lp://staging/~vanvugt/compiz-core/fix-981703

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp://staging/compiz-core at revision 3093

Sam Spilsbury: Approve on 2012-04-16

Diff: 19 lines (+1/-3)

1 file modified

src/window.cpp (+1/-3)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #981703: regression / unable to interact with window-titlebar (window decoration) after minimizing/unminimizing gnome-terminal

High

Fix Released

Link a bug report

Related blueprints

3093. By Daniel van Vugt on 2012-04-16

Revert r3092 because it was causing a worse regression (LP: #981703)

This reverts the fix for LP: #980026.

3092. By Alan Griffiths on 2012-04-13

Avoid dumping all the windows into first workspace when closing (LP: #980026)

Reverts to correct logic used before r2838.

3091. By Daniel van Vugt on 2012-04-12

Avoid comparison between signed and unsigned integers (LP: #972519)

Strange. All versions of gcc should have reported this as an error. But
obviously gcc prior to 4.7 had bugs and failed to find all such errors.

3090. By Daniel van Vugt on 2012-04-11

Minor corrections to the 0.9.7.6 NEWS

3089. By Sam Spilsbury on 2012-04-10

Fix shadows being clipped incorrectly (LP #976467)

1) Don't include _NET_WM_WINDOW_TYPE_MENU windows in the clip groups ...
   those windows are a lot different from what you expect of them
2) Also update clip regions when the shadow output extents change even
   if the window isn't resized.

3088. By Daniel van Vugt on 2012-04-06

Bump version to 0.9.7.7 while in between releases.

3087. By Sam Spilsbury on 2012-04-06

Release 0.9.7.6

3086. By Sam Spilsbury on 2012-04-05

Remove DecorWindow::computeShadowRegion and replace with a smarter class.
(LP: #969101)

DecorClipGroup (split into GenericDecorClipGroup and MatchedDecorClipGroup)
take a DecorClippableInterface which where entry and exit from the group is
guarded by the implementation and are treated as one layer for the purposes
of shadowing - eg the input region of each member of the clip group will clip
all of the shadows of the members, so that none of the members appear to be
overlapping.

This algorithm is a lot more optimized in that the shadows are only updated
when relevant windows are updated.

Tested too.

3085. By Sam Spilsbury on 2012-04-05

Stab in the dark (a possible but unconfirmed fix for LP: #940603)

I think whats going on for bug 940603 is that windows aren't being removed
from the window list if they are unreparented by the application itself in
very special circumstances (not sure what because I can't reproduce the bug).

This code adds some more sound checks to see where the window is going and
destroys it based on that, rather than using the same logic for both
reparented and unreparented windows.

Also removed the whole destroyedFrameWindows tracking code which was more or
ineffectual and could potentially result in a frame window leak.

3084. By Sam Spilsbury on 2012-04-01

Use gtest_add_tests and format the code so that it picks them up