It doesn't make sense to keep modifing more than two just for testing pourposes. If you insist, I will wait for feedback on the previous paragraph and then I will modify it.

tacaswell · 2013-12-20T06:18:24Z

@fariza Fair enough on just doing two at first.

tacaswell · 2013-12-23T06:50:46Z

I am coming around to your suggestion of only exporting FigureCanvas in the case where the canvas being exported is fully useful.

I had not realized how new that naming scheme was.

@pwuertz Can you comment on this?

pwuertz · 2013-12-23T07:49:00Z

@tacaswell When switching to a different backend for saving figures, FigureCanvasBase.switch_backends() requires a reference to the new FigureCanvas class. Each known file type XY had its own print_XY method for a hardcoded import of FigureCanvasXY from module backendXY.

After matplotlib/matplotlib@8159c81 there is just a mapping from file types to backend modules, which can be modified and extended at runtime. You still need to know the name of a module's FigureCanvasClass though when doing the import. Instead of mapping that name as well I introduced the canonical name FigureCanvas - which basicly adds polymorphism to the backend module system and simplifies the whole process.

At the moment it is safe to omit this alias if you are not planning to use that backend for file saving.

Pdf xobject linewidth

This fixes some possible heap buffer overflows, such as the following triggered by our cmmi10.ttf: ``` ERROR: AddressSanitizer: heap-buffer-overflow on address 0x617000235709 at pc 0x7f95efd3c48a bp 0x7ffe41b6ecc0 sp 0x7ffe41b6ecb0 READ of size 1 at 0x617000235709 thread T0 #0 0x7f95efd3c489 in utf16be_to_ascii extern/ttconv/pprdrv_tt.cpp:178 #1 0x7f95efd3c489 in Read_name(TTFONT*) extern/ttconv/pprdrv_tt.cpp:339 #2 0x7f95efd499ef in read_font(...) extern/ttconv/pprdrv_tt.cpp:1325 #3 0x7f95efd4c602 in get_pdf_charprocs(...) extern/ttconv/pprdrv_tt.cpp:1420 #4 0x7f95efd35c22 in py_get_pdf_charprocs src/_ttconv.cpp:217 0x617000235709 is located 1 bytes to the right of 648-byte region [0x617000235480,0x617000235708) allocated by thread T0 here: #0 0x7f9612262a38 in __interceptor_calloc (/usr/lib64/libasan.so.4+0xdea38) #1 0x7f95efd3b261 in GetTable(TTFONT*, char const*) extern/ttconv/pprdrv_tt.cpp:140 ```

commit a05a486 Merge: 6ec4f8f e751286 Author: Ida Hjorth <idahjorth@gmail.com> Date: Sun Apr 29 19:07:32 2018 +0200 Merge pull request #3 from magnunor/idahj_anchored_directionarrows Idahj anchored directionarrows commit e751286 Merge: 3ba79fb 6ec4f8f Author: Magnus Nord <magnunor@gmail.com> Date: Sun Apr 29 09:00:34 2018 +0200 Merge branch 'NEW_AnchoredDirectionArrows' of github.com:idahj/matplotlib into idahj_anchored_directionarrows commit 6ec4f8f Author: Ida <idahjorth@gmail.com> Date: Fri Aug 25 13:14:26 2017 +0100 AnchoredDirectionArrows: Docstring, what's new commit 19f4a38 Author: Ida <idahjorth@gmail.com> Date: Thu Aug 24 14:27:23 2017 +0100 AnchoredDirectionArrows: Demo docstring commit d024684 Author: Ida <idahjorth@gmail.com> Date: Thu Aug 24 14:09:52 2017 +0100 AnchoredDirectionArrows: Fix error commit 27b6bc5 Author: Ida <idahjorth@gmail.com> Date: Thu Aug 24 13:58:58 2017 +0100 AnchoredDirectionArrows: make variable not private commit c0762d2 Author: Ida <idahjorth@gmail.com> Date: Thu Aug 24 13:02:38 2017 +0100 AnchoredDirectionArrows: Add docstring in demo commit 88d04d0 Author: Ida <idahjorth@gmail.com> Date: Wed Aug 23 14:19:06 2017 +0100 AnchoredDirectionArrows: docstring, what's new commit efcca59 Author: Magnus Nord <Magnus.Nord@glasgow.ac.uk> Date: Wed Aug 23 13:31:32 2017 +0100 AnchoredDirectionArrows: add test which uses many of the arguments commit e200c44 Author: Magnus Nord <Magnus.Nord@glasgow.ac.uk> Date: Wed Aug 23 12:06:05 2017 +0100 AnchoredDirectionArrows: minor docstring changes commit b6b0ea2 Author: Ida <idahjorth@gmail.com> Date: Tue Aug 22 13:28:56 2017 +0100 AnchoredDirectionArrows: PEP8 commit 5fde65a Author: Ida <idahjorth@gmail.com> Date: Tue Aug 22 13:23:25 2017 +0100 AnchoredDirectionArrows: Add demo commit 23f7d5b Author: Ida <idahjorth@gmail.com> Date: Tue Aug 22 11:16:41 2017 +0100 AnchoredDirectionArrows: Add working unit test commit 459ae07 Author: Ida <idahjorth@gmail.com> Date: Tue Aug 22 10:37:44 2017 +0100 AnchoredDirectionArrow: first attempt unit test Currently not working commit f4ec0e2 Author: Ida <idahjorth@gmail.com> Date: Wed Aug 16 19:31:05 2017 +0100 Cleanup, bug fix, begin merge overlapping arrows commit 75e52b0 Author: Magnus Nord <magnunor@gmail.com> Date: Mon Aug 14 23:05:30 2017 +0100 AnchoredDirectionArrows: PEP8 fixes, and improve docstring commit 4fff775 Author: Ida <idahjorth@gmail.com> Date: Mon Aug 14 22:04:22 2017 +0100 Add AnchoredDirectionArrows class in axes_grid1

