=== eeejay is now known as eeejay_is_afk [04:15] Good morning [04:16] good morning pitti! [04:17] Hey pitti! [04:18] the meeting point of US, AU, and EU! [04:18] kenvandine: still busy catching up on email? :-) [04:19] yup :) [04:19] hey RAOF, how are you? did you see that rodrigo updated colord in ubuntu? we needed to get the stuff in for b2, so we couldn't wait for a sync from Debian; but of course we should sync again once 0.1.12 is in debian [04:19] i am actually in california for a couple days [04:19] pitti: Yeah, I did. [04:19] so a few hours further back in time :) [04:20] pitti: I didn't realise that it was wanted for b2, otherwise I would have uploaded it myself. [04:20] pitti: can't sync, bad version [04:21] argh [04:21] * micahg thinks lintian should barf on that, but hasn't gotten around to fixing it [04:22] i wish it would [04:22] * kenvandine makes that mistake too often :/ [04:23] bug 782308 if anyone is feeling adventerous [04:23] Launchpad bug 782308 in lintian "Lintian should warn about if there is a -XubuntuY without a corresponding Debian entry" [Wishlist,New] https://launchpad.net/bugs/782308 [04:32] pitti: good morning [04:33] hey jbicha, how are you? [04:35] pitti: good, I had to set my dput to sftp so that big uploads work [04:35] could you add nautilus to the desktop set? [04:37] jbicha: I can't (at least not permanently), I'm afraid; I'm afraid you need to email cjwatson to ask for this [04:38] jbicha: I'll sponsor it; however, I don't think it'll land in b2 [04:38] jbicha: uploaded [04:38] ok, thanks! [04:38] * micahg wonders why nautilus wouldn't be in the desktop set [04:39] pitti: Do you have any insight into PolicyKit development? I've got http://lists.freedesktop.org/archives/polkit-devel/2011-September/000353.html which will enable us to switch polkit support back on in colord. [04:40] RAOF: probably best to directly discuss that with davidz (he's Boston time) [04:40] RAOF: he probably will prefer a bz.fd.o bug for it [04:42] smspillaz, if I get a XConfigureRequest, should it always be responded to with a XConfigureWindow? [04:43] pitti: And davidz would be in what IRC channels? :). I'll also go file a bug. [04:44] RAOF: I usually catch him in #udev on freenode [04:44] but he's usually pretty good in seeing bz mail [04:46] RAOF: when you package colord for Debian, can you please add the missing liblcms2-dev dependency to libcolord-dev? [04:46] Oooh, certainly. [04:46] we got a control-center build failure due to that [04:49] RAOF: actually, seems the -dev package is missing a whole lot more [04:49] In fact, I'll add all the requires there. [04:49] heh, snap :) [04:49] dbus, gobject, etc. [04:49] Yeah. [04:50] * debian/control: Add missing liblcms2-dev, libdbus-1-dev, libglib2.0-dev [04:50] dependencies to libcolord-dev, to fix FTBFS of reverse dependencies. [04:50] something like that [04:50] Yup. [04:51] Done [04:52] RAOF, any idea why onboard would send three configure requests? http://paste.ubuntu.com/693515/ [04:53] Animation? [04:54] Not really, no. [04:54] RAOF, it wants to be normal sized, then requests two tiny windows [04:55] Does it do that all at once? [04:56] RAOF, yes, before the map request [04:56] No idea. [04:57] RAOF, what's the correct response as a window manager - do XConfigureWindow each time? [04:57] I don't know sorry. [04:57] Sam would know, obviously, as might DBO :) [04:59] RAOF, perhaps you know this one. How does focus work in X? Do the windows request focus and the window manager gives it to them? It appears the default behaviour is the window under the cursor has focus [05:00] That's my understanding of the default focus mode, yes. Well, ish. [05:00] robert_ancell: http://standards.freedesktop.org/wm-spec/wm-spec-latest.html#id2550738 [05:01] aha [05:01] cheers broder [05:02] robert_ancell: when in doubt, guess "ClientMessage to the root window with what you want as the message type" :-P [05:02] Although clients can just XSetInputFocus if they so desire. [05:03] Now I think I understand why wm developers mutter about focus stealing so much... [05:03] RAOF: on a totally unrelated note, do you know if anybody's modified xtrace or written something like it that can filter on certain extensions or do any filtering at all? [05:04] broder: What do you mean by filtering? xtrace itself can filter out extensions to the client running under it. [05:04] RAOF: no, i want it to pass everything through, but i only want it to print out, say, RandR messages [05:04] Oh. I just pass everything through grep :) [05:05] ugh :) [05:05] i guess the messages are all confined to one line so it's not that bad [05:28] robert_ancell: it depends [05:29] robert_ancell: if you choose to do nothing, you must send a synethetic ConfigureNotify to the client window with the same position it alrady had [05:29] otherwise you're allowed to do whatever you want with it [05:30] smspillaz, right, that's what I figured. Any idea why onboard would send these bogus 1x1 requests? [05:30] robert_ancell: it may be for a toplevel unmapped window / inputonly window and not actually for the keyboard window itself [05:31] good morning [05:31] smspillaz, the requests are all for the same window, but it does seem to produce a panel style window if you kill the main window. So perhaps it's testing if the window manager will let it hide? [05:31] bonjour didrocks, ca va? [05:32] pitti: guten morgen pitti. I'm ok, thanks! you? [05:32] didrocks: bit tired still, was a rather short night, but ok [05:32] didrocks: pondering what to do with bug 779382 [05:32] Launchpad bug 779382 in unity-2d "update-notifier not visible under unity" [Critical,Confirmed] https://launchpad.net/bugs/779382 [05:33] didrocks: I think we should just whitelist u-notifier, but that doesn't seem to work with u-2d [05:33] didrocks: what's the process for raising DX' awareness to fix the whitelist handling for -2d? [05:33] * pitti can't remember whether it's subscription, assignment, upstream task, milestone, etc. [05:34] pitti: yeah, the whitelist is different in unity-2d and unity-3d… [05:34] let me look at the bug first [05:34] didrocks: ah, so it's just a matter of setting the whitelist in two different places? that'd be easy then [05:34] pitti: isn't what design wants to kill (update-notifier icon) and just let update-manager appearing? [05:35] didrocks: yes, that's the default now, but people may have configured it to show the icon instead [05:35] hum, weird, the schema changed, it should be the same [05:36] the code is different, but it's looking at the same key now [05:37] pitti: do we really want to support that? Last time I wanted to add something to the systray, it was a long discussion with mpt who didn't want to and let people edit the whitelist [05:37] otherwise, the whitelist would be "All" (like all posts on OMGUbuntu are doing) [05:37] * didrocks still thinks that this whitelist is useless [05:37] didrocks: well, update-notifier is a bit special here, as it regressed only recently (after beta-1) for unity-2d [05:38] before that it apparently appeared in 2d's systra [05:38] y [05:38] pitti: indeed, because there was no whitelist for unity-2d [05:38] but it was already blacklisted in unity-3d for a cycle [05:38] so this was a feature/ui change [05:38] right, for -3d [05:38] slangasek's concern is primarily for 2d, AFAIUI [05:38] robert_ancell: perhaps [05:38] it's just making -2d the same as 3d [05:39] why the concern is only on 2d and not 3d? [05:39] robert_ancell: I mean, the default behaviour compiz adopts is to just do what the client asks for as long as the window is managed really [05:39] and as it changed behaviour after FF/UIF, and thus stops people who got used to the icon from upgrading from beta to final, we should at least whiltelist that one [05:39] smspillaz, it's weird then, I wonder what it does with onboard to avoid being 1x1 [05:39] I'm concerned about both - but for beta-2 I'm concerned about unity-2d in particular because up to now, unity-2d users have been getting their notifications of updates without interruption [05:39] if we break that, some users may not notice they're running insecure systems [05:40] it's also in the session indicator now, but of course you need to know about this [05:40] robert_ancell: it probably resizes itself back [05:41] session indicator> heh, yes, I would never have thought to look there [05:41] anyway, I think we should be coherent between both [05:42] that is, enabling in 3d and 2d if we are going to enable it [05:42] so let's whitelist it in both then? [05:42] sounds good to me [05:42] normally, it's the wmclass which is used [05:42] so, we should have a try [05:42] didrocks: can we JFDI this without a huge process? I'll take the bullets, you can blame me :) [05:42] pitti: will do that, and I was hoping today would be quiet :) [05:42] * pitti never really believed in deliberatly breaking our own software even [05:43] didrocks: well, I don't think it's a beta-2 blocker, we can't wait for that anyway [05:43] pitti: yeah, I spent a faire amount of time with design to discuss that the whitelist is useless and dangerous [05:43] and now, softwares in ppa editing the whitelist in their postinst [05:44] sedding in the gsetting