W/ xfwm 4.11.2, probable regression from 4.10 - if i have a fullscreen app (say, terminal or ffx) in front of a non-fullscreen app and i focus the latter, focus is transmitted but the fullscreen app stays on top (and still hides the other one) instead of having it brought up to the front. Pretty sure in 4.10 this wasnt the case...
< gaston+> hmm another potential xfwm4 regression in 4.11.. have a fullscreen app, alt-tab -> the other apps are not brought in front of the fullscreen win, only focused..
<kalikian+> gaston: does the fullscreen app stay in front whilst the focus changes to the destined app? if so I'm seeing exactly that
On that matter, i have STR on a xinerama setup. have the window on top of fullscreen app, alt-tab to the fullscreen app -> the window disappears. Alt-tab to the hidden window -> doesnt come up. Focus other window on other screen with mouse, focus back the fullscreen window, focus back 'other window on other screen' - the previously hidden windowed app comes on top of fullscreen app, unfocused.
just to clarify, this doesnt happen with maximized windows, only fullscreen ones.
I can also confirm it's a regression - on two computers with xfwm 4.10, having a fullscreen window (be it a term or a firefox window), alt-tabbing to another window correctly brings it to the top, showing the window on top of the fullscreen one (and the xfce panel, if the said window is not fullscreen too)
Bisect revealed the regression is caused by http://git.xfce.org/xfce/xfwm4/commit/?id=7c87af29befcc91ba2e890dd822a4a69cb3263d9
Specifically this piece of code from the above commit causes the problem:
c2 = myScreenGetClientFromWindow (screen_info, GDK_WINDOW_XID (gtk_widget_get_window ( passdata.tabwin->tabwin_list->data)), SEARCH_FRAME);
g_message ("%p", c2);
clientSetFocus (screen_info, c2, ev->time, NO_FOCUS_FLAG);
Nick, any idea ?
I can confirm that commenting out these 3 lines fixes the issue i'm seeing - will check if i get other regressions...
And i also have the strange feeling it's also fixing bug 11166 while at it.. will continue using it this way to confirm.
What bother me slightly is that it's unclear to me what the patch that introduced the regression was supposed to fix in the first place...
*** Bug 11120 has been marked as a duplicate of this bug. ***
Should be fixed with git commit 0151f6b
yeah, reconfirming since that commit removes the lines i had to comment out to workaround it when i tested in comment 7.. that leaves bug 11166.