Update backend_svg.py

The attempt to simplify the stride calculation in `a06f343dee3cebf035b74f65ea00b8842af448e9` seems to be the cause of the problem. Using an approach equivalent to what was there before hopefully resolves matplotlib#24092.

tacaswell and others added 30 commits December 3, 2013 22:02

split backend_qt4 into two parts, with and without Gcf

0b60477

added a static method to FigureManagerQt to gracefully patch in the Gcf.destroy call back. As with the _key_press_handler call back this may be a stop-gap if that logic gets re-factored.

split backend_qt4agg into two parts.

7f1247d

One part (_backend_qt4Agg.py) has no dependencies on Gcf or pyplot, which are all collected in backend_qt4agg.py. Should be fully back-wards compatible.

Added a demo-file to show how to use the FigureManager classes to

70b5af1

get pop-up graph windows as part of a larger gui application.

removed un-needed import of Gcf

b682427

pep8 on backend_gtk.py

a070748

Two line-length violations on long urls in comments are left.

pep8 clean up in backend_gdk

14cd596

removed un-needed Gcf import

8ae109b

changed imports to _backend_bases to avoid implicit Gcf import

split backend_gcf into two parts,

09f1ce8

one which includes Gcf and one that does not.

pep8 on backend_gtkagg.py

fd33282

split backend_gktagg.py in to two parts

1451226

_backend_gtkagg.py has no dependence on `Gcf`, backend_gtkagg.py depends on `Gcf` and maintains reverse compatibility.

updated exclude list

2ba6fa4

pep8 clean up on backend_gtk3.py

c5b7c29

Same long url as backend_gkt.py is the lone violation

split backend_gtk3 into two parts

467f54d

_backend_gtk3.py does not depend on `Gcf`. backend_gkt3.py does depend on `Gcf` and maintains reverse compatibility

updated coding standards

2ab49f4

pep8 on backend_gtkcairo.py

93040ad

split backend_gtkcairo.py into Gcf dependent and independent

6173745

parts.

import from _backend_bases in backend_agg to avoid implicit Gcf import

2563216

pep8 on backend_agg.py

58badfd

split backend_gtk3agg.py into two parts, the gcf dependent and

6927e3c

gcf independent parts. Same poor naming scheme as everything else in this branch.

updated coding standards exclude list

854c890

pep8 on backend_gtk3cairo.py

c065a40

pep8 on backend_cairo.py

da76714

changed backend_cairo.py to import from _backend_bases to avoid

8c68cd5

Gcf

updated coding standards

28f37cb 8000

Split backend_gtk3cairo.py into two parts, one dependent on Gcf

59c8bb3

and one not. Same naming scheme as rest.

Updated embedding_with_qt4_manager.py

7993710

updated example to reflect that the class the class `FigureManagerQTAgg` has been removed.

Improved class-aliases in _backend_qt*.py

146a6eb

Added `FigureCanvas` and `FigureManager` aliases to `_backend_qt4agg.py` and `_backend_qt4.py`.

adjusting imports

b34d58a

Merge pull request #2 from fariza/qt4_example_correction

4f5187f

adjusting imports

Streamlining the imports

1efef7a

fariza closed this Feb 6, 2014

fariza deleted the imports_cleanup branch February 6, 2014 19:02

tacaswell pushed a commit that referenced this pull request Oct 19, 2014

Merge pull request #3 from mdboom/pdf-xobject-linewidth

878b681

Pdf xobject linewidth

tacaswell pushed a commit that referenced this pull request May 21, 2020

Merge pull request #3 from cover-me/fix-pep8

932a863

Update backend_svg.py

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Streamlining the imports #3

Streamlining the imports #3

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Streamlining the imports #3

Streamlining the imports #3

Uh oh!

Conversation

Figuremanager and FigureCanvas

Callbacks setting

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!