schemas… [05:44] * didrocks tries to get the wmclass of the window [05:44] pitti: I have targeted the bug to beta-2, because of the issue of unity-2d users upgrading to beta-2 and then having no idea why they're not getting updates... [05:44] Oh, my. [05:45] slangasek: TBH, we have the same issue with people who upgraded to natty then [05:45] not a blocker for images, but I do think it should be a blocker for beta-2 [05:45] slangasek: they set the value to on, but never saw any update with unithy [05:45] unity [05:45] slangasek: for upgrades from natty? ok [05:45] hence my suprise it's a blocker on a beta, but not on a previous stable release [05:45] didrocks: yes, and I think it's a major issue there as well, and one I would like to SRU for [05:45] if we can get an upload right now, it'd still be possible to squeeze in [05:45] didrocks: it's a blocker on beta because I didn't know about it for the previous stable release ;) [05:45] as I want to shuffle image builds to wait for the FTBFSed control-center [05:46] didrocks: is that hard to change? I didn't see an existing gsettings override file in unity [05:46] didrocks: but for the "rush" one, I guess -2d is enough [05:46] pitti: there is none, I edit upstream [05:46] (unless it's using the very same blacklist) [05:46] pitti: it's the same [05:46] so it needs an unity upload [05:47] and maybe a -2d if it doesn't work there [05:47] * didrocks is looking for the wmclass [05:47] didrocks: ah; but as this is a distro specific change until we port update-notifier to indicator, a dh_installgsettings override file might be easier? [05:47] pitti: all unity settings change are done upstream, we don't override our software until someone else ship unity [05:48] ok [05:48] pitti: so better to keep the upstream in sync as they are using jhbuild for them [05:48] is there a way to force update-notifier showing its icon? [05:49] (apart from the gconf key, but just as a test) [05:49] you mean if you have up to date software? [05:49] downgrade $randompkg? [05:49] yeah, ok, will downgrade [05:53] ah, found the gsettings corresponding key [05:53] great, I have notifications, but didn't see the reject for showing the icon [06:02] * didrocks sigh at the debugging message removed to find the wmclass of the icon (as clicking on unity gives compiz if you whitelist it) [06:05] ok, the WMCLASS is Update-notifier [06:08] works on 2d as well, phew [06:08] uploading unity [06:15] didrocks, FYI bug 854448 [06:15] Launchpad bug 854448 in unity "Skype is not in whitelist for Systray" [Undecided,New] https://launchpad.net/bugs/854448 [06:15] jasoncwarner_ just filed that [06:15] kenvandine: ? [06:15] jasoncwarner_ is aware about sni-qt isn't it? [06:15] we discussed that a lot [06:16] oh yeah... we did talk about that before my vacation [06:16] it isn't working for some reason [06:16] jasoncwarner_, ^^^ [06:16] jasoncwarner_, hadn't you filed a bug about that before? [06:16] I also filed the weekly report with status on it… [06:16] hehe [06:17] I thought I did [06:17] i know you did... now i remember that [06:17] and there is the rationale in the changelog when I removed it from the systray [06:17] didrocks: so, skype should work without the whitelist? because it was not on a fresh install today [06:18] which is why I thought I needed to file the bug....even with the fix to sni-qt [06:18] jasoncwarner_: it should, you should have sni-qt [06:18] jasoncwarner_: is sni-qt installed? [06:18] jasoncwarner_: just tried skype here, and I have the icon with sni-qt showing it [06:19] didrocks: yes, it is installed... [06:19] didrocks: trying something [06:19] It should be shown as an indicator [06:20] with an additional "Activate" menu that we had manually to some applications as you can't just right click on them [06:20] didrocks: just removed it from my whitelist....going to check...one sec [06:20] you need 32bit version of sni-qt [06:20] looking at bug 838470 [06:20] Launchpad bug 838470 in sni-qt "Skype not showing in indicator" [Medium,Fix released] https://launchpad.net/bugs/838470 [06:21] jasoncwarner_: you need to restart unity to get it worked [06:21] do you have that? [06:21] yeah, we discussed about the issue where it didn't work because of multi-arch [06:21] and we fixed it for you and ping you jasoncwarner_ with agateau :) [06:22] * slangasek hugs didrocks - thanks for the unity upload! [06:22] * didrocks hugs slangasek back, you're welcome [06:22] * slangasek installs sni-qt:i386 here for good measure [06:23] didrocks: yay you, thanks [06:24] slangasek: should be correctly transitionned, I didn't test it here but agateau did [06:24] pitti, if you are waiting on something for images... it would be nice to get my ido upload in [06:24] (to multiarch) [06:25] it fixes the unclickable calendar in indicator-datetime [06:25] pitti, and is very low risk imo [06:25] kenvandine: *nod* [06:25] thx [06:25] didrocks: yeah, I now have both sni-qt and sni-qt:i386 installed, looks good :) [06:26] didrocks: seems sni-qt isn't working...restarted and it isn't there. I must have had it in my whitelist in my other install...this is a fresh install today so testing a whole bunch of stuff ;) [06:26] jasoncwarner_: amd64? [06:26] didrocks: yeah [06:26] didrocks: I did install skype:i386 this time ;) [06:26] slangasek: is there anything particular to do in the seed? to install the i386 version? ^ [06:27] didrocks: you can't seed packages from a different architecture; shouldn't we ask for the skype:i386 package to recommend/depend on it? [06:28] slangasek: it would be the case for a lot of Qt application (sni-qt is for all Qt applications showing something in the systray) [06:28] like mumble and such [06:28] right, but there are fairly few i386 binary-only qt apps that I know of [06:28] but regardless, there really is no way to seed it [06:29] (and certainly no space for it...) [06:29] slangasek: sni-qt is already seeded by default [06:29] it seems the issue is only for amd64 application with 32 bits only [06:29] I mean there's no way to seed the foreign-arch one [06:29] ok, so right now, the skype recommends seems better [06:29] ... which would require pulling a second copy of qt onto the CDs, which is definitely not possible [06:30] ah, indeed, even it's a plugin, it will pull the 32 bits version of Qt [06:31] slangasek: however, that won't fix the situation when people install skype from the skype website [06:31] jasoncwarner_: you downloaded skype from the partner repo? [06:32] didrocks: well, except skype is a partner package... provided by skype... so whatever they put in one place, they'll put in the other :) [06:32] didrocks: I enabled the source in S-C and then I apt-get installed skype:i386 [06:32] installing via S-C didn't get me the correct version (which is something else I have to track down) [06:32] tremolux and MVO ^^ I was going to talk to you guys about that... [06:33] slangasek: ah ok, do you know about who is in charge of asking them/us doing, this tweak? [06:33] didrocks: it also means we don't have access to directly modify the skype:i386 package anyway. Could you talk to Brian Thomason about this? He's the contact [06:33] jasoncwarner_: each issue after another? :) [06:33] slangasek: excellent, will do, thanks! [06:33] jasoncwarner_: so, at least, we have the explanation with sni-qt, did reinstalling the 32 bits version of it fixed it for you? [06:33] didrocks: I'm like the perfect user tester ;) I keep breaking my machine...mostly id10t style ;) [06:34] jasoncwarner_: stop breaking things! it's 8:40 and I still didn't get any coffee :) [06:35] jasoncwarner_, libido got accepted, so that fix will be coming soon [06:36] my body still feels like it is in europe... but i think i need to force myself to sleep soon [06:49] good morning === tkamppeter__ is now known as tkamppeter [06:50] i have add some comment on the cogl/clutter update on the pad [06:50] didrocks, pitti ^ is this worth the trouble to get it into beta2? [06:50] Morning, all! [06:51] good morning Sweetshark [06:51] hey Sweetshark [06:52] ricotz: ABI break should certainly be done after b2 at this point [06:52] hey ricotz, Sweetshark [06:52] pitti, yeah, i am thinking the same [06:52] didrocks, hi [06:58] pitti: 3.4.3-2 is blocked at debian because of debian bug 642176 (caused by dokos gcc update there). But I would fix ubuntu bugs #835153 and ubuntu bug #799535, so should we still release that merged as 3.4.3-1ubuntu2 for beta2? [06:58] Debian bug 642176 in g++-4.6 "g++-4.6: libreoffice/mozilla build fails with gcc >= 4.6.1-10 due to gcc PR 50442" [Important,Open] http://bugs.debian.org/642176 [06:58] Ubuntu bug 835153 in libreoffice "oosplash.bin crashed with SIGSEGV in XSetForeground()" [High,In progress] https://launchpad.net/bugs/835153 [06:58] Ubuntu bug 799535 in libreoffice "package libreoffice-common (not installed) failed to install/upgrade: tentata sovrascrittura di "/usr/share/mime/packages/openoffice.org.xml" presente anche nel pacchetto openoffice.org-debian-menus 3.3-9556" [Medium,Confirmed] https://launchpad.net/bugs/799535 [06:59] Sweetshark: too late for b2 [06:59] pitti: ah, ok. [06:59] but we can upload it afterwards of course [07:02] didrocks, hey, do you know much about