8000 Makes Range#sum an O(1) operation instead of an O(n) one. by soc · Pull Request #16 · scala/scala · GitHub
[go: up one dir, main page]

Skip to content

Makes Range#sum an O(1) operation instead of an O(n) one. #16

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Dec 2, 2011
Merged

Makes Range#sum an O(1) operation instead of an O(n) one. #16

merged 1 commit into from
Dec 2, 2011

Conversation

soc
Copy link
Contributor
@soc soc commented Dec 2, 2011

Partially fixes SI-4658. NumericRange stays slow, thanks to the brilliant idea that Numeric doesn't need a division operation.

Some notes:

  • I did use if instead of match because I was not sure if there is a difference in performance especially with the work on the new pattern matcher.
  • I used (len.toLong * (head + last) / 2) instead of (len.toLong * (head.toLong + last) / 2) because I couldn't find a situation in which working with a Long here avoided an overflow error in the result. (There are no inputs where head + last would overflow but the result wouldn't.) The first one is more or less "overflow error compatible" to the slow O(n) way, while the second one produces different overflow errors. The only case where it avoided an overflow error was in Ranges with a single element, but the code before catches this, thanks to notnoop spotting it (Makes Range#sum an O(1) operation instead of an O(n) one. legacy-svn-scala#83).

Partially fixes SI-4658. NumericRange stays slow, thanks to the brilliant idea that Numeric doesn't need a division operation.
@paulp paulp merged commit db7bd31 into scala:master Dec 2, 2011
gkossakowski added a commit to gkossakowski/scala that referenced this pull request Jan 11, 2012
Factory manifests compiler plug-in.
heathermiller pushed a commit to heathermiller/scala that referenced this pull request Oct 11, 2012
retronym referenced this pull request in retronym/scala Nov 6, 2014
In Scala 2.8.2, an optimization was added to create a static
cache for Symbol literals (ie, the results of `Symbol.apply("foo"))`.
This saves the map lookup on the second pass through code.

This actually was broken somewhere during the Scala 2.10 series,
after the addition of an overloaded `apply` method to `Symbol`.

The cache synthesis code was made aware of the overload and brought
back to working condition recently, in scala#3149.

However, this has uncovered a latent bug when the Symbol literals are
defined with traits.

One of the enclosed tests failed with:

	  jvm > t8933b-run.log
	java.lang.IllegalAccessError: tried to access field MotherClass.symbol$1 from class MixinWithSymbol$class
	        at MixinWithSymbol$class.symbolFromTrait(A.scala:3)
	        at MotherClass.symbolFromTrait(Test.scala:1)

This commit simply disables the optimization if we are in a trait.
Alternative fixes might be: a) make the static Symbol cache field
public / b) "mixin" the static symbol cache. Neither of these
seem worth the effort and risk for an already fairly situational
optimization.

Here's how the optimization looks in a class:

	% cat sandbox/test.scala; qscalac sandbox/test.scala && echo ":javap C" | qscala;
	class C {
	  'a; 'b
	}
	Welcome to Scala version 2.11.5-20141106-145558-aa558dce6d (Java HotSpot(TM) 64-Bit Server VM, Java 1.8.0_20).
	Type in expressions to have them evaluated.
	Type :help for more information.

	scala> :javap C
	  Size 722 bytes
	  MD5 checksum 6bb00189166917254e8d40499ee7c887
	  Compiled from "test.scala"
	public class C

	{
	  public static {};
	    descriptor: ()V
	    flags: ACC_PUBLIC, ACC_STATIC
	    Code:
	      stack=2, locals=0, args_size=0
	         0: getstatic     #16                 // Field scala/Symbol$.MODULE$:Lscala/Symbol$;
	         3: ldc           #18                 // String a
	         5: invokevirtual #22                 // Method scala/Symbol$.apply:(Ljava/lang/String;)Lscala/Symbol;
	         8: putstatic     #26                 // Field symbol$1:Lscala/Symbol;
	        11: getstatic     #16                 // Field scala/Symbol$.MODULE$:Lscala/Symbol$;
	        14: ldc           #28                 // String b
	        16: invokevirtual #22                 // Method scala/Symbol$.apply:(Ljava/lang/String;)Lscala/Symbol;
	        19: putstatic     #31                 // Field symbol$2:Lscala/Symbol;
	        22: return

