returns the state set by the user - these two may differ e.g. while hover-unshaded.
Therefore titlebar buttons should use isSetShade(), otherwise they become
a bit useless with hover-unshading.
Also fix updating the button state - the styles duplicate way too much code,
and if one of them has a bug, most of them definitely have it as well. Some
of the code should probably move to the decoration lib :-/ .
svn path=/trunk/kdebase/kwin/; revision=366147
Initial work on kwin rules, i.e. #36377 , per window specific settings.
So far only desktop/above/below work, no GUI, and settings from the old
'Save window settings' are ignored for now.
svn path=/trunk/kdebase/kwin/; revision=315446
workaround, don't make their user timestamp newer than the active window's
one (unless a real user activity takes place in them).
As they are belong to the active application and just fail to say so,
this makes sure they won't prevent that application from getting focus
by having newer timestamp. E.g. Alt+F2, typing URL, kio_uiserver dialog
shows (has workaround), SSL certificate dialog shows (shown by kdesktop),
and it wouldn't get focus, because kio_uiserver's timestamp would be later.
svn path=/trunk/kdebase/kwin/; revision=298357
reportedly improves move/resize performance on some systems
for some reason which I fail to see, but oh well.
(http://lists.kde.org/?t=107302193400001&r=1&w=2)
svn path=/trunk/kdebase/kwin/; revision=293755
on the whole group, so that the possible 'close?' dialog has a recent
timestamp even if the window about to be closed wasn't active.
svn path=/trunk/kdebase/kwin/; revision=291859
Qt3.3 capitalizes it (part of my own patch IIRC, hehe :-/), so all
the workarounds for specific apps no longer work -> force resource class
to be always lowercase.
svn path=/trunk/kdebase/kwin/; revision=290521
stacking order for Qt-3.2.x fullscreen windows not patched to use
NETWM fullscreens. Tested by Laurent Montel.
svn path=/trunk/kdebase/kwin/; revision=281817
because if one timestamp is reused for activating window A and then
immediatelly window B, and A will want to set focus to one
of its subwindows after receiving WM_TAKE_FOCUS, and will lag a bit,
it will actually set the focus after B, and will win. Fixes#71704.
If somebody still gets the warning, I'd be interested in the backtrace.
svn path=/trunk/kdebase/kwin/; revision=278208
set, it remembers the contents of the unmapped client window, therefore
the ugly effect actually stayed.
svn path=/trunk/kdebase/kwin/; revision=273891
"smart" apps from doing stupid things before they get MapNotify.
This avoids the recent Java+OracleInstaller strange problem too.
The patch may look huge, but it's only adding two KWin wrappers
for X(Un)GrabServer() and uncommenting the calls in Client::manage().
svn path=/trunk/kdebase/kwin/; revision=271152