[Webkit-unassigned] [Bug 144166] Function allocation sinking shouldn't be performed on singleton functions

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=144166

Filip Pizlo <fpizlo at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #251587|review?, commit-queue?      |review+
              Flags|                            |

--- Comment #5 from Filip Pizlo <fpizlo at apple.com> ---
Comment on attachment 251587
  --> https://bugs.webkit.org/attachment.cgi?id=251587
Fix ChangeLog typos

View in context: https://bugs.webkit.org/attachment.cgi?id=251587&action=review

Holding off cq because the ChangeLog comment could be refined.

> Source/JavaScriptCore/ChangeLog:14
> +        watchpoints invalidating code that depends on it being a singleton).  According
> +        to fpizlo, this happens to be sinkable right now, but things may not stay that
> +        way in the future (e.g. it would not be the case for CreateActivation), and
> +        fixing this now would help prevent hard-to-find bugs later in that case.

What won't stay that way in the future?  That singleton function allocations are sinkable?

I think it's enough to say that we don't want to sink code that has effects if we haven't done the work to prove that the effect wasn't observed.  Object allocation sinking phase does no such analysis because it assumes that the act of allocating is not effectful.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20150427/5c9876d3/attachment.html>