Summary: | Improve support for function call nodes in SquirrelFish | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Cameron Zwarich (cpst) <zwarich> | ||||||||||
Component: | JavaScriptCore | Assignee: | Nobody <webkit-unassigned> | ||||||||||
Status: | RESOLVED FIXED | ||||||||||||
Severity: | Normal | CC: | ggaren, mjs | ||||||||||
Priority: | P2 | ||||||||||||
Version: | 528+ (Nightly build) | ||||||||||||
Hardware: | Mac | ||||||||||||
OS: | OS X 10.5 | ||||||||||||
Attachments: |
|
Description
Cameron Zwarich (cpst)
2008-03-30 23:55:33 PDT
Created attachment 20231 [details]
Proposed patch
Created attachment 20232 [details]
Proposed patch (now with email!)
When I made a new ChangeLog after Geoff's latest patch I forgot to add my email address.
Comment on attachment 20232 [details]
Proposed patch (now with email!)
r=me
After r31455, I hit a SquirrelSpider regression of about 3%. I'll try to work it out. Created attachment 20234 [details]
Revised proposed patch
Here is a fairly trivial patch containing only the support for FunctionCallValueNode. This doesn't regress SunSpider.
Comment on attachment 20234 [details]
Revised proposed patch
r=me
Committed in revision 31456. This isn't actually fixed, because I had to remove part of the patch due to a performance regression. It is required functionality, so hopefully I can find a way to avoid the performance problems by permuting the code a bit to make GCC happy. Created attachment 20469 [details]
Proposed patch
I figured I would revisit this, and it turned out to be a SunSpider progression. What a strange world we live in.
Comment on attachment 20469 [details]
Proposed patch
r=me
Committed r31810 |