indicators? I'm wondering if it's safe to load them in a separate thread, because it's behaving really badly in unity-greeter [07:03] robert_ancell: hey, how badly? FYI, in unity, we have it in a service (unity-panel-service) which communicates to unity through dbus [07:03] rodrigo_: good morning [07:04] rodrigo_: FYI, the gnome-control-center FTBFS, I retried it after I fixed libcolord-dev depenendies; armel failed again, need another publisher; but it's being handled, so you can ignore it [07:04] ah, at least, oneconf syncing is working on staging now :-) [07:04] pitti: btw, reminder about the team meeting report [07:05] argh, that, too [07:05] * pitti doing 5 things at the same time [07:05] didrocks: cheers [07:05] didrocks, getting weird crashes [07:05] didrocks, using new Indicator.Object.from_file (filename) and loading them all in parallel [07:05] but they're not put into the menubar in the UI thread from then on [07:05] rather they are put into the menubar in the UI thread from then on [07:06] robert_ancell: I guess definitively threading it will be nicer, but see with tedg [07:06] robert_ancell: the only way we got some stability in natty is to put them in a separate process (it was in the same process in maverick and was the culpurit of most UNE crashers) [07:07] didrocks, is each one in it's own process? [07:08] robert_ancell: no, there are all in the same process (unity-panel-service), [07:08] didrocks, so they are loaded sequentially in one process? [07:08] that's my next plan of attach, to load them like that [07:08] attack [07:08] late :) [07:09] robert_ancell: no, they are all loaded one after another. First indicator_object_new_from_file() and then cycling on the libraries resulting from that call [07:09] didrocks, cycling on the libraries? [07:10] basically something like that (from memory): [07:10] while (((name = g_dir_read_name (dir)) != NULL) && g_str_has_suffix (name, ".so")) [07:11] path = g_build_filename (INDICATORDIR, name, NULL); [07:11] … [07:11] and using indicator_object_new_from_file on that [07:12] oh, ok [07:12] they take *ages* to load. They really should be loaded in parallel [07:12] robert_ancell: indeed, you can see that when unity is starting (all indicators appearing one after another) [07:12] robert_ancell: I guess we can try that as we are protected in a separate process [07:13] robert_ancell: but in unity-greeter, if you are in the same process, try to protect from crash… :/ [07:13] didrocks, I'll get back to you, I think it may be broken [07:13] didrocks, so, how do you pass the objects over d-bus? [07:13] then I'll fork it in unity-greeter which will be much safer [07:13] robert_ancell: that's more than possible, we discovered a lot of issues with indicators, even if normally the backends are in separate processes (all the indicator*services) and so should be easily multithreadable [07:14] robert_ancell: dbusmenu [07:24] apt-get update just informed me that one signature was invalid. How do I check which repository it belongs to? [07:25] BigWhale: it seems like there is a server issue currently [07:25] never mind [07:26] with the recent crack into kernel.org and such, I got sort of paranoid :> [07:27] pitti: do you know if we ever had a openoffice.org-debian-menus package? I suspect it has always been from a third party. bug #799535 [07:27] Launchpad bug 799535 in libreoffice "package libreoffice-common (not installed) failed to install/upgrade: tentata sovrascrittura di "/usr/share/mime/packages/openoffice.org.xml" presente anche nel pacchetto openoffice.org-debian-menus 3.3-9556" [Medium,Confirmed] https://launchpad.net/bugs/799535 [07:28] BigWhale: what was the error message you got? [07:28] mvo: W: GPG error: http://archive.ubuntu.com oneiric Release: The following signatures were invalid: BADSIG 40976EAF437D05B5 Ubuntu Archive Automatic Signing Key [07:30] BigWhale: thanks! I start some investigation on this now [07:30] mvo: you're welcome. and thank you. :) [07:30] I wonder if more people are affected, I get this too on two machines [07:38] Sweetshark: to check: https://launchpad.net/ubuntu/+source/openoffice.org-debian-menus/+changelog [07:38] Sweetshark: I can't reach LP or the DC right now [07:38] mvo: I've seen someone in #ubuntu-de having it too [07:39] geser: thanks, its under investigation, server is out of sync, admins working on it [07:39] morning [07:39] didrocks, seems much more stable now [07:40] hey robert_ancell [07:40] Bah! *That's* why avahi backtraces don't get retraced properly? All their symbols are in avahi-dbg! [07:41] rodrigo_, hey [07:41] can anyone else reach launchpad or chinstrap? [07:41] * pitti toddles off for breakfast, perhaps it'll come back by then [07:44] pitti: that link doesnt exist, so it is third party right? [07:45] good morning everyone [07:46] pitti: LP is fine here [07:56] hey [07:57] salut seb128 [07:57] lut didrocks, ca va bien ? [07:57] seb128: on fait aller, et toi? [07:57] ca va bien je pense, je te dirais après un café ;-) [07:58] didrocks, oh, slangasek talked you about whitelisting update-notifier it seems ;-) [07:58] Sweetshark: right [07:58] bonjour seb128 [07:59] hey pitti, how are you? [08:00] seb128: bit hectic, but fine, thanks! how about yourself? [08:00] I'm fine thanks [08:00] colord/armel published now, giving back gnome-control-center again [08:00] pitti, is there anything I can help on? [08:01] hi seb128, pitti [08:01] seb128: currently fighting with DC/mirroring problems for image builds, not much there, I'm afraid [08:01] hey chrisccoulson [08:01] hey chrisccoulson, how are you? [08:01] i'm good thanks, how are you? [08:01] seb128: indeed :) [08:01] pitti, ok, just say the word if you need me for something [08:01] seb128: but if you could have a look at the ~ 4 c-desktop-team assigned bugs on http://status.qa.ubuntu.com/reports/ubuntu-desktop/index.html, and see if they should be assigned/upstreamed/etc, that'd be nice [08:02] pitti, ok [08:02] seb128: pitti: chrisccoulson: jasoncwarner_: seems the 3rd tarball of c-p-m was still wrong. Adding a distro-patch which explained why we got those artefacts to the ppa. [08:02] didrocks, yeah, DBO explained it last night [08:02] seb128: in particular, bug 827921 should be fixed now [08:02] Launchpad bug 827921 in nautilus "Pressing Ctrl-V puts path to a copied file into search box instead of actually running paste action" [High,Fix committed] https://launchpad.net/bugs/827921 [08:02] the version in the PPA made it pretty much unusable ;) [08:02] pitti, that one should be fixed in the nautilus update from yesterday [08:02] pitti, I will check and close it [08:03] chrisccoulson: he explained it to you? that would be nice that I got an email at least [08:03] seb128: I couldn't reproduce the bug with the old nautilus either, but yes, should be fixed [08:03] didrocks, yeah, DBO said the wrong trunk was used, he said the trunk trunk is the one we want, not the oneiric trunk [08:03] seb128: seems compiz process is totally broken, we got 6 bad tarballs in 4 days [08:03] didrocks, read the channel log from the evening,night [08:03] and lost numerous hours [08:03] didrocks, yeah, DBO said he would fix that [08:04] and I would have liked to not spent some time this morning to discover that again :/ [08:04] pitti, it was trivial to reproduce ctrl-C on a file, ctrl-V [08:04] well, anyway, let's build and push a fix [08:04] seb128: I tried in normal and list view, both worked [08:04] *shrug*, who knows [08:05] pitti, ok, weird, it opens a text entry on the bottom right with the filename there [08:05] rather than copying the file [08:05] a search box? [08:05] yes [08:05] yes, that's what the bug was about [08:06] but I wasn't able to reproduce [08:06] it does it every time there [08:06] well anyway, I will update now and test [08:06] i get that too [08:06] although, i have seen it work correctly on occasion, but i get the text entry most of the time [08:09] it seems a race in the events [08:10] * didrocks fixes the patch which, of course, doesn't apply [08:12] wow, i''ve only got this week left, then i'm on vacation for 2 weeks :) [08:12] that's probably the worst time for me to take vacation isn't it? ;) [08:13] pitti - do you happen to have any bootcharts from when startup time was quick (lucid?). i want to see where pulseaudio starts [08:15] chrisccoulson, http://people.canonical.com/~seb128/bootchart/ [08:15] chrisccoulson, I've some there [08:15] thanks === tseliot_ is now known as tseliot [08:16] chrisccoulson, http://people.canonical.com/~seb128/bootchart/seb-dellmini-lucid-20100125-3.png [08:16] should be about right [08:16] ok, it seems to start when everything else starts [08:16] i think i can probably make it do that again then [08:16] i wonder what changed there :/ [08:17] does that matter? [08:17] we block the whole session to load that at the moment [08:17] why do we block on it? [08:17] chrisccoulson, holiday> great, do you have everything in shape you needed done for oneiric? [08:17] seb128, the media-keys plugin connects to pulseaudio before the main loop runs [08:17] https://bugs.launchpad.net/ubuntu/+source/gnome-settings-daemon/+bug/854101/comments/4 [08:17] Ubuntu bug 854101 in ubuntu-boot-speed "gnome-settings-daemon