Bug 101665 – DFG should know that int == null is always false

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Filip Pizlo]

Patch forthcoming.

[Filip Pizlo]

Created attachment 173146 [details]
the patch

[Filip Pizlo]

Landed in http://trac.webkit.org/changeset/133990

[Darin Adler]

Comment on attachment 173146 [details]
the patch

View in context: https://bugs.webkit.org/attachment.cgi?id=173146&action=review

> Source/JavaScriptCore/ChangeLog:3
> +        DFG should know that int == null is always false

Seems more specific than I would have expected. Isn’t double == null also always false? And string == null? And lots of other cases like that?

[Filip Pizlo]

(In reply to comment #3)
> (From update of attachment 173146 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=173146&action=review
> 
> > Source/JavaScriptCore/ChangeLog:3
> > +        DFG should know that int == null is always false
> 
> Seems more specific than I would have expected. Isn’t double == null also always false? And string == null? And lots of other cases like that?

You're right, this is too specific.  On the other hand, I was playing it safe because even for int == null Gavin and I ended up scratching our heads over whether or not 0 is equal to null.

Also, whereas we have test coverage for int == null being compiled correctly (V8/crypto), we don't have such coverage for the other cases.  So, that patch will require somewhat more work since I don't want to add optimizations that aren't covered.

I've opened a bug for this: https://bugs.webkit.org/show_bug.cgi?id=101680