[compiz] incomplete window contents when unminimizing

[Erkin Bahceci]

I submitted a bug report to Mozilla for this:

https://bugzilla.mozilla.org/show_bug.cgi?id=378293


On 4/13/07, David Reveman <davidr at novell.com> wrote:
> On Wed, 2007-04-11 at 03:29 -0500, Erkin Bahceci wrote:
> > Hi,
> >
> > Using Xgl, when a window is unminimized, sometimes (especially with
> > Firefox), the window contents are not immediately visible (partially
> > or completely) until an unminimize animation is completed. That is,
> > sometimes there are empty (transparent) rectangles within the window
> > (maybe corresponding to some Mozilla controls), and sometimes the
> > whole window is empty (only the decoration is visible). This is more
> > of an issue with cpu-hungry effects (well, most of them if you are
> > using Xgl + fglrx).
> >
> > To avoid this, I want to wait until the window contents are *fully*
> > available before I start any unminimize animations. My question is how
> > can I know the right time? Is there a certain place in core that I can
> > maybe throw a notification event from for this? If not, what is the
> > best way to solve this problem? Should I just wait for a certain
> > amount of time for certain applications (like Firefox)?
>
> The sync request protocol (_NET_WM_SYNC_REQUEST in the EWMH spec) allow
> this and compiz should automatically wait with drawing windows until
> window contents is *fully* available for applications that support it.
>
> A damageWindowRect call with "initial" argument set to TRUE is the
> indication that a newly mapped window has become available for drawing.
>
> It's not much we can do for applications that don't support the sync
> request protocol. Compiz will currently consider the first damage event
> an indication that the window is available for drawing and I don't think
> we can do much better than that.
>
> All gtk apps should automatically support this protocol but I'm not sure
> about Firefox. Waiting for a certain amount of time is a bad idea. If an
> application doesn't support the sync request protocol you should bug the
> developers.
>
> - David
>
>