GNOME Bugzilla – Bug 668573
Fix rendering of the handle in panel applet frame widget
Last modified: 2012-03-02 15:26:40 UTC
Created attachment 205973 [details] [review] git patch 1/2 Attached git patches fix the rendering of the handle in the panel applet frame widget. NB - Although patches are against branch 3.2, they should apply cleanly to git master.
Created attachment 205974 [details] [review] git patch 2/2
Hrm, could you attach before/after screenshots? I'm not sure which theme you're using, but it doesn't make any difference here with Adwaita (where the handle is really just not painted at all, afaict).
(In reply to comment #2) > Hrm, could you attach before/after screenshots? Sure. > I'm not sure which theme you're > using, but it doesn't make any difference here with Adwaita (where the handle > is really just not painted at all, afaict). Probably bugs in theme/engine. I use Radiance (with the unico engine), and had to fix bugs in both wrt gnome-panel. First, Radiance uses a "border-width: 1" for everything, so things that don't draw frames (like PanelApplet and PanelAppletFrame) show a box artifact because on background wiping the fill is reduced by the border amount (same as stock engine). Second, the unico engine uses the border-width as the line width for stroking the handle and separator -- which is bad when the "border-width: 0". Third, css styling was wrong for the showdesktop-button so the background was being wiped to the default button style (thus no 'transparent' effect).
Created attachment 206658 [details] After patch - handle rendered in correct location with correct background Cropped screenshot of handle rendered after patches applied (+ bug fixes to theme and engine)
What's the situation with these patches? Is more information required here to push these into master?
Created attachment 208854 [details] [review] Fixes regression introduced by patch 2/2 above This additional patch fixes a regression introduced by the git patch 2/2 submitted earlier. Please let me know if a single squashed patch is preferable. The regression occurred when the system theme used a gradient background image. In this case, a background pattern exists but the pattern matrix needs to remain the identity matrix.
I've squashed the patches in a single patch, and pushed, thanks!