	  public C();
	    descriptor: ()V
	    flags: ACC_PUBLIC
	    Code:
	      stack=1, locals=1, args_size=1
	         0: aload_0
	         1: invokespecial #34                 // Method java/lang/Object."<init>":()V
	         4: getstatic     #26                 // Field symbol$1:Lscala/Symbol;
	         7: pop
	         8: getstatic     #31                 // Field symbol$2:Lscala/Symbol;
	        11: pop
	        12: return
	}
retronym referenced this pull request in retronym/scala Nov 6, 2014
In Scala 2.8.2, an optimization was added to create a static
cache for Symbol literals (ie, the results of `Symbol.apply("foo"))`.
This saves the map lookup on the second pass through code.

This actually was broken somewhere during the Scala 2.10 series,
after the addition of an overloaded `apply` method to `Symbol`.

The cache synthesis code was made aware of the overload and brought
back to working condition recently, in scala#3419.

However, this has uncovered a latent bug when the Symbol literals are
defined with traits.

One of the enclosed tests failed with:

	  jvm > t8933b-run.log
	java.lang.IllegalAccessError: tried to access field MotherClass.symbol$1 from class MixinWithSymbol$class
	        at MixinWithSymbol$class.symbolFromTrait(A.scala:3)
	        at MotherClass.symbolFromTrait(Test.scala:1)

This commit simply disables the optimization if we are in a trait.
Alternative fixes might be: a) make the static Symbol cache field
public / b) "mixin" the static symbol cache. Neither of these
seem worth the effort and risk for an already fairly situational
optimization.

Here's how the optimization looks in a class:

	% cat sandbox/test.scala; qscalac sandbox/test.scala && echo ":javap C" | qscala;
	class C {
	  'a; 'b
	}
	Welcome to Scala version 2.11.5-20141106-145558-aa558dce6d (Java HotSpot(TM) 64-Bit Server VM, Java 1.8.0_20).
	Type in expressions to have them evaluated.
	Type :help for more information.

	scala> :javap C
	  Size 722 bytes
	  MD5 checksum 6bb00189166917254e8d40499ee7c887
	  Compiled from "test.scala"
	public class C

	{
	  public static {};
	    descriptor: ()V
	    flags: ACC_PUBLIC, ACC_STATIC
	    Code:
	      stack=2, locals=0, args_size=0
	         0: getstatic     #16                 // Field scala/Symbol$.MODULE$:Lscala/Symbol$;
	         3: ldc           #18                 // String a
	         5: invokevirtual #22                 // Method scala/Symbol$.apply:(Ljava/lang/String;)Lscala/Symbol;
	         8: putstatic     #26                 // Field symbol$1:Lscala/Symbol;
	        11: getstatic     #16                 // Field scala/Symbol$.MODULE$:Lscala/Symbol$;
	        14: ldc           #28                 // String b
	        16: invokevirtual #22                 // Method scala/Symbol$.apply:(Ljava/lang/String;)Lscala/Symbol;
	        19: putstatic     #31                 // Field symbol$2:Lscala/Symbol;
	        22: return

	  public C();
	    descriptor: ()V
	    flags: ACC_PUBLIC
	    Code:
	      stack=1, locals=1, args_size=1
	         0: aload_0
	         1: invokespecial #34                 // Method java/lang/Object."<init>":()V
	         4: getstatic     #26                 // Field symbol$1:Lscala/Symbol;
	         7: pop
	         8: getstatic     #31                 // Field symbol$2:Lscala/Symbol;
	        11: pop
	        12: return
	}

fixup
retronym referenced this pull request in retronym/scala Nov 7, 2014
In Scala 2.8.2, an optimization was added to create a static
cache for Symbol literals (ie, the results of `Symbol.apply("foo"))`.
This saves the map lookup on the second pass through code.

