Bug 154927 – RegExpPrototype should check for exceptions after calling toString and doing so should not be expensive

• Component: JavaScriptCore
• Status: RESOLVED FIXED

[Filip Pizlo]

Patch forthcoming.

[Filip Pizlo]

Created attachment 272673 [details]
the patch

[Saam Barati]

Comment on attachment 272673 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=272673&action=review

> Source/JavaScriptCore/runtime/JSCJSValue.cpp:391
>      if (exec->hadException())

this can be "vm.exception()"

> Source/JavaScriptCore/runtime/JSCJSValue.cpp:395
> +    if (exec->hadException())

this can be "vm.exception()"

[Filip Pizlo]

(In reply to comment #2)
> Comment on attachment 272673 [details]
> the patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=272673&action=review
> 
> > Source/JavaScriptCore/runtime/JSCJSValue.cpp:391
> >      if (exec->hadException())
> 
> this can be "vm.exception()"
> 
> > Source/JavaScriptCore/runtime/JSCJSValue.cpp:395
> > +    if (exec->hadException())
> 
> this can be "vm.exception()"

Thanks, fixed!

[Filip Pizlo]

Created attachment 272679 [details]
patch for landing

[Build Bot]

Comment on attachment 272679 [details]
patch for landing

Attachment 272679 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.webkit.org/results/912394

Number of test failures exceeded the failure limit.

[Build Bot]

Created attachment 272682 [details]
Archive of layout-test-results from ews100 for mac-yosemite

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100  Port: mac-yosemite  Platform: Mac OS X 10.10.5

[Build Bot]

Comment on attachment 272679 [details]
patch for landing

Attachment 272679 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/912390

Number of test failures exceeded the failure limit.

[Build Bot]

Created attachment 272683 [details]
Archive of layout-test-results from ews104 for mac-yosemite-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews104  Port: mac-yosemite-wk2  Platform: Mac OS X 10.10.5

[Build Bot]

Comment on attachment 272679 [details]
patch for landing

Attachment 272679 [details] did not pass mac-debug-ews (mac):
Output: http://webkit-queues.webkit.org/results/912384

Number of test failures exceeded the failure limit.

[Build Bot]

Created attachment 272684 [details]
Archive of layout-test-results from ews116 for mac-yosemite

The attached test failures were seen while running run-webkit-tests on the mac-debug-ews.
Bot: ews116  Port: mac-yosemite  Platform: Mac OS X 10.10.5

[Filip Pizlo]

Created attachment 272687 [details]
the patch

Lets try this again.

[Saam Barati]

Comment on attachment 272687 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=272687&action=review

> Source/JavaScriptCore/runtime/RegExpPrototype.cpp:104
> +    if (!string)
> +        return JSValue::encode(jsUndefined());

It might be worth adding a debug assert that an exception was thrown to make the call site contract of toStringFast more explicit

> Source/JavaScriptCore/runtime/RegExpPrototype.cpp:115
> +        return JSValue::encode(jsUndefined());

ditto

[Filip Pizlo]

(In reply to comment #12)
> Comment on attachment 272687 [details]
> the patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=272687&action=review
> 
> > Source/JavaScriptCore/runtime/RegExpPrototype.cpp:104
> > +    if (!string)
> > +        return JSValue::encode(jsUndefined());
> 
> It might be worth adding a debug assert that an exception was thrown to make
> the call site contract of toStringFast more explicit
> 
> > Source/JavaScriptCore/runtime/RegExpPrototype.cpp:115
> > +        return JSValue::encode(jsUndefined());
> 
> ditto

I want to call toStringFast() in a lot of places.  My next patch will add several more calls.  The idea is that null-checking the return is even faster than vm.exception() in this case.

So, maybe we can either agree that "toStringFast()" means this, or we can come up with a better name for the function that does convey this meaning?  I'm fine with renaming it if the new name is not long.  If the alternatives use a word that is longer than "fast" then I think "fast" is least bad.

[Geoffrey Garen]

Comment on attachment 272687 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=272687&action=review

> Source/JavaScriptCore/runtime/JSCJSValue.h:256
> +    JSString* toStringFast(ExecState*) const; // On exception, this returns null, to make exception checks faster.

I would call this toStringOrNull. It's two letters / one word longer than "fast" but I think it conveys the meaning, and the action required by the caller, much more clearly. So, I think it's worth it.

A shorter option, which I like less but still keeps with project naming conventions, is "tryToString".

[Filip Pizlo]

(In reply to comment #14)
> Comment on attachment 272687 [details]
> the patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=272687&action=review
> 
> > Source/JavaScriptCore/runtime/JSCJSValue.h:256
> > +    JSString* toStringFast(ExecState*) const; // On exception, this returns null, to make exception checks faster.
> 
> I would call this toStringOrNull. It's two letters / one word longer than
> "fast" but I think it conveys the meaning, and the action required by the
> caller, much more clearly. So, I think it's worth it.

OK.

> 
> A shorter option, which I like less but still keeps with project naming
> conventions, is "tryToString".

tryToString would not be consistent - that is used in other string code to mean "try to give me the string without doing anything effectful".

[Filip Pizlo]

Landed in http://trac.webkit.org/changeset/197485