=== Guest16260 is now known as felipe_ | ||
* yofel is fairly certain now that kubuntu_42_fix_icon_themes.diff in qt4-x11 is what causes those BadWindows | 00:06 | |
yofel | more debugging tomorrow | 00:06 |
---|---|---|
yofel | ok, one more thing before I go to bed: The X errors are caused by kile and have been there already, but that patch seems to pull KDE apps into a gtk context where those errors are fatal, in kde they're not unless KDE_FATAL_X_ERROR is set | 00:49 |
apachelogger | yofel: the world is cruel like that | 01:06 |
apachelogger | yofel: also there appeared a gdk error thread on kde-core-devel today | 01:07 |
apachelogger | will recover kubotu tomorrow | 01:07 |
apachelogger | ENOKEYS | 01:07 |
apachelogger | yofel: upstream reports are to like 95% related to ubuntu | 01:09 |
apachelogger | (the other 5% are listed as source compiled, which of course does not say much about qt...) | 01:09 |
apachelogger | so I'll pretty much agree that it is a patch that is causing it | 01:10 |
apachelogger | -> bed | 01:10 |
smartboyhw | The first PM I received today is a stranger directing foul languages at me:O | 02:04 |
manchicken | JontheEchidna: Hey man, I'm on. Are you on? | 03:35 |
JontheEchidna | manchicken: for a little bit, yup | 03:35 |
manchicken | JontheEchidna: So I added some debug code into what I believe dbus is calling from the libqapt | 03:35 |
manchicken | bool WorkerDaemon::writeFileToDisk(const QString &contents, const QString &path) | 03:36 |
manchicken | That one. | 03:36 |
JontheEchidna | uh huh | 03:36 |
manchicken | I added some qDebug() prints and they're not coming up, so I'm thinking that the dbus calls aren't happening. | 03:36 |
manchicken | Unless dbus calls end up in a different place :) | 03:37 |
JontheEchidna | That is indeed what should end up being called | 03:37 |
manchicken | I didn't think linking kdebug stuff in was worth the trouble since you didn't have it linking in. | 03:37 |
JontheEchidna | heh, yeah | 03:37 |
manchicken | So I'm wondering if you have a moment to help me troubleshoot this. | 03:38 |
JontheEchidna | I do | 03:38 |
JontheEchidna | I'm wondering if qaptworker2 ever gets started | 03:38 |
manchicken | How do I test for that? | 03:38 |
JontheEchidna | top maybe? It should come up in the process list when you call a function in libqapt that requires the worker daemon | 03:39 |
JontheEchidna | dbus starts it automatically when the dbus call is made | 03:39 |
manchicken | Let me put htop up and see if I can watch it in my chroot while it's running. | 03:40 |
manchicken | I think it's very possible that I'm just missing something in my chroot. | 03:41 |
JontheEchidna | probably. the issue you're having seems to be a very... fundamental error, heh | 03:41 |
manchicken | Yeah... so I htop while the program is running and I don't see qaptworker2 going at all. | 03:42 |
manchicken | Do you know which package it's in? | 03:43 |
JontheEchidna | manchicken: libqapt2-runtime | 03:44 |
JontheEchidna | manchicken: though it should be installed if you're running libqapt git | 03:44 |
manchicken | Okay, I do have it installed, but it's not running. | 03:44 |
JontheEchidna | It could be that the dbus service file isn't being installed in a place where the chroot's dbus can see it | 03:45 |
manchicken | I think I saw a build warning about that, let me rebuild libqapt and see if that pops out. | 03:45 |
JontheEchidna | org.kubuntu.qaptworker2.service is the file | 03:46 |
manchicken | So dbus is sort of an RPC service then? | 03:48 |
JontheEchidna | yes, inter-process too | 03:48 |
JontheEchidna | like dcop back in the day | 03:48 |
manchicken | WARNING: Installation prefix does not match PolicyKit install prefixes. You probably will need to move files installed in POLICY_FILES_INSTALL_DIR and by dbus_add_activation_system_service to the /usr prefix | 03:48 |
manchicken | That's what I get when I `cmake .` in the libapt source root. | 03:49 |
manchicken | The toplevel of the git repo | 03:49 |
JontheEchidna | it seems that that's causing problems | 03:50 |
manchicken | -- Up-to-date: /usr/local/share/dbus-1/system-services/org.kubuntu.qaptworker2.service | 03:50 |
