[Webkit-unassigned] [Bug 175584] We are too conservative about the effects of PushWithScope

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=175584

Saam Barati <sbarati at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #318133|review?, commit-queue?      |review-, commit-queue-
              Flags|                            |

--- Comment #2 from Saam Barati <sbarati at apple.com> ---
Comment on attachment 318133
  --> https://bugs.webkit.org/attachment.cgi?id=318133
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=318133&action=review

> Source/JavaScriptCore/ChangeLog:7
> +

It's worth a sentence describing the reasoning here.

Given the semantics of PushWithScope, I think it's also worth changing Clobberize and DoesGC (based on making Clobberize not report it clobbers things).

Can you also add tests that would stress a few things:
- we CSE around a PushWithScope
- cases where PushWithScope throws an exception in the DFG because the "o" is null/undefined
- cases where PushWithScope allocates (e.g, the "o" is a number/string/Symbol)

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170815/e89c318c/attachment-0001.html>