I'm using oxygen-gtk2 1.2.5 with a dark kde theme on a fresh Slackware64-current installation. I spoke before this report with Ruslan Kabatsayev and he mentioned: > As for the glitch, it looks very familiar to me, but I can't remember > when I saw similar thing (but it seems to have been very long ago). > This is likely a bug in cairo, you might have too old (or new) > version. I don't reproduce this bug with cairo 1.8.10 and the version > in ubuntu 12.04. Have a look at the attached screenshot, there a 2 bright vertical stripes on top of some GTK applications. My installed Packages: - KDE 4.8.95 - GTK+2 2.24.10 - Cairo 1.10.2 Reproducible: Always Steps to Reproduce: 1. Open a GTK Application like Pidgin Actual Results: You can see the glitch on the top of the application. Expected Results: Draw the Background/Widget (?) correctly.
Created attachment 72358 [details] Here is a screenshot of the glitch.
sadly enough can't reproduce either with cairo-1.10.2, gtk-2.24.10 (just like you), oxygen-gtk from trunk (of course), and kde a mixture of 4.8.4 and trunk (but that should not matter anyway: there is no dependency of oxygen-gtk on kde) See: http://wstaw.org/m/2012/07/07/plasma-desktopGI2735.png Cant reproduce with oxygen-gtk-1.2.5 either. So that I am pretty clueless. What's your graphics card ?
I use the Intel HD4000 GPU (i7 3770K). With: - xf86-video-intel-2.19.0 - mesa-8.0.3 - xorg-server-1.12.2
... I have intel also (i915), x11-server-xorg-1.11.4-2.mga2 mesa-8.0.2 x11-driver-video-intel-2.19.0-2.mga2 So older x11 server and slightly older mesa ... mmm.
@Hermann Could you check if the glitch reproduces with earlier oxygen-gtk releases? If it doesn't, then you could do a git-bisect between working and non-working versions to find the commit which introduces the bug.
Sure, I recompiled some versions and tested them and here are my results. oxygen-gtk2-master [ok] oxygen-gtk2-1.2.5 [fail] oxygen-gtk2-1.2.4 [ok] So only 1.2.5 produces this glitch. Sounds quite good for now.
So its fixed in master ? That's good news, since there will be no new release of oxygen-gtk2 1.2 series. Next one (around July 14) will be oxygen-gtk 1.3.0, branched from master. Still, would be nice to know what actually fixed it :(
@Hermann, If you feel like testing (have time for this), could you please do git-bisect between master and v1.2.5 marking buggy state as good and working as bad, so that we knew what fixed this? @Hugo If we know which commit fixed it, maybe we'll release one more bug-fix release (maybe minor, like 1.2.5-1) along with 1.3.0? Though there's not been any activity in 1.2 branch I guess, so dunno if one commit is worth it...
@Ruslan, I'm fine to make another 1.2.x release, but it means that one has to figure out whether the bug is fixed in 1.2 branch from git (and likely it is not), and if not, figure which commit in master, fixed it (your git-bisect suggestion), in order to backport it in the 1.2 branch.
Yeh, its fixed in master. Good to hear the next branch is not far away so the next stable tag will fix it. > Still, would be nice to know what actually fixed it :( Mhh, there are 21 commits from 1.2.5 to master, and testing this is a bit stressy. But I do it for you ;) Thank you for your time and help. :) Ps: This a good piece of software. Great work.
@Hermann, with git bisect (look for the doc), you don't need to go through all 21 commits one by one (that's the whole point). It uses binomial seach for doing it, so you would need something like 3 to 5 rounds to test. (and thanks for the encouragement)
good-start f03ac25c3ab057d51e7e1a8f02f0fc2b9569d3f8 good 9ec1e6a0da437e6b1fad99cccb2796264be119c4 good 71cde7b5cef52a234c25459b7d18285ef44516d2 bad 595ee7e74a62cfadfc7e851a155d00c70bc45eeb bad 17c95dccbd74c3b7c6eb2a73de47c06d049a1f4c bad-start c3774b43266ebd4d9115e17dec8743a2868eb5f4 Commit 71cde7b5cef52a234c25459b7d18285ef44516d2 fixed the problem. > (and thanks for the encouragement) No problem. You're welcome. :)
Hmm... unbelievable... the commit looks completely unrelated to the glitch... Could you double-check that before this commit ("git reset --hard 71cde7b^") the bug is present and after it ("git reset --hard 71cde7b") it's not? (Just to check that you've not mistaken during bisection)
I've manually checked both commits out, builded them, restarted the machine and started the demo app. Like: git checkout 595ee7e74a62cfadfc7e851a155d00c70bc45eeb. (I've tried it first with git-bisect, maybe I should have answer the question inverted (good-bad), but there I didnt get usefull results. So I manually bisected, through my results on the last comment. So I'm really sure this commit fixed my problem).