manchicken | So I think this may be a polkit issue then/ | 03:50 |
JontheEchidna | On a live system it's in /usr/share/dbus-1/system-services/org.kubuntu.qaptworker2.service | 03:51 |
manchicken | Ah, so I need to get cmake not to use /usr/local then | 03:51 |
JontheEchidna | I live life on the edge and do qapt/muon development against my live system :P | 03:51 |
JontheEchidna | I always pass -DCMAKE_INSTALL_PREFIX=/usr to cmake at the top level | 03:53 |
manchicken | Yeah, it's going now. | 03:53 |
manchicken | That is living on the edge. | 03:53 |
JontheEchidna | it's fairly safe. I can always fall back to apt-get if things get too horribly broken in Muon Land at any given time | 03:54 |
manchicken | That's true. | 03:54 |
manchicken | I usually use aptitude anyway. | 03:55 |
ScottK | manchicken: You should use apt-get. Things have changed while you were away and the apt resolver is generally considered better these days. | 03:56 |
JontheEchidna | Did they ever get MultiArch support sorted for aptitude? | 03:57 |
JontheEchidna | I assume they must have at least rudimentarily, since it's Debian's default package manager | 03:58 |
manchicken | JontheEchidna: Fixed the prefix, killed polkit, restarted it and then ran again. Still won't write the file. | 04:00 |
JontheEchidna | :( | 04:00 |
JontheEchidna | manchicken: I have to shuffle off to bed now; sorta getting up early tomorrow | 04:01 |
JontheEchidna | hit me up tomorrow though if you're still having issues. | 04:02 |
manchicken | JontheEchidna: No worries. I'll keep plugging away at it, I'm gonna be headed to bed in an hour or two myself. | 04:02 |
manchicken | QDBusConnection: session D-Bus connection created before QCoreApplication. Application may misbehave. | 04:03 |
manchicken | I get that when I start /usr/lib/kde4/libexec/polkit-kde-authentication-agent-1 | 04:03 |
manchicken | I'm gonna google | 04:04 |
manchicken | I'm wondering if I need to start qdbus in some way before I fire this off. I wonder if my chroot not having a full session is causing this. | 04:06 |
manchicken | shadeslayer_: You around? | 04:14 |
manchicken | That's an interesting bug... when apt-get asks if you want to continue using fr localization, it says [O|n] instead of [Y|n]. If you type o for "oui," then it aborts. It wants 'y', even though it is acting in fr locale and even prompted for o/n. | 04:19 |
manchicken | Okay, taking a crack at lxc, I'll see if that makes life easier :) | 04:27 |
manchicken | JontheEchidna: I don't think this is it, I can't get this to play nice even in my normal box. | 04:50 |
manchicken | I'm gonna head to bed. This isn't moving anywhere fast... | 04:51 |
manchicken | Later. | 04:51 |
lordievader | Good morning. | 07:29 |
smartboyhw | Riddell, ping | 10:07 |
BluesKaj | Hiyas all | 10:24 |
soee | good morning | 10:51 |
BluesKaj | 'morning soee | 10:52 |
manchicken | Anybody got an idea why a qDebug() call that I added to libqapt for debugging wouldn't be printing from the calling program? I'm wondering if there's something akin to kdebugdialog involved as well. | 12:22 |
manchicken | JontheEchidna: This whole thing was totally my fault. | 13:13 |
=== Guest30692 is now known as skfini | ||
tester56 | yofel: Adding "Option "TripleBuffer" "True"" and "Option "CoolBits" "1" " to /etc/X11/xorg.conf reduces Cpu consupmtion massively for me | 16:49 |
tester56 | (under "Section "Device"") | 16:50 |
tester56 | if feel like trying out, I would be very happy if you could report back if it helps for you too! | 16:51 |
tester56 | yofel: Cpu consumption of kwin, as we talked about lately | 16:54 |
mikecb | tester56: what driver are you running? | 16:55 |
tester56 | nvidia-325 | 16:55 |
tester56 | mikecb: are you experiencing the high cpu consumption too? | 16:56 |
mikecb | kwin runs consistently at a few percent, but nothing to make the fans go crazy | 16:57 |
mikecb | also 325.08 | 16:57 |
tester56 | mikecb: but much more than it used to be in kde 4.10 ? | 16:58 |
tester56 | mikecb: what vsync option are you using? | 16:58 |
mikecb | default. I haven't added anything. | 16:58 |
tester56 | okay that should mean Automatic | 16:58 |