This actually was broken somewhere during the Scala 2.10 series,
after the addition of an overloaded `apply` method to `Symbol`.

The cache synthesis code was made aware of the overload and brought
back to working condition recently, in scala#3149.

However, this has uncovered a latent bug when the Symbol literals are
defined with traits.

One of the enclosed tests failed with:

	  jvm > t8933b-run.log
	java.lang.IllegalAccessError: tried to access field MotherClass.symbol$1 from class MixinWithSymbol$class
	        at MixinWithSymbol$class.symbolFromTrait(A.scala:3)
	        at MotherClass.symbolFromTrait(Test.scala:1)

This commit simply disables the optimization if we are in a trait.
Alternative fixes might be: a) make the static Symbol cache field
public / b) "mixin" the static symbol cache. Neither of these
seem worth the effort and risk for an already fairly situational
optimization.

Here's how the optimization looks in a class:

	% cat sandbox/test.scala; qscalac sandbox/test.scala && echo ":javap C" | qscala;
	class C {
	  'a; 'b
	}
	Welcome to Scala version 2.11.5-20141106-145558-aa558dce6d (Java HotSpot(TM) 64-Bit Server VM, Java 1.8.0_20).
	Type in expressions to have them evaluated.
	Type :help for more information.

	scala> :javap C
	  Size 722 bytes
	  MD5 checksum 6bb00189166917254e8d40499ee7c887
	  Compiled from "test.scala"
	public class C

	{
	  public static {};
	    descriptor: ()V
	    flags: ACC_PUBLIC, ACC_STATIC
	    Code:
	      stack=2, locals=0, args_size=0
	         0: getstatic     #16                 // Field scala/Symbol$.MODULE$:Lscala/Symbol$;
	         3: ldc           #18                 // String a
	         5: invokevirtual #22                 // Method scala/Symbol$.apply:(Ljava/lang/String;)Lscala/Symbol;
	         8: putstatic     #26                 // Field symbol$1:Lscala/Symbol;
	        11: getstatic     #16                 // Field scala/Symbol$.MODULE$:Lscala/Symbol$;
	        14: ldc           #28                 // String b
	        16: invokevirtual #22                 // Method scala/Symbol$.apply:(Ljava/lang/String;)Lscala/Symbol;
	        19: putstatic     #31                 // Field symbol$2:Lscala/Symbol;
	        22: return

	  public C();
	    descriptor: ()V
	    flags: ACC_PUBLIC
	    Code:
	      stack=1, locals=1, args_size=1
	         0: aload_0
	         1: invokespecial #34                 // Method java/lang/Object."<init>":()V
	         4: getstatic     #26                 // Field symbol$1:Lscala/Symbol;
	         7: pop
	         8: getstatic     #31                 // Field symbol$2:Lscala/Symbol;
	        11: pop
	        12: return
	}

fixup
retronym referenced this pull request in retronym/scala Jul 14, 2015
And do the same for the specialized variants.

Tested by a Java source file that uses lambda syntax to create
instances of generic and specialized `Function{0,1}`.

Here's how the interfaces look now:

```
% javap -c -classpath /tmp/function 'scala.Function1'
Compiled from "Function1.scala"
public interface scala.Function1<T1, R> {
  public abstract R apply(T1);

  public <A> scala.Function1<A, R> compose(scala.Function1<A, T1>);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  #18                 // Method scala/Function1$class.compose:(Lscala/Function1;Lscala/Function1;)Lscala/Function1;
       5: areturn

  public <A> scala.Function1<T1, A> andThen(scala.Function1<R, A>);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  #24                 // Method scala/Function1$class.andThen:(Lscala/Function1;Lscala/Function1;)Lscala/Function1;
       5: areturn

  public abstract java.lang.String toString();

  public int apply$mcII$sp(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokestatic  #110                // Method scala/Function1$class.apply$mcII$sp:(Lscala/Function1;I)I
       5: ireturn

  public long apply$mcJI$sp(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokestatic  #115                // Method scala/Function1$class.apply$mcJI$sp:(Lscala/Function1;I)J
       5: lreturn
    ...
}

% javap -c -classpath /tmp/function 'scala.Function1$mcII$sp'
Compiled from "Function1.scala"
public interface scala.Function1$mcII$sp extends scala.Function1<java.lang.Object, java.lang.Object> {
  public java.lang.Object apply(java.lang.Object);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  #16                 // Method scala/runtime/BoxesRunTime.unboxToInt:(Ljava/lang/Object;)I
       5: invokeinterface #19,  2           // InterfaceMethod apply:(I)I
      10: invokestatic  #23                 // Method scala/runtime/BoxesRunTime.boxToInteger:(I)Ljava/lang/Integer;
      13: areturn

  public abstract int apply$mcII$sp(int);

  public int apply(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokeinterface #30,  2           // InterfaceMethod apply$mcII$sp:(I)I
       7: ireturn
}
```
retronym referenced this pull request in retronym/scala Jul 23, 2015
And do the same for the specialized variants.

