Created attachment 291627 [details, diff] Backport Martin Gräßlin's performance work from KDE trunk to KDE 4.7.3 People on the forums are complaining that "KDE effects are slow and laggy" and that the "efforts to speed up the effects" "target KDE 4.8", so I decided to backport Martin Gräßlin's performance work from KDE trunk to KDE 4.7.3. http://forums.gentoo.org/viewtopic-t-895068-postdays-0-postorder-asc-start-0.html The backport involved commits bb6d4873259691b189a8b706f9dfc2e858f54609 and 5e8be6288cc34702bfabcc730bb1168e1c4e9eee from git://anongit.kde.org/kde-workspace.git. Commit bb6d4873259691b189a8b706f9dfc2e858f54609 didn't apply cleanly to kwin 4.7.3, but manually merging the failures was straightfoward. I have only tested this on amd64 unstable. I wanted to have some people in IRC test this before I filed this bug report, but ABCD in #gentoo-kde on freenode asked me to file anyway, so here it is.
For the eager, running the following command (all one on line) will unpack kwin, download the patch, apply it and rebuild kwin with the patch: ebuild $(equery which kwin) unpack && wget -O - "https://389467.bugs.gentoo.org/attachment.cgi?id=291627" | patch -d /var/tmp/portage/kde-base/kwin-4.7.3/work/ -p1 && ebuild $(equery which kwin) merge
Talked to MG and this is way to risky to go into main tree 4.7.3. I'll make an ebuild for the overlay later on, but without keywords - "for the brave or insane".
(In reply to comment #2) > Talked to MG and this is way to risky to go into main tree 4.7.3. I'll make an > ebuild for the overlay later on, but without keywords - "for the brave or > insane". Would you be more specific as to why he said that this is risky? Looking through the patch, it seems that the only real issue would be that this breaks API compatibility with out of tree effects, but the in-tree effects should be fine. I haven't had a single problem since making this change to my system.
http://blog.martin-graesslin.com/blog/2011/08/rendering-at-60-frames/comment-page-1/#comment-44945
Given how vocal Martin is in opposing, I dont think we should put any effort into it. After all, 4.8.0 will also happen fairly soon (beta1 release planned for 23/nov/11).
(In reply to comment #5) > Given how vocal Martin is in opposing, I dont think we should put any effort > into it. After all, 4.8.0 will also happen fairly soon (beta1 release planned > for 23/nov/11). This is why I wanted to try to find testers on IRC before I filed a bug report. Anyway, I have no issues with the patch so far. I suspect that some of the regressions Martin mentioned were caused by other changes that are going into HEAD and wouldn't occur in 4.7.x, but I will try hunting for regressions when I have more time.