[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Ayatana] [Fwd: Re: Fwd: Open Letter: The issues with client-side-window-decorations]



> > From: Martin Gräßlin <kde@xxxxxxxxxxxxxxxxxxxx>
> > It would be unacceptable that windows cannot be closed if they 
> > hung.

Hi Martin,

Sorry I've been absent from this discussion, I haven't really felt I had
much to say yet.  And I generally avoid Ayatana list because I have the
same problem you have with it.. all the email goes to my @canonical.com
but whatever I write from that address gets blocked, while mail from my
@gnome.org goes through for some reason.

Anyway, I just wanted to comment on this statement and point out that
this is not a hugely difficult problem to solve with some cooperation
from a WM.  I would assume a WM can monitor _NET_WM_PING and either
popup a dialog, or we could create a hint to expose a region of the
close button to the WM for this purpose.

So one reason I have not been commenting on this discussion is that
quite frankly, I agree with you that right now they may not be ready to
land in a distribution due to many of the problems you have cited.  It
seems sort of clear to me that there needs to be more development time
in order for CSD to catch up in certain respects.  However, this in
itself is not a reason to abandon the development effort altogether.

>From what I can tell, window manager decorators need to jump through an
extraordinary number of hoops in order to make a window decoration and
the application appear to be one.  KWin seems to do this really well,
and I think we're all really impressed with it.  But at the same time,
why should we not explore the opposite approach as well?  Rather than
having two processes trying to draw what is essentially the same
application and do a lot of work to make them look the same, why not try
drawing them from the same process and do less work to get the
functionality back (e.g., the close button hint I described above)?
This strikes me as being a lot less work than trying to do in our WM
what you've done in KWin.

I disagree that our approach is fundamentally flawed.  But I admit that
I do agree we perhaps should not be pushing it into our distro yet, and
we should be working with you guys on wm-spec-list more to discuss
issues like the hints I described above before we try to really release
it.  This is my fault, not some Canonical conspiracy to try to fuck you
guys over or anything. ;)  I'm dealing with other projects, and due to
the relatively short development cycle for Ubuntu it sometimes makes it
difficult for me to plan and execute proper coordination with upstreams,
especially ones that I'm not very familiar with (such as wm-spec-list).
I'll take full responsibility for that, so please don't hold it against
Ubuntu or Canonical.  Hold it against me if you want to. ;)

/ Cody