Tested by a Java source file that uses lambda syntax to create
instances of generic and specialized `Function{0,1}`.

Here's how the interfaces look now:

```
% javap -c -classpath /tmp/function 'scala.Function1'
Compiled from "Function1.scala"
public interface scala.Function1<T1, R> {
  public abstract R apply(T1);

  public <A> scala.Function1<A, R> compose(scala.Function1<A, T1>);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  #18                 // Method scala/Function1$class.compose:(Lscala/Function1;Lscala/Function1;)Lscala/Function1;
       5: areturn

  public <A> scala.Function1<T1, A> andThen(scala.Function1<R, A>);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  #24                 // Method scala/Function1$class.andThen:(Lscala/Function1;Lscala/Function1;)Lscala/Function1;
       5: areturn

  public abstract java.lang.String toString();

  public int apply$mcII$sp(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokestatic  #110                // Method scala/Function1$class.apply$mcII$sp:(Lscala/Function1;I)I
       5: ireturn

  public long apply$mcJI$sp(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokestatic  #115                // Method scala/Function1$class.apply$mcJI$sp:(Lscala/Function1;I)J
       5: lreturn
    ...
}

% javap -c -classpath /tmp/function 'scala.Function1$mcII$sp'
Compiled from "Function1.scala"
public interface scala.Function1$mcII$sp extends scala.Function1<java.lang.Object, java.lang.Object> {
  public java.lang.Object apply(java.lang.Object);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  #16                 // Method scala/runtime/BoxesRunTime.unboxToInt:(Ljava/lang/Object;)I
       5: invokeinterface #19,  2           // InterfaceMethod apply:(I)I
      10: invokestatic  #23                 // Method scala/runtime/BoxesRunTime.boxToInteger:(I)Ljava/lang/Integer;
      13: areturn

  public abstract int apply$mcII$sp(int);

  public int apply(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokeinterface #30,  2           // InterfaceMethod apply$mcII$sp:(I)I
       7: ireturn
}
```
SethTisue pushed a commit to SethTisue/scala that referenced this pull request Aug 7, 2015
And do the same for the specialized variants.

Tested by a Java source file that uses lambda syntax to create
instances of generic and specialized `Function{0,1}`.

