[Webkit-unassigned] [Bug 56929] [Qt] Tiled painting still causes synchronous layout when accelerated compositing and texture mapper are enabled

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=56929





--- Comment #9 from Benjamin Poulain <benjamin at webkit.org>  2011-03-28 09:51:11 PST ---
(In reply to comment #6)
> > Maybe change the signature to use QFlags and avoid the ugly cast?
> 
> I would gladly do that, but I would like to do that on renderRelativeCoords as well in this case. Shouldn't I?

Yep, sounds reasonable.

> What do you think about renderFrameExtras / renderFrameVisuals ?

Yep, I like both, I let you pick the best one.

> > > Source/WebKit/qt/Api/qwebframe.cpp:394
> > > +    if (!layer & (QWebFrame::PanIconLayer | QWebFrame::ScrollBarLayer))
> > > +        return;
> > This could be an assertion instead of a runtime check.
> 
> This is a runtime check copied from renderRelativeCoords. It cannot simply be turned into an ASSERT as renderRelativeCoords can cause false assertions. 

Yep, I was obviously thinking about changing also the way it is called, like masking the layers.


> I moved it here since I considered that the robustness added by it overweighs its low cost. I figured that most callers would have to do the test anyway so I save some memory by putting it here, rather then to do the check before every call. If you want I can totally remove the check in which case I uselessly walk through the rects of the region painting nothing if the function is called with neither of the interesting layers specified, or I can move the test to renderRelativeCoords and put an ASSERT in the function.

I was not talking about performance but code robustness. Your call on that, I don't mind the if() if you think it is more robust..

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