takes waaaaaaaay too long to start" [Undecided,Confirmed] [08:18] chrisccoulson, do we need to cover for anything while you are not there? [08:18] urg [08:18] chrisccoulson: https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis has links to my three boot charts (lucid/maverick/oneiric) on the mini 10v [08:18] chrisccoulson: wrt. xrandr, didn't we fix that by not calling it at all unless the user has a ~/.config/monitors.xml? [08:19] chrisccoulson: nice finding wrt. pulseaudio! indeed the session doesn't need to block for that [08:19] seb128, i think most things should be ok. the firefox/thunderbird dailies are pretty much automatic. the only thing that really needs to keep up-to-date are the firefox-next/thunderbird-next PPA's, with the 8.0 beta releases [08:19] and we advertise those as pretty much officially supported :) [08:20] chrisccoulson: will there be firefox/tbird releases in the next two weeks which we need to package? [08:20] pitti - the 7.0 release is on 27th, but we'll actually get the final builds this week [08:21] there might not actually be a beta release until i get back, as the first ones tend to be 10-14 days after the last update [08:21] so that might work out ok [08:21] pitti, ok, so I should ignore the build failure of g-c-c, right? [08:22] rodrigo_: [08:22] rodrigo_: right [08:22] urgh, metacity just hung [08:22] pitti, what was the colord problem? it needed a newer version? [08:22] on a lock in ca_context_play [08:23] i've seen the same hang in both firefox and thunderbird in the last few days too :/ [08:23] rodrigo_: libcolord-dev was missing dependencies on liblcms2-dev, libglib2.0-dev, and a third one [08:23] oh ok [08:23] rodrigo: see Requires in /usr/lib/x86_64-linux-gnu/pkgconfig/colord.pc [08:23] ah, libdbus-1-dev [08:24] rodrigo_: RAOF committed that to debian, too [08:24] rodrigo_: btw, next time can you please use -0ubuntu1? you used -1ubuntu1, so we can't sync [08:24] oh, missed that, sorry [08:24] chrisccoulson, do you have tb contacts syncing working? [08:24] so, whenever we use a different version than debian, it's 0ubuntu?, right? [08:25] seb128, desktopcouch is still broken, so it doesn't even work in evolution yet [08:25] chrisccoulson, did you look at the tb messaging issue on pop? my friend still get spams added to the counter :-( [08:25] rodrigo_: right [08:25] seb128, yeah, i'm not sure what's going on there, and i don't have a POP account to test. with imap, we don't get new message notifications until the message filters have run [08:28] chrisccoulson, http://mail.google.com/support/bin/answer.py?answer=13273 [08:28] chrisccoulson, gmail does pop [08:28] chrisccoulson, get a gmail account! [08:28] oh, i didn't realize that [08:28] i've got a gmail account, but i use imap ;) [08:29] is anyone going to work on network-manager updates? [08:29] I suppose cyphermox will [08:29] rodrigo_, check with cyphermox when he's online [08:30] urgh, what on earth is going on? firefox, thunderbird, metacity, gnome-settings-daemon all spinning my CPU in the same call (ca_context_play)? [08:30] ok [08:30] my laptop feels like it's going to melt [08:30] rodrigo_, but it's getting late in the cycle [08:30] chrisccoulson, is that a canberra pulseaudio thing? [08:30] seb128, late for updating NM? [08:30] rodrigo_, yeah [08:30] ok, I'll check with cyphermox [08:30] seb128, yeah. i just noticed pulseaudio isn't even running here [08:31] chrisccoulson, could be the issue ;-) [08:31] everything is greyed out in the sound panel [08:31] if i start pulseaudio manually, then everything starts working [08:31] this has happened several times in the last few days :/ [08:32] open a bug against pulseaudio [08:32] let it run in debug somewhere maybe so you see if it goes down or something [08:33] David Henningsson might look at your bug if you open one ;-) [08:38] rodrigo_: there: https://launchpad.net/ubuntu/+source/gnome-control-center/1:3.1.92-0ubuntu1, all built [08:38] pitti, ok thanks! [08:39] pitti, " * Drop 05_correct_gpg_agent_error_code.patch, upstream now." in gnome-keyring [08:39] did you check? [08:40] seb128: I think so, did I look wrong? [08:40] some users said on bug #828756 that the bug is back with yesterday update [08:40] Launchpad bug 828756 in gnome-keyring "getting "connection is untrusted" warnings" [High,Fix released] https://launchpad.net/bugs/828756 [08:40] "The bug was solved by the gnome-keyring 3.1.91-0ubuntu4 upgrade about a [08:40] week ago, but now it's back with 3.1.92-0ubuntu1 :(" [08:40] pitti, but maybe it's a different issue [08:40] checking git [08:40] seb128: I have gotten untrusted certificate errors in empathy for weeks now [08:41] pitti, it was fixed with 3.1.91-0ubuntu4 for most users [08:41] * pitti checks, perhaps the patch didn't apply, and it only looked like it was fixed [08:41] pitti, btw confirmed that nautilus ctrl-V is fixed so I'm closing that one [08:42] https://bugzilla.gnome.org/show_bug.cgi?id=659080 says "fixed", too [08:42] Gnome bug 659080 in gpg-agent "gpg-agent: Bad error code in response to GETINFO" [Normal,Resolved: fixed] [08:42] pitti, ok, yeah, the user comments were misleading then [08:42] sorry about the noise [08:42] seb128: yes, confirmed, it's upstream [08:42] seems like "upstream issue" [08:42] this seems to be a telepathy-something bug [08:43] right [08:43] (and I get it, too) [08:43] I will check with kenvandine when he's online [08:45] * pitti goes to unbreak our Pango GIR [08:45] pitti, what is broken? [08:45] enums don't have type information, and TabAlign.LEFT is mis-spelled TabAlign.TAB_LEFT [08:45] it works fine in jhbuild [08:46] type information sounds like a similar issue to the out-of-tree GTK 3 breakage that we had [08:46] ok, git got some fixes for builddir != srcdir builds [08:46] ah, it's multibuild, too [08:46] but not sure if those fix your issues [08:46] pitti, vala 0.14.0 is still in the queue, right? [08:47] hm, do they ever do a new release? [08:47] rodrigo_: yes [08:47] ok [08:47] http://git.gnome.org/browse/pango/commit/?id=2c3683a35bb6212b4ba2e1f30bdbed38f076ef42 [08:47] aah [08:47] that's most probably the fix which causes gtimelog to not segfault any more [08:53] pitti, they = pango? I can ping behdad about that later if you want [08:53] seb128: that'd be helpful; they don't seem to align to GNOME milestones? [08:53] seb128: otherwise I'll just package a git snapshot, there's quite a lot of fixes that we want [08:53] pitti, pango is not "actively maintained" nowadays [08:54] well let's say it's mostly doing the job so it's in maintainance mode and behdad is working on harfbuzz [08:54] seb128: do you mind if I just do a make dist from git? [08:55] i guess i should probably take bug 854101 ;) [08:55] Launchpad bug 854101 in ubuntu-boot-speed "gnome-settings-daemon takes waaaaaaaay too long to start" [Undecided,Confirmed] https://launchpad.net/bugs/854101 [08:55] oh, i'm already assigned [08:55] hah :) [08:56] pitti, you can do a snapshot, no issue from me, I will ping behdad about a tarball anyway, that would still be nice to have [08:56] right [08:56] chrisccoulson_: yeah, I'm reviewing http://reports.qa.ubuntu.com/reports/ubuntu-desktop/oneiric.html and just assigned it to you 2 minutes ago ;-) [09:00] rodrigo_, could you review the patch from tkamppeter on bug #842768, the bug is beta2 milestoned there [09:00] Launchpad bug 842768 in system-config-printer "Cups notifies "printer ' xxx ' may be not connected " although printer is OK and printing is OK too" [Medium,Confirmed] https://launchpad.net/bugs/842768 [09:00] seb128, ok, looking [09:02] rodrigo_, it's probably not a beta2 blocker but seems like something to "fix" for Oneiric [09:02] ok [09:03] rodrigo_, if you have no opinion on it maybe get pedro to upstream the bug and see what upstream (marek?) says ;-) [09:03] I think it makes sense to use 'connecting...' instead of 'maybe connected' for sure [09:03] but yes, that would be for 3.4, so yes, will talk to marek [09:05] thanks [09:07] ugh, again gnome-online-accounts rejected, I always forget I have no permissions [09:09] rodrigo_, I can sponsor it for you, did you email cjwatson about it? [09:11] seb128, yes, iirc [09:11] let me check my mails [09:11] ok [09:11] rodrigo_, you pushed to the vcs? I can sponsor from it? [09:12] oh yes, he said it should be added to the desktop set automatically [09:12] seb128, yes, in the vcs, ready to sponsor [09:12] need to reboot, brb [09:12] ok [09:12] how "added automatically"? [09:12] it's a recommends from gnome-control-center? === chrisccoulson_ is now known as chrisccoulson [10:09] seb128: is this a known bug in the latest nautilus? http://people.canonical.com/~amilone/nautilus_bug.png [10:09] tseliot, dpkg -l | grep nautilus? [10:10] rodrigo_,seb128: added gnome-online-accounts to the ubuntu-desktop package set now [10:10] cjwatson, thanks [10:11] rodrigo_, i've sponsored the update already so no need to reupload ;-) [10:11] seb128: http://pastebin.ubuntu.com/693665/ [10:11] tseliot, upgrade to 0ubuntu2 [10:12] seb128: ok, thanks [10:12] yw [10:12] pitti: chrisccoulson: seb128: finally, the correct c-p-m should be in the ppa now and published (hopefully) [10:13] didrocks, \o/ [10:13] didrocks, thanks [10:13] didrocks: you rock! that must have been "fun" to untangle *hug* [10:13] will switch back to 3d in a bit and give it a try [10:13] pitti: yeah, for some definition of fun :-) [10:14] hmmm, i'm getting half a dozen emails every day from chromium users [10:16] seb128: also, do you have any news on bug #820103 ? [10:16] Launchpad bug 820103 in pidgin "Disconnecting connection 0x" [Medium,Confirmed] https://launchpad.net/bugs/820103 [10:16] chrisccoulson: see, you should start using it! :) [10:16] lol [10:16] i'm happy with my current browser ;) [10:17] didrocks: the c-p-m added to the ppa yesterday by mterry contained different animation settings - is that fixed now or even intended? [10:17] htorque, see tha backscroll [10:17] htorque: no, it was wrong, I fixed it [10:17] good, thanks (no backscroll :()! [10:19] seb128, cjwatson: ok, thanks [10:21] dobey, pitti: will we get ubuntuone-client-gnome back on the CD? or make sure that upgrades get it? njpatel didn't have it installed on his Oneiric system and I didn't either, it seems like natty->oneiric upgraders might loose the nautilus integration if that one doesn't get pulled in [10:22] jibel, mvo: ^ do you know if that's a know issue? if you do natty to oneiric updates could you check if that one is correctly installed on the upgraded systems? [10:22] seb128: looks like it, but only after b2 [10:22] shouldn't be an issue for upgrades, though? [10:23] pitti, well I'm wondering by it was not installed for me or njpatel [10:23] but maybe it got removed during the cycle and nothing pulled back in [10:23] I installed O fresh on this system ~ a2 [10:23] seb128, it wasn't installed for me on a fresh install last week [10:24] njpatel, chrisccoulson: ok, so that might be normal with their "installer", will be fixed after b2 apparently [10:24] ah, ok [10:24] we will go back to pre-install u1 things [10:24] hey kamstrup ...hate to say this, but I just did a fresh install of oneiric on a new disk and I got the same (i think) zg error at boot :( [10:24] hey jasoncwarner_ [10:24] hey seb128 [10:29] jasoncwarner_: that's just really really odd [10:29] kamstrup: you want the crash file? [10:29] jasoncwarner_: seems there's a subset of people with some really odd persisting problems with the fts extension [10:29] jasoncwarner_: why not :-) [10:29] kamstrup: the weird thing, like I said, is that I did a brand new, fresh install... [10:30] jasoncwarner_: yeah, it's the same computer still, right? [10:30] yeah, ubt a diff disk (ssd vs 7200rpm) [10:30] kamstrup: http://paste.ubuntu.com/693677/ [10:31] jasoncwarner_: yeah, that shouldn't matter... what I am thinking is that it's some odd platform specific error in the SWIG bindings for libxapian... [10:31] jasoncwarner_: ah, that trace, is another bug :-) [10:31] kamstrup: I have ZERO idea if this is related, but I feel like I got the moment I started Skype...then again...the moment I started skype caused all sorts of problems. ;) [10:31] jasoncwarner_: it's the old zeitgeist-daemon --restart bug [10:31] kamstrup: another bug? sheesh...this machine and zg ) [10:32] kamstrup: I have it back? [10:32] jasoncwarner_: nah, that bug is known [10:32] jasoncwarner_: I thought we had it distropatched or something, but mayb enot [10:33] kamstrup: well, I feel better that it is already known ;) perhaps a fix will be waiting for me when I wake up in the morning! :) I'm about to get to bed.... [10:35] jasoncwarner_: looking into this, sleep tight dude :-) [10:36] kamstrup: as always, mucho appreciated! [11:10] chrisccoulson, I found (and fixed) the evo-couchdb issue [11:10] rodrigo_, excellent, thanks :) [11:11] chrisccoulson, so, could you please build lp:~ubuntu-desktop/couchdb-glib/ubuntu and test with thunderbird? [11:18] rodrigo_, hmm, i still get "Cannot open book: Could not create DesktopcouchSession object" [11:18] chrisccoulson, hmm, you maybe need to recreate the desktopcouch config [11:18] let me find the instructions [11:19] that error is when it can't find the tokens in the keyring [11:20] chrisccoulson, http://www.freedesktop.org/wiki/Specifications/desktopcouch/Documentation/Troubleshooting [11:20] see the killing and restarting... section [11:24] rodrigo_, \o/ [11:24] * rodrigo_ dances [11:24] i added a contact in thunderbird, and checked it exists with futon [11:24] seems to work fine :) [11:24] cool [11:24] ok, i'll upload my little extension to handle the initial addressbook creation [11:24] then it will all work :) [11:24] yes, and we can go out for beer celebration :) [11:25] can we do the beer bit first? [11:25] let's not touch anything couchdb-related in the next few weeks though, just in case something else breaks :) [11:25] chrisccoulson, yes, sure :) [11:25] heh [11:32] \o/ running retracers [11:41] what plays the login sound when the session loads? [11:42] chrisccoulson, /usr/share/gnome/autostart/libcanberra-login-sound.desktop [11:42] seb128, ah, thanks [11:42] chrisccoulson, btw I still have a pending him to "disable login sound in oneiric" [11:42] chrisccoulson, if you want to do it you"re welcome [11:43] yeah, i don't mind. that's the first thing i turn off on a fresh install anyway [11:43] perhaps we should make the login sound conditional on having accessibility enabled? [11:43] I wonder if that .desktop works [11:43] i can see how it might be useful there [11:43] /usr/bin/canberra-gtk-play --id="desktop-login" --description="GNOME Login" [11:43] Failed to play sound: File or data not found [11:44] chrisccoulson, is it useful with a11y? the login screen ready sound is, but is the session sound useful? [11:45] yeah, good point [11:45] orca starts talking when you get to the desktop [11:46] drums at the start of ubiquity on the live CD is a big deal though [11:50] pitti: The category icons for Songs and Albums are missing in unity-lens-music. Does this need a UI freeze exception, or can I just fix it? [11:51] kamstrup, just fix it [11:51] kamstrup: JFDI [11:52] like [11:52] pitti, is there any chance to let in the applications updates from the queue once we have a first set of isos? [11:53] seb128: we can do that, yes; but still building ISOs right now, I better not risk uninstallability right now [11:53] pitti, no hurry ;-) [11:54] morning folks! [11:55] hey pedro_ [11:55] pedro_, how were your holidays? :) [11:55] seb128: did you ever look at and understand bug 797000 by chance? [11:55] Launchpad bug 797000 in desktop-file-utils "/etc/gnome/defaults.list uses non-existent inode/directory=nautilus-folder-handler.desktop" [High,Triaged] https://launchpad.net/bugs/797000 [11:55] hey pedro_, welcome back! [11:55] pitti, looked at, yes, understood no [11:56] rodrigo_, hola! , very relaxing , almost the only 'noise' i've heard were birds singing ;-) so pretty good [11:56] pitti, hello! thanks :-) [11:56] salut seb128! [11:56] hey pedro_, how are you? [11:56] * pedro_ swimming in the mail flood [11:56] pitti, but doesn't seem worth tracking for the team or for Oneiric [11:56] seb128, pretty good and you? [11:56] pedro_, I'm good thanks [11:56] pedro_, catching up after holidays [11:57] pedro_, ah cool, so you didn't see all the bugs we assigned to you? [11:58] rodrigo_, i'm going trough those now, later i'll run my 'assigns-all-the-bugs-to-the-spanish-guy' script :-P [11:59] ugh :) [12:01] is that a spanish war happening there? ;-) [12:04] yeah, I have a spam-pedro-with-bug-mail script, so he'd better not run his script [12:04] lol === MacSlow is now known as MacSlow|lunch [12:06] desrt, what are the chances that https://bugzilla.gnome.org/show_bug.cgi?id=654563 gets fixed this cycle? ;-) [12:06] Gnome bug 654563 in gio "info capplet: Failed to calculate disk space" [Normal,Unconfirmed] [12:06] Hello, any idea about when gnome-documents and sushi will be packaged? [12:08] jjardon, hello! whenever somebody interested steps up to package those for debian or ubuntu [12:08] jjardon, or said differently "no idea, lack of resources and packaging those is not the priority for Oneiric" [12:09] but maybe ricotz or somebody else will want to pick it up ;-) [12:10] seb128: yeah maybe we have to convince some packager ;) [12:11] jjardon, we will look at it after the Oneiric freeze if nobody else does it before [12:13] seb128, jjardon, currently not ;), sorry [12:13] seb128: only said that because the two packages are part of GNOME 3.2, and will appear in the release notes, so users will expect them avaliable [12:13] jjardon, yeah, and I would like to test shushi as well ;-) [12:13] let's when it's a bit less crazy [12:14] * rodrigo_ lunch [12:15] seb128: FYI we are in hard code freeze in GNOME, so do not expect lot of changes ;) [12:15] seb128: thanks anyway! [12:15] jjardon, right, yw, thank you for pointing that those are needed ;-) [12:15] bah [12:16] Well, that was interesting [12:17] Upgraded to Oneiric, restarted, and all my windows were invisible [12:17] we have mimetype issues [12:17] brb, testing another g-s-d speedup fix [12:28] seb128, pitti - https://launchpadlibrarian.net/80414933/farnsworth-oneiric-20110920-2.png :) [12:28] the time taken for gsd to start is shorter now [12:28] compared with http://ubuntuone.com/0xKppDPyZiHP9aDzmzzPoR [12:30] chrisccoulson, 11s to 5s, not bad [12:30] ;-) [12:30] heh :) [12:31] there is some variation with the xrandr plugin. it seems to take much longer with my laptop docked [12:31] but the changes i'm testing here seem to save ~2.5s consistently [12:35] so, i can't really see any other obvious wins in g-s-d, other than making the xrandr plugin not take so long. i wonder if that's just because i have a local monitor config [12:35] brb [12:37] * didrocks got his OneConf syncing working \o/ [12:38] ok, xrandr plugin still sucks without any monitors.xml [12:38] it looks like it's not meant to do anything in that case [12:38] i'll see if i can figure out what's going on [12:38] chrisccoulson: btw, is your stacking bug vanished? [12:38] didrocks, not sure, i'm on 2d atm [12:39] ok :) [12:39] i'll try 3d in a bit [12:39] but i'm quite happy with 2d now :) [12:39] mvo, hi [12:39] mvo, could you please look at https://code.launchpad.net/~zkrynicki/command-not-found/fix-839609/+merge/75440 [12:39] mvo, I believe it fixes the annoying c-n-f crashes-on-localized-input bug [12:40] would be nice to have a better alt+tab switcher in 2d :) [12:41] chrisccoulson: that's the pet project I want to trigger when I can get some spare time :) [12:41] would be a nice first QML app [12:41] can someone please explain alt-tab in unity now? Is there any other way I'm supposed to just switch between applications on the same desktop? [12:41] didrocks, that would be awesome :) [12:43] zyga: sure, will do [12:43] http://cdimage.ubuntu.com/daily-live/20110920/ is finally there for testing [12:56] didrocks, what was wrong with my ghost windows patch? [12:57] mterry: it was taken from compiz-plugins-main trunk, which reverted some of ours distro-patch (smspillaz has embeeded our distro patches in another branch and roll as part of the tarball). It contained additional commits we don't want as well [12:57] didrocks, mine did? It was a two liner [12:58] * didrocks looks at the diff again [12:58] didrocks, whoa, I think you're right [12:59] mterry: https://launchpadlibrarian.net/80368147/compiz-plugins-main_1%3A0.9.5.94%2Bbzr20110919-0ubuntu1~ppa1_1%3A0.9.5.94%2Bbzr20110919-0ubuntu1~ppa2.diff.gz [12:59] didrocks, I had a two-liner version that I intended to upload [12:59] didrocks, I must have forgotten to re-debuild... [12:59] mterry: no worry, just scary :-) [12:59] mterry: it's all fine now, I pushed the correct version (hopefully) [12:59] * mterry is really confused, remembers looking at a debdiff twice to be sure [13:00] mterry: ah, that was the issue! only twice. Three is mandatory with compiz :-) [13:00] didrocks, oh! [13:00] * didrocks hugs mterry, no worry ;) [13:01] didrocks, god, I'm all sorts of confused. the two liner patch was for qtwebkit for a completely different thing. The compiz-plugins-main thing went down just like you described. [13:01] * mterry drinks a large coffee [13:01] mterry: seems you had processes memory separation issues :)- [13:01] :-) [13:01] heh [13:01] and corrupted memory! [13:01] didrocks, anyway, this new compiz seems great [13:02] fixes all the stacking issues I've had [13:02] smspillaz, ^ [13:02] mterry: good news! hoping that we will be able to push it after beta2 :) [13:02] (still some known issues with the gimp and libro though) [13:02] ah, don't happen to use em [13:03] neither do I, but it seems some are :-) [13:03] crazy when you have LaTeX for everything :) [13:08] pitti, did we get to turn off hibernate by default in Oneiric? [13:08] didrocks: those issues aren't really manifesting themselves that much [13:08] seb128: not so far, I think [13:08] well I guess "no, because we didn't get the ui to turn it back on"? [13:08] seb128: also, we forgot to disable the startup sound :/ [13:08] didrocks: I think by that point we'll fix it in an SRU once I've got some time to really dig into it [13:08] seb128: that, too [13:09] smspillaz: let's see how it goes and what the feedback "in real use" will be [13:09] pitti, yeah, I was mentioning to chrisccoulson before lunch, the startup sound [13:09] wow, so g-s-d takes 0.5 seconds each time it tries to restore the backup, intended and default configurations, even if they don't exist [13:09] that's definitely dixable [13:09] pitti, does it still play for you? [13:09] **fixable [13:09] chrisccoulson, "backup"? [13:10] seb128, yeah, monitors.xml.backup [13:10] ah [13:10] then it tries monitors.xml [13:10] and then the system default [13:10] none of these exist in my case, yet it still spends 0.5s trying each one [13:10] that seems broken ;) [13:10] pitti, /usr/share/gnome/autostart/libcanberra-login-sound.desktop has "/usr/bin/canberra-gtk-play --id="desktop-login" --description="GNOME Login"" and that command doesn't work for me [13:10] $ /usr/bin/canberra-gtk-play --id="desktop-login" --description="GNOME Login" [13:10] Failed to play sound: File or data not found [13:10] seb128: this is a silly bug. why didn't it get fixed already? [13:10] chrisccoulson, indeed [13:11] desrt, not enough people to look at bugs, everybody is busy...? [13:11] seb128: yes, I still hear the sound in a guest session [13:11] seb128: glib developers are so lazy [13:11] pitti, ok, it's on my beta2 list of things to check on [13:11] Failed to play sound: Sound disabled [13:11] desrt, GNOME bug #658188 seems a lot as well [13:11] seb128: ^ here [13:11] Gnome bug 658188 in gio "g_app_info_set_as_last_used_for_type generates a broken mimeapps.list" [Normal,Unconfirmed] http://bugzilla.gnome.org/show_bug.cgi?id=658188 [13:11] but that's my personal account, where I indeed disabled it [13:12] desrt, it means every time you open the g-c-c info panel, default application without doing any chance, it screws all your preferred applications [13:12] seb128: do you always wait until the day after hard code freeze to tell me all of the things you want fixed? :) [13:12] desrt, broken since GNOME3, still broken in 3.2 [13:12] desrt, lol, yes, wouldn't be fun otherwise ;-) [13:13] desrt, sorry, I think they are not stopper for .0, maybe to fix for .1 [13:13] seb128: I'm just booting the current i386 live CD, login sound works there, too [13:13] the first one is probably easy enough [13:13] pitti, ok, I will look at turning it off then [13:13] i'm taking a look now [13:13] desrt, thanks [13:13] seb128: that command works fine in a default user account [13:14] desrt, the second one is "fun", opening the g-c-c info panel leads to gwibber being the default image viewer for me ;-) [13:14] pitti, it's probably my user config then [13:14] seb128: it's gnome's new approach to social media [13:14] ;-) [13:15] seb128: you "view" images by posting them to your twitter page and asking your friends to describe to you what they saw [13:15] wow [13:15] so, gnome_rr_config_apply_from_filename_with_time does a gnome_rr_screen_refresh before even checking if the configuration exists [13:15] i think it's a great way to spend more time talking to your friends! [13:15] pitti, I need to think what would be the best way, we could disable the autostart by default but that's probably wrong for those who will want to set a login sound back [13:16] i think we're going to have g-s-d startup down to < 1s in a bit ;) [13:16] seb128: I'd just disable the default in gsettings [13:16] chrisccoulson: yay [13:16] desrt, it should just connect you to random community users [13:16] "easy community building" ;-) [13:17] GNOME: "helping you meet new people!" [13:17] chrisccoulson, hurry, it seems ubiquity is broken, we can get respins :p [13:18] now I hide before pitti start chasing me with a big stick [13:18] seb128: btw, I remember we discussed about sound-theme-freedesktop brought by libcanberra0, seems it's still the case, so we have system sounds installed as well? [13:19] didrocks, yes, but it seems it made some users happy and it's small enough [13:19] since nobody pushed to revert and we have the CD space for it now I didn't bother [13:19] I hate those sounds, time to disable them :) === m_conley_away is now known as m_conley [13:19] didrocks, they should be off by default [13:19] but there for users who want to set a sound theme [13:19] hum, it's not there, maybe my old old configuration set them back [13:20] if it's off by default, I'm happy, I let my configuration this way to not forget about it [13:21] ok, the obvious question is how to disable them [13:21] heh [13:21] juste mute then I guess [13:21] just* [13:21] ahhhhhhhh way better :) [13:22] when I think I let those on to remember to continue the discussion about the depends with the kubuntu guys… [13:22] so much pain for nothing ;) [13:22] how /do/ you change the sound theme? [13:23] seems you can't change it in the ui, just mute the current one and change the default alert sound [13:23] rodrigo_: I'm checking whether it would be worth updating NM so late. there are some bugs fixed and I think it also fixes the agents crash I've been fighting, but it also bumps soname for libnm-util [13:23] Laney, didrocks: yeah, I was wondering the same, seems quite broken :-( [13:26] :( [13:27] pedro_, can you upstream bug #854401? [13:27] Launchpad bug 854401 in nautilus "the text on desktop icons doesn't adapt to the background color" [High,Confirmed] https://launchpad.net/bugs/854401 [13:28] seb128, sure, its going there in a bit [13:28] seb128: when did you/njpatel upgrade? [13:29] pedro_, thanks [13:29] dobey, unping, njpatel did an oneiric install so he got the installer and not things preinstalled [13:29] ok [13:29] dobey, I upgraded earlier and probably got it dropped when it became a new source or when we upgraded nautilus to gtk3 before you ported u1 [13:30] seb128: on upgrade though, if you've already got stuff installed, it should still be installed after, unles spackages are broken right? [13:30] right [13:30] ah ok; yeah, that's why i asked. if you upgraded when the package went away for a short time, you would have been missing it :) [13:30] cyphermox, ok, keep me posted, there is libnm-gtk in n-m-applet, which g-c-c depends on [13:31] the what? [13:32] brr [13:34] hey cyphermox [13:34] hey seb128! [13:35] cyphermox, gnome-control-center, they use widgets from libnm-gtk [13:35] yeah [13:35] mvo, thanks [13:36] cyphermox, they used nm-applet before but gnome-shell conflicts with it so they moved code around it seems [13:36] yeah, see that now [13:36] it's going to be a little painful but it should be fine [13:37] nm-applet is just bugfix, nm looks like it is too (checking) [13:37] well, we reverted the g-c-c commit for now [13:37] if it's late in the cycle to update nm we can probably adapt g-c-c [13:37] or let the commit reverted since we still use nm-applet [13:38] well, the biggest issue really is NM, not nm-applet [13:39] nm changing the soname for libnm-util just means updating g-c-c (which would be done anyway), and the vpn plugins, afaict [13:39] well, that's what g-c-c needs I think, nm [13:39] if they moved the code there [13:39] no [13:39] it's a new lib in the network-manager-applet source pkg :) [13:39] oh ok [13:40] right, easier than I though then [13:40] we just need to update nm-applet then from a g-c-c perspective [13:41] seb128: no, because that new nm-applet also requires the new NM, I think [13:41] ok [13:41] but I think this could be reverted [13:42] your call [13:42] cyphermox, otherwise, do you plan to work on the evolution, evolution-exchange, gtkhtml updates as well? I put your name for those in the etherpad ;-) [13:43] they are not on the CD so no hurry [13:43] yeah, I plan to [13:43] great [13:43] I was going to try and fix this NM crasher first, upload after beta [13:43] but I'm also kind of stuck, it [13:44] it's hard to reproduce so very hard to confirm it's fixed :) [13:47] desrt, bug #839876 (https://launchpadlibrarian.net/78809203/Stacktrace.txt) [13:47] Launchpad bug 839876 in evince "evince crashed with SIGSEGV in g_settings_delay()" [Medium,Confirmed] https://launchpad.net/bugs/839876 [13:47] desrt, do you know if that's an application bug or a gsettings one? [13:51] seb128: I think we should retitle bug #854503 with "Take real launcher size to get icons area next to the launcher", wdyt? [13:51] Launchpad bug 854503 in nautilus "Using Unity setting "Hide Launcher - never", files on desktop can not be placed close to the launcher" [Low,New] https://launchpad.net/bugs/854503 [13:51] seb128: offhand it looks like a missing schema [13:51] seb128: not for oneiric anyway, as -2d and -3 doesn't have the same values, and with gsettings maybe segfaulting back next release if the value isn't there, it makes a dep between nautilus and unity-common :) [13:51] didrocks, works for me [13:52] desrt, that's when you tell me that it was clear at least when you were aborting on missing schemas? ;-) [13:52] seb128: i think next cycle i'll be bringing back insta-fatal for missing schemas again [13:52] seb128: in a modified form that lets you explicitly turn it off [13:52] thank you zyga! [13:52] this is totally insane [13:53] desrt, this? [13:53] seb128: what's your take on critical being fatal by default? [13:53] mvo, thanks, sorry for not proposing this during your absence :/ [13:53] desrt, things like https://launchpadlibrarian.net/78230385/Stacktrace.txt are missing schemas as well I guess? [13:53] seb128: but able to be disabled in language bindings (like python) where it gets converted to an exception [13:53] desrt, ok during unstable cycle, not on stable versions [13:54] seb128: yes. things like that are also missing schemas [13:54] desrt, you really want to minimize things closing on stable version, often applications can keep running enough for you to not have work lost [13:54] desrt, ok thanks, I wish I know how users land with missing schemas [13:55] they are in the deb, that doesn't make much sense [13:55] seb128: this is actually an interesting problem [13:55] or corruptions happen often... [13:55] zyga: no worries, still time until final freeze [13:55] we don't want to explode on missing schemas [13:55] but when looking through traces we do want to be able to know about them [13:55] (ie: that they were missing) [13:56] seb128: i guess it comes down to being able to get the xsession-errors [13:56] fedora does a cool trick where it greps for the name of the process that crashed and sends only lines that match that [13:56] desrt, we have .xsession-errors [13:57] desrt, https://launchpadlibrarian.net/78809028/XsessionErrors.txt that's for the evince one [13:57] makes it nicer not to have to download and dig through 14 meg :) [13:57] well apport does filter [13:57] but I think the filters need improvement [13:57] odd. no gsettings errors from evince. [13:58] it grabs only on some keywords like Gtk-Warnings [13:58] mvo, good :) [13:58] seb128: this gnome-control-center bug is pretty obvious [13:58] desrt, the disk or the mimetype one? [13:58] seb128: you have "none" in your fstab instead of "swap", which is what we expect [13:58] for the mountpoint of the swap partition [13:59] # swap was on /dev/sda5 during installation [13:59] UUID=d077cd26-da2d-4c0f-ad00-276c510fd4ae none swap sw 0 0 [13:59] ya. the 'none' is the trouble [13:59] change it to 'swap' and the trouble disappears [13:59] we have a few possible fixes here [13:59] desrt, it's the "mount point" field [13:59] which should it be "swap"? [13:59] why [13:59] seb128: i'm just explaining what works [13:59] desrt, swap has no mount point [13:59] right [14:00] I knew about that workaround ;-) [14:00] we have this code if (strcmp (fstab->fs_vfstype, "swap") == 0) continue; [14:00] so there :p [14:00] that's why :) [14:00] ok [14:00] anyway... so there are 3 fixes we could do [14:00] first is that you change your fstab [14:00] next? ;-) [14:00] next is that i modify this part of glib to use our own big internal list of stuff to ignore [14:01] but [14:01] # [14:01] next is that we patch control-center not to call query_filesystem_info() on files that don't exist [14:01] i like the last option best. it's easiest and least likely to explode. [14:01] shouldn't "fstab->fs_vfstype" be the third field? [14:01] fs mount type [14:01] which is fstab for me [14:01] UUID=d077cd26-da2d-4c0f-ad00-276c510fd4ae none swap sw 0 0 [14:01] yes. fedora has "UUID=blahblah swap swap .." [14:02] well the line you show me check fs_vfstype [14:02] which should be swap in Ubuntu as well [14:02] sorry [14:02] i'll show another [14:02] mvo, one more thing, could you please make sure to upgrade the version number stored internally in the application? this is the only thing that users report most of the time and differentiating between versions helps. I changed the version this time but it has to be maintained continuously (before it was stuck on some older release) [14:02] if ((strcmp (mntent->mnt_dir, "ignore") == 0) || (strcmp (mntent->mnt_dir, "swap") == 0)) continue; [14:02] and [14:02] if ((strcmp (mntent.mnt_mountp, "ignore") == 0) || (strcmp (mntent.mnt_mountp, "swap") == 0)) continue; [14:03] so i think we should try to fix gio, but that can wait until after the release [14:03] desrt, what about adding "none" to those? [14:04] meanwhile, this is a heuristic and gio will never be 100% perfect in all cases [14:04] right [14:04] so probably gnome-control-center should be more robust [14:04] agree on fixing g-c-c [14:04] feel free to bounce the bug back there ;-) [14:04] we can probably fix it by just treating the NULL case as 0 bytes [14:04] or to clone it [14:04] meh [14:05] i'll do the quick patch for GIO now [14:05] plus the g-c-c patch [14:05] and we'll look at fixing it properly after the release [14:05] can keep it all on the same bug. will make release-team decisions easier [14:07] desrt, thanks [14:08] seb128: does this bug impact debian, in your opinion? [14:09] * desrt needs to know where to assign the blame :) [14:09] desrt, don't you have a debian server? ;-) [14:09] actually, no [14:10] I would tend to say that yes but easy to verify by somebody who has a debian box [14:10] i.e pitti [14:10] best i have is a laptop [14:10] seb128: don't YOU have a debian server? :p [14:10] I've no server :p [14:10] more shame for you than me, i say :) [14:10] eh. compiz just crashed because I re-enabled the task switcher. [14:10] I'm a desktop guy [14:10] ;-) [14:10] now unity is gone, but at least I have alt-tab working :) [14:11] hallyn: the task switcher have some keys conflicting like unity, (alt + tab), so you probably acked the fact to remove unity [14:12] didrocks: why would it not just