Here's how the interfaces look now:

```
% javap -c -classpath /tmp/function 'scala.Function1'
Compiled from "Function1.scala"
public interface scala.Function1<T1, R> {
  public abstract R apply(T1);

  public <A> scala.Function1<A, R> compose(scala.Function1<A, T1>);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  scala#18                 // Method scala/Function1$class.compose:(Lscala/Function1;Lscala/Function1;)Lscala/Function1;
       5: areturn

  public <A> scala.Function1<T1, A> andThen(scala.Function1<R, A>);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  scala#24                 // Method scala/Function1$class.andThen:(Lscala/Function1;Lscala/Function1;)Lscala/Function1;
       5: areturn

  public abstract java.lang.String toString();

  public int apply$mcII$sp(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokestatic  scala#110                // Method scala/Function1$class.apply$mcII$sp:(Lscala/Function1;I)I
       5: ireturn

  public long apply$mcJI$sp(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokestatic  scala#115                // Method scala/Function1$class.apply$mcJI$sp:(Lscala/Function1;I)J
       5: lreturn
    ...
}

% javap -c -classpath /tmp/function 'scala.Function1$mcII$sp'
Compiled from "Function1.scala"
public interface scala.Function1$mcII$sp extends scala.Function1<java.lang.Object, java.lang.Object> {
  public java.lang.Object apply(java.lang.Object);
    Code:
       0: aload_0
       1: aload_1
       2: invokestatic  scala#16                 // Method scala/runtime/BoxesRunTime.unboxToInt:(Ljava/lang/Object;)I
       5: invokeinterface scala#19,  2           // InterfaceMethod apply:(I)I
      10: invokestatic  scala#23                 // Method scala/runtime/BoxesRunTime.boxToInteger:(I)Ljava/lang/Integer;
      13: areturn

  public abstract int apply$mcII$sp(int);

  public int apply(int);
    Code:
       0: aload_0
       1: iload_1
       2: invokeinterface scala#30,  2           // InterfaceMethod apply$mcII$sp:(I)I
       7: ireturn
}
```
adriaanm pushed a commit that referenced this pull request Mar 31, 2016
Rather than in implementation of the abstract method in the
expanded anonymous class.

This leads to more more efficient use of the constant pool,
code shapes more amenable to SAM inlining, and is compatible
with the old behaviour of `-Xexperimental` in Scala 2.11,
which ScalaJS now relies upon.

Manual test:

```
scala> :paste -raw
// Entering paste mode (ctrl-D to finish)

package p1; trait T { val x = 0; def apply(): Any }; class DelambdafyInline { def t: T = (() => "") }

// Exiting paste mode, now interpreting.

scala> :javap -c p1.DelambdafyInline
Compiled from "<pastie>"
public class p1.DelambdafyInline {
  public p1.T t();
    Code:
       0: new           #10                 // class p1/DelambdafyInline$$anonfun$t$1
       3: dup
       4: aload_0
       5: invokespecial #16                 // Method p1/DelambdafyInline$$anonfun$t$1."<init>":(Lp1/DelambdafyInline;)V
       8: areturn

  public final java.lang.Object p1$DelambdafyInline$$$anonfun$1();
    Code:
       0: ldc           #22                 // String
       2: areturn

  public p1.DelambdafyInline();
    Code:
       0: aload_0
       1: invokespecial #25                 // Method java/lang/Object."<init>":()V
       4: return
}

scala> :javap -c p1.DelambdafyInline$$anonfun$t$1
Compiled from "<pastie>"
public final class p1.DelambdafyInline$$anonfun$t$1 implements p1.T,scala.Serializable {
  public static final long serialVersionUID;

  public int x();
    Code:
       0: aload_0
       1: getfield      #25                 // Field x:I
       4: ireturn

  public void p1$T$_setter_$x_$eq(int);
    Code:
       0: aload_0
       1: iload_1
       2: putfield      #25                 // Field x:I
       5: return

  public final java.lang.Object apply();
    Code:
       0: aload_0
       1: getfield      #34                 // Field $outer:Lp1/DelambdafyInline;
       4: invokevirtual #37                 // Method p1/DelambdafyInline.p1$DelambdafyInline$$$anonfun$1:()Ljava/lang/Object;
       7: areturn

  public p1.DelambdafyInline$$anonfun$t$1(p1.DelambdafyInline);
    Code:
       0: aload_1
       1: ifnonnull     6
       4: aconst_null
       5: athrow
       6: aload_0
       7: aload_1
       8: putfield      #34                 // Field $outer:Lp1/DelambdafyInline;
      11: aload_0
      12: invokespecial #42                 // Method java/lang/Object."<init>":()V
      15: aload_0
      16: invokespecial #45                 // Method p1/T.$init$:()V
      19: return
}

scala> :quit
```

Adriaan is to `git blame` for `reflection-mem-typecheck.scala`.
OlivierBlanvillain pushed a commit to OlivierBlanvillain/scala that referenced this pull request Jun 27, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants
0