On Tue 10 Sep 2013 at 17:26:31 -0400, Stefan Monnier wrote:
> > The patches otherwise look good to me, so I would propose to commit them
> > (after removing the debugging prints that I added) to my branch.
> 
> I assume you can do that for me.  And thanks for your help,

Yes, I've done it now (in the branch
"free.lp.se:X.ctwm.rhialto.cleanup") and pushed to the repo.  I got a
bit bogged down in virtual screen stuff... that stuff is really crufty
and I suspect noone uses it. It has too many weird bugs that apparently
never got reported or fixed.

I think I know what issues you have with virtual screens and your
OnTopPriority code. With virtual screens, each screen has a separate
root window, and each of those has its own window stack. Your
consisitency check however assumes there is one global stack though,
like actually most of the ctwm code does.

That isn't too difficult to code around in the OtpCheckConsistency()
function. I also had to modify ReparentWindow(). It tries to re-stack a
window in the correct position. However it needs to find a proper window
as reference, i.e. one that has the same parent (= the same vscreen).

The next issue is another feature that is probably pretty much unused:
WindowBoxes. Those basically also function as separate root windows.
And they have their own quirks, as I noticed quickly when I tried the
example from the manual. Positioning went wrong already when placing the
window (the coordinate systems were mixed up; inside the box it got
location (x,y) where (x,y) were (probably) the location relative to the
Root window).

I've changed the OtpCheckConsistency() to avoid checking windows in
WindowBoxes but that doesn't seem to be enough.t I'll have to ponder
that some more.

>         Stefan
-Olaf.
-- 
___ Olaf 'Rhialto' Seibert  -- The Doctor: No, 'eureka' is Greek for
\X/ rhialto/at/xs4all.nl    -- 'this bath is too hot.'

Attachment: pgpKiMpF5AdCY.pgp
Description: PGP signature

Reply via email to