[compiz] bug in today's git snapshot with maximizing windows

[David Reveman]

On Fri, 2006-11-10 at 15:00 -0500, Mike Cook wrote:
> On Wed, 2006-11-08 at 15:16 -0700, David Reveman wrote:
> > On Wed, 2006- 11- 08 at 14:36 - 0700, Mike Cook wrote:
> > > Per David's response to my comment, I'm willing to try hacking up a patch
> > > for the workarea- per- output idea.  Then updates in window.c and maybe
> > > the place plugin could use that and we'd get more proper placement and
> > > window maximizing per output.  However, I've noticed some discussions on
> > > the _NET_WORKAREA and multiple heads and it sounds like there needs to
> > > be changes eventually to the spec to better handle these multiple output
> > > cases.  But in the meantime, I think we can make it work mostly right.  ;)
> > 
> > Great, I look forward to seeing patches for this. Please keep the
> > per- output workarea addition and the updates to window.c and eventually
> > updates to some of the plugins in separate patches. I think at least
> > place, move and wobbly plugins will need updates but just getting the
> > per- output workarea into head will be a good start.
> > 
> > _NET_WORKAREA is a rectangle but our internal representation in compiz
> > will be a set of rectangles. We should try to get the EWMH spec updated
> > sometime soon as being able to communicate a non- rectangular workarea to
> > apps and toolkits is important for the dynamic multi- head support that
> > compiz will be able to do.
> > 
> > - David
> 
> Alright, here's an initial patch which implements a workarea in the CompOutput
> structure, and sets it up each call to updateWorkareaForScreen.  I've been
> working on some updates to window and plugins to properly use it, but haven't
> had time to finish that so wanted to get this much submitted in the meantime...

Thanks, I applied you're patch. I found one issue with the current code
though.. We shouldn't use the window position and size to determine if a
window should affect the workspace of a output or not. The
_NET_WM_STRUT_PARIAL hint contains everything we need to determine if it
should affect an output or not.

I removed the MIN_SANE_AREA stuff as I would like it to be possible to
have a zero workarea on some outputs.

I changed these things and pushed it out. You should have a look at it
and let me know what you think.

Thanks,

-David