[Webkit-unassigned] [Bug 66650] [Qt] Flash in a QGraphicsWebView using OpenGL is slow

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=66650





--- Comment #20 from Christophe Oosterlynck <christophe.oosterlynck at dzine.be>  2011-11-03 14:23:21 PST ---
(In reply to comment #19)
> (From update of attachment 113167 [details])
> I'm sorry, I'm giving this an r-. At the very least because the ChangeLog doesn't make much sense
> by saying that "Plugins are now drawn on a QPixmap instead of on a X11 Pixmap when the OpenGL backend is used".
> 
> X11 windowless plugins can only be rendered onto X11 pixmaps (technically it's a general purpose X11 drawable, but a pixmap
> by all means). The following line is responsible for passing the information _where_ the plugin should paint to:
> 
>     exposeEvent.drawable = qtDrawable.handle();
> 
> That means qtDrawable must be a QPixmap that is backed by an X11 pixmap.
> 
> This is the case with the current code and unless you're using the OpenGL graphics system, a regular QPixmap instance allocated
> with Qt compiled for X11 will _also_ be backed by an x11 pixmap. So the statement "...draw onto QPixmap instead of on an X11 Pixmap" doesn't make sense because QPixmap is _also_ backed by an X11 pixmap, returned by handle().
> 
> If however the OpenGL graphics system is used, then QPixmaps will not be backed by X11 pixmaps and QPixmap::handle() will return 0. A 0
> drawable is certainly not going to make the plugin paint.
> 
> 
> So either I am missing something (in which case please elaborate in the ChangeLog) or the patch will not have any effect and certainly break plugin rendering when the OpenGL graphics system is used, because QPixmap::handle() will return 0 unless the QPixmap was constructed using QPixmap::fromX11Pixmap.

Hmm, it does make sense what you're saying and indeed makes my patch questionable. Although I'm pretty sure I saw performance increase with my patch when using OpenGl 1 (backed up by profiling) and the plugin (flash) was still visible.

But it seems that I am following the wrong path to solve my issue. I don't have much time these days to work on my issue anymore, but I'll try to look into it in the near future and provide a better patch (or maybe get it fixed in Qt if the problem lies not within webkit) and be back here with an update.

Thanks for the input!

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