[Webkit-unassigned] [Bug 135046] Turn width/height to presentation attributes

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=135046





--- Comment #15 from Dirk Schulze <krit at webkit.org>  2014-07-22 05:47:53 PST ---
Created an attachment (id=235285)
 --> (https://bugs.webkit.org/attachment.cgi?id=235285&action=review)
Performance test

(In reply to comment #14)
> (From update of attachment 235181 [details])
> View in context: https://bugs.webkit.org/attachment.cgi?id=235181&action=review
> 
> > Source/WebCore/svg/SVGElement.h:104
> > +        // Trigger style recalculation for "elements as resource" (e.g. referenced by feImage).
> > +        setNeedsStyleRecalc(InlineStyleChange);
> 
> Is this going to be a big performance hit?

I attached a performance test. I commented out the line in question and it did not have a measurable performance impact. In general with the patch we have 30%-40% more runs/s than without the patch (tested on tip, both with debug build). I wouldn't take this result too serious though. Even if the results are stable (~390 runs/s w/o patch ~610 runs/s with patch), it just might be that we bypassed some special testing that we just compile for debug builds. Testing release builds would make more sense.

For now it is just important that we do not regress with setNeedsStyleRecalc on performance.

To the test: It switches the relative width of a couple of dozens rectangles between 98% and 99%. Just elements that are NOT resources by other elements are from interest for this line.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.