tester56 | mikecb, but can you confirm that there is higher cpu consumption than in kde 4.10? | 16:59 |
mikecb | I don't know. I never really looked. | 17:00 |
tester56 | mikecb: play a video, do nothing else and watch top kwin cpu usage | 17:01 |
mikecb | jumps to 30% on an old avi in dragon | 17:01 |
tester56 | mikecb: that means high | 17:02 |
mikecb | yup | 17:02 |
mikecb | want me to try the tripbuf and coolbits and see if it does anything? | 17:02 |
tester56 | would be very nice as I am in talk with a kwin developer | 17:03 |
tester56 | remember to restart x :D | 17:03 |
mikecb | yup | 17:03 |
mikecb | brb | 17:05 |
tester56 | now try to watch the video again ... in my case kwin dropped to a third or less | 17:08 |
mikecb | yeah, now between 5-6% | 17:09 |
mikecb | that's excellent | 17:09 |
tester56 | thing is the developer i am in talk with does not yet know the reason of the problem | 17:10 |
tester56 | but it is nice for me to see the issue confirmed by someone else ... | 17:10 |
tester56 | mikecb: thank you very much! | 17:11 |
mikecb | yw. let me know if you need any logs or anything. | 17:11 |
tester56 | mikecb: are you a member of the kde mailing list? | 17:15 |
mikecb | nope | 17:15 |
=== Quintasan_ is now known as Quintasan | ||
yofel | tester56: checking (I only have 319 though) | 17:35 |
tester56 | yofel: that should not matter, I had 319 until yesterday too | 17:36 |
yofel | tester56: TripleBuffer seems to help, CoolBits has no noticible effect | 17:50 |
tester56 | yofel: is it significantly better for you? | 17:50 |
tester56 | yofel: thank you very much for testing btw. | 17:51 |
yofel | well, it is better in the sense that disabling VSync doesn't seem to change much now | 17:51 |
yofel | it's hard to measure a ~5% difference | 17:52 |
tester56 | have you tried watching a video in idle? | 17:52 |
tester56 | and comparing cpu usage? | 17:52 |
tester56 | there is a second issue though: if you restart kwin, watch a video you will notice kwin cpu usage raising if you have vsync to reuse or full scene repaint | 17:53 |
tester56 | it will raise after a minute or so (sometimes even faster, despite doing nothin else but watching) | 17:54 |
yofel | I have no idea what it was before, but now my kwin cpu usage while watching a movie is ~20% | 17:54 |
yofel | which is a bit high indeed :/ | 17:54 |
mikecb | yofel: mine went from 30% without the options to 5% with | 17:55 |
tester56 | yofel: yeah ... what vsync option? | 17:55 |
yofel | well, here it's using ~20% with "Automatic" and "None", so that's not it | 17:56 |
tester56 | yofel: could you also check if it is raising for you during watching a video instantly after a kwin restart (you need vblank reuse or full scene repaint)? | 17:56 |
yofel | kwin restart? i.e. killing it? | 17:57 |
tester56 | kwin --replace & | 17:57 |
yofel | hm, now it's ~11% | 17:58 |
tester56 | yeah and raises after some time | 17:59 |
tester56 | does it? (it shoudl remain constant then) | 18:00 |
yofel | uh, it stayed at 11 for a while, now jumped to 18 and is staying there again | 18:03 |
yofel | this is with vblank reuse | 18:03 |
tester56 | yeah ... i am seeing the same issue | 18:04 |
yofel | tester56: hm, this is not constant actually. it went from 11 to 18, then to 16, stayed for a while on 24, now it's on 11 again. | 18:10 |
yofel | so I don't think it actually rises because of the video | 18:10 |
tester56 | are doing anything else during watching the video? | 18:10 |
tester56 | even moving the cursor has an impact | 18:10 |
tester56 | in my case it rises because of the video, as there is nothing else on my desktop and I am not even moving the cursor, just watching video and top | 18:11 |
yofel | well, I think the quassel notification when you pinged me had an effect | 18:11 |
yofel | otherwise no | 18:11 |
yofel | I now fetched my EeePC so I can write somewhere else ^^ | 18:12 |
yofel | ok, now it's again on 18 - for no obvious reason | 18:12 |
tester56 | restarting kwin should show you the minimum again | 18:13 |
yofel | 11 | 18:13 |
=== claydoh_ is now known as claydoh |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!