Hello, I have the exact same problem (strange lines in menu bars, i'll attach a screenshot). My setup is : OpenSuse 12.3 oxygen-gtk2 : 1.3.4 KDE : 4.10.5 GTK2 : 2.24.18 Cairo : 1.12.8 I have no idea on how to find out where is the problem. But if you want more info or some tests ; i'm in.
Created attachment 81550 [details] Same glitch on latest oxygen-gtk
@brancaleone What video card are you using?
Well i'm using a laptop with 2 cards : one ATI and one Intel. I use the ATI proprietary driver so i'm able to switch between them, but i'm mostly using the Intel display (power saving). That made me think to test it using the ATI display and it doesn't show the glitches. So the problem only appear when i'm using the Intel display. Version : ATI catalyst : 13.4 xf86-video-intel : 2.20.19 Models : ATI : Radeon 7600M Intel : Intel(R) Sandybridge Mobile (GT2+) (?)
Thanks. So, as I thought, it looks like a bug in intel driver. I'd suggest you reporting this to Xorg. On my side, I wasn't able to reproduce this with i915 (maybe it appears only on SandyBridge devices).
Any news on this problem? I'm experiencing the same glitch in Debian Jessie (testing) with KDE 4.14.2 and oxygen-gtk2 version 1.4.6. I have an Intel graphics card with the i915 driver, too. My laptop is a SandyBridge device, so maybe Ruslan Kabatsayev is right about only these models being affected. Has anybody reported the bug to Xorg yet?
I think no one has. Usually reporters give the links to upstream bug reports if then do report them.
yeah, i kind of gave up on this, because it was only affecting apps that i don't use regularly. However, i upgraded to OpenSuse 13.2 a couple of week ago, and the bug is now gone (it was still present in 13.1). Same hardware of course. oxygen-gtk2 : 1.4.6 KDE : 4.11.14 xorg server : 1.16.1 xf86-video-intel : 2.99.916
All right, thanks for the information. I'm trying to find more people who are affected by this bug (see my post on the KDE forums: https://forum.kde.org/viewtopic.php?f=17&t=123852&p=325192#p325192) to see if it is really the Intel driver. I will also try compiling older versions of oxygen-gtk as soon as I have a bit of time.
*** Bug 341498 has been marked as a duplicate of this bug. ***
I compiled and installed oxygen-gtk2 version 1.4.5 three days ago and haven't seen the glitch since. Before, with version 1.4.6 installed from the Debian repos, the vertical stripes appeared and disappeared within hours for no apparent reason. So it seems that version 1.4.5 is not affected, while 1.4.6 is.
(In reply to Gita Benadi from comment #25) > I compiled and installed oxygen-gtk2 version 1.4.5 three days ago and > haven't seen the glitch since. Before, with version 1.4.6 installed from the > Debian repos, the vertical stripes appeared and disappeared within hours for > no apparent reason. So it seems that version 1.4.5 is not affected, while > 1.4.6 is. Could you also try to compile 1.4.6 (instead of using Debian repos) ? As far as I know there is no change in oxygen code between 4.5 and 4.6 that would explain the bug
(In reply to Hugo Pereira Da Costa from comment #26) > Could you also try to compile 1.4.6 (instead of using Debian repos) ? > As far as I know there is no change in oxygen code between 4.5 and 4.6 that > would explain the bug I've done that now. So far, there are no vertical stripes in my gtk apps. I'll use this version for a couple of days and report whether or not the glitch reappears.
Today the stripes reappeared. So I uninstalled version 1.4.6 and installed 1.4.5 again. But the stripes are still there! Apparently, 1.4.5 is affected, too, but the glitch is not always visible. I'll try to find a pattern in its appearances and disappearances, but so far it seems totally random to me.
I'm having the same issue on Debian 8 Jessie with KDE. I have a Sandy Bridge notebook with a Intel HD 4000 Graphics.
Debian 8 user here: upgrading xserver-xorg-video-intel to jessie-backports solved this problem.