[Webkit-unassigned] [Bug 52311] [chromium] Add command-line flag to enable composite to offscreen texture.

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=52311





--- Comment #49 from W. James MacLean <wjmaclean at chromium.org>  2011-02-01 07:41:53 PST ---
(In reply to comment #48)
> 
> They colorMask calls should be removed anyway as they're not consistent anymore with the state that the compositor uses (compositor expects color to be unmasked most of the time).  

Done.

> My only other guess for the differences you're seeing is that something is off by a fraction of a pixel and you're seeing slight stretching. Our textures are by default using LINEAR filtering.  You could try setting the TEXTURE_MIN_FILTER, TEXTURE_MAG_FILTER to NEAREST .  You could do that in RenderSurface::prepareContentsTexture, after the call to reserve() you can bind the texture and set the filter modes.

This seems to have helped. 12 of the 14 failing tests now pass, and the two remaining failures have considerably fewer pixels failing. The the remaining failures are

fast/canvas/canvas-text-alignment.html
fast/canvas/canvas-text-baseline.html

All of the remaining pixel mis-matches are along high-contrast edges (vertical and horizontal lines).

> The reason I'd like to get to the bottom of this is that it would be nice to know that you can use the off-screen path as your primary render path which means that you'd need the layout tests to work.

Rest assured, I don't want *any* failing testswhen we finish, but I also want to head off regressions that might sneak in in the meantime since the offscreen path is not being tested at present.

I'll test the new version of the patch on Mac and then upload it.

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