obsolete the other keybinding? [14:12] (shift-alt-tab would have been the one) [14:12] seb128: ya. it's a debianism. [14:12] hallyn: because compiz-config-settings-manager doesn't know how to do that? patches welcomed :) [14:13] fair enough :) [14:13] (I thought it used to do that years ago though!) [14:13] (guess i was wrong) [14:13] desrt, yeah, I just looking on alioth, it's none there as well [14:13] hallyn: I don't think so, it just remove the plugin unfortunatly [14:13] or you can ignore the conflict [14:13] desrt, on a debian 6.0.2 [14:13] and then, weird things can happen :) [14:14] didrocks: ok, i can just manually disable the other - but, can i ask, [14:14] is there some place where the motivation for the new alt-tab behavior is documented? [14:14] hallyn: right, disable the other, then reenable unity [14:14] hallyn: I guess it's a question for the ayatana mailing list [14:14] to talk with the designers [14:14] didrocks: ok, thanks [14:16] yw [14:29] seb128: the patch in that bug is tested and works now if you want to vendor-patch it [14:30] https://bugzilla.gnome.org/show_bug.cgi?id=654563#c5 [14:30] Gnome bug 654563 in gio "info capplet: Failed to calculate disk space" [Normal,Unconfirmed] [14:30] desrt, thanks [14:30] doing g-c-c now [14:37] dobey, ping [14:38] rodrigo_: hey [14:38] dobey, fixed the evo-couchdb issue [14:38] rodrigo_: awesome [14:38] dobey, seems recent changes to e-d-s made it not happy with SoupSessionAsync being used on threads [14:38] ah [14:39] so, just changed that to use SoupSessionSync, since we do sync operations anyway [14:39] makes sense [14:40] now, replication doesn't work for me though :( [14:40] does it for you? [14:42] rodrigo_: that's mostly a server issue [14:42] yes, BAD_REQUEST [14:42] it's known and being worked on; but we have some SRUs to do of desktopcouch for old versions [14:42] ah ok [14:42] hrmm, you shouldn't get a 400 [14:42] should be getting a 503 [14:42] i seem to get 503s in replication log [14:43] I consistently get 400's [14:44] weird [14:44] rodrigo_: can you make a new user and set up ubuntuone with same account in that user, and see if it gets the same errors? [14:47] yes [14:55] rodrigo_: cool; thanks again [14:55] * dobey goes to get some lunch and run some errands [15:00] seb128, pitti - https://bugs.launchpad.net/ubuntu/oneiric/+source/gnome-desktop3/+bug/854101/comments/7 :) [15:00] Ubuntu bug 854101 in ubuntu-boot-speed "gnome-settings-daemon takes waaaaaaaay too long to start" [Undecided,Confirmed] [15:00] i can't believe how fast the unity-2d session starts here now [15:00] i can barely blink after typing my password before the panel has loaded [15:01] chrisccoulson: wow, that looks awesome! [15:01] chrisccoulson: X.org still takes an awful amount of time on your machine [15:01] i shall push the changes to the desktop PPA in a bit, and would appreciate some testers [15:01] to make sure i've not broken anything ;) [15:01] yeah, i'm not sure why Xorg takes so long here [15:01] chrisccoulson, count me in for testing from the ppa when you put it there ;-) [15:02] chrisccoulson: great work! did you already got a response from upstream for your previous patch? [15:02] chrisccoulson, oh, and stop using 2d :p [15:03] pitti - which patch was that? [15:03] seb128, i prefer 2d ;) [15:04] no blurry icons in the dash :) [15:04] chrisccoulson, bah, your g-s-d work should work for me on 3d as well ;-) [15:04] chrisccoulson: I thought you had one for the media plugin (pulse), and before one for xrandr? [15:04] pitti - i don't think so. i only started on this yesterday evening [15:04] i intend to send them upstream, or at least start a discussion on the issues i've found :) [15:05] (like, doing 3 reprobes, which is insane) [15:05] probably won't help on http://people.canonical.com/~pitti/bootcharts/daniel-oneiric-20110914-disable-indicator-session.png as both CPUs are full anyway [15:05] but on faster CPUs this should be quite noticeable [15:06] could buy 3.5 s on http://people.canonical.com/~pitti/bootcharts/donald-oneiric-20110914.png (that's my X201 workstation) [15:08] chrisccoulson: sorry, I didn't follow earlier, but what those gsd configuration are for? [15:08] didrocks, which ones? [15:08] pitti: [15:08] * didrocks wonders why with nvidia he always get on gsd now a "can't find suitable size" and enormous fonts now at each startup [15:08] [ 14.607] (II) intel(0): Initializing HW Cursor [15:08] [ 17.020] (II) intel(0): RandR 1.2 enabled, ignore the following RandR disabled message. [15:08] are those timestamps in my Xorg log? [15:08] chrisccoulson: the backup, intended, default [15:09] didrocks, oh, those are for the display configuration [15:09] by default, there aren't any configurations to load, but gsd still spends 0.5s iterating each one [15:09] seems what is maybe broken there then :) [15:09] but i've fixed that locally now :) [15:09] yeah, I'll tell if you that makes vanish the dialog there [15:10] didrocks, hmm, it probably won't fix your problem [15:10] chrisccoulson: where are they saved? maybe I can rm them [15:10] didrocks, ~/.config/monitors.xml is probably the one [15:10] (that's the intended configuration) [15:11] ok, I have one [15:11] let me remove it [15:11] sorry to interrupt (again), but this doesn't influence bug 828112, right? [15:11] Launchpad bug 828112 in unity-greeter "Password field feedback slow at times" [High,Triaged] https://launchpad.net/bugs/828112 [15:11] it's the thing that keeps popping up a window on login [15:11] ah seems that was it :) [15:11] that happened to me too [15:11] thanks chrisccoulson, not sure how a bad one was saved though [15:11] htorque, i don't think that anything i'm doing will affect that bug [15:11] jcastro: ah! [15:12] i'm focused on the session start only atm [15:12] didrocks: it's something else that happened, I get it on 2 machines [15:12] jcastro: I just rm the file and it "fixed" it [15:12] but still not sure why it's saved [15:12] the display thingey saves there [15:13] cjwatson: could yelp-xsl and yelp-tools be added to the desktop set also? [15:13] jcastro: nvidia? [15:13] yep [15:13] pitti - so, that "17.020" timestamp in my xorg log lines up nicely with the the greeter starting [15:13] oh, no, I don't think the nvidia tool saves there, I think the GNOME one does [15:13] i guess that's something i should bug RAOF about [15:13] ok, maybe we really should blacklist the display dialog with nvidia [15:13] but I am not certain [15:13] perhaps he can figure out what Xorg is doing for 3 seconds there [15:13] jcastro: yeah, I guess so, we should hide the display with nvidia blob driver anyway [15:13] hide the gnome display right? [15:14] indeed [15:14] because you need the nvidia one [15:14] oh ok [15:14] just a thought [15:14] I agree [15:14] are we allowed to put things in g-c-c yet or does that incur hatred from upstream still? [15:14] jcastro: you mean, without distro-patching? [15:14] yeah [15:15] it's a distro-patch right now, all the thing we put it there [15:15] jasoncwarner, Sweetshark, bryceh, chrisccoulson, didrocks, tremolux, Riddell, kenvandine, cyphermox, mterry, rodrigo_, seb128, tkamppeter, pedro_: meeting reminder in 15 mins [15:15] w00t [15:15] yup yup [15:15] jcastro: put things into gcc> that's a long lost cause [15:15] * pedro_ waves [15:15] we are so bad, we even put ubuntuone there TWICE! [15:15] chrisccoulson, didrocks: the backup monitors.xml config thing is in case you apply a broken config I think, so it can go revert to the working one after the delay [15:16] iirc [15:16] by the way, will the extra ubuntuone be gone before beta2? [15:16] the third file seems weird [15:16] pitti: depends hadess told me at desktop summit that he wasn't against putting unity settings there if unity runs! [15:16] jbicha, what extra ubuntuone? [15:16] seb128: there are two ubuntuone settings in c-c [15:16] not here [15:16] seb128: it's not a backup one, it's read when gsd starts apparently… [15:16] hey before your meeting, I have a question. I want to tell users about the new DVD image and all it's goodness, is this the canonical list of extra packages or are there more somewhere? http://people.canonical.com/~ubuntu-archive/seeds/ubuntu.oneiric/dvd [15:16] both open the control panel [15:16] I've seen a bug about it [15:16] but it doesn't happen for me [15:16] seb128: maybe one is u1-installer, the other u1-control-panel? [15:17] oh right [15:17] I don't have the installer installed [15:17] install the installer :-) [15:17] yeah, that's what they are [15:19] is jcastro's link right? [15:19] if so, what's that mono stuff doing therE? [15:19] didrocks, http://pastebin.ubuntu.com/693835/ that's the gsd code [15:20] didrocks, it's a bit weird, but it seems they try to be smart about backup configs on start as well, in case you have a non working configuration [15:20] like you tried to apply something that didn't work, screwed your xorg and forced you to reboot [15:21] jbicha, you probably better drop an email to cjwatson for those, easier to track than irc pings usually [15:22] seb128: indeed, but does those backup configuration file works with nvidia? it seems not [15:22] (the one which was generated) [15:22] and maybe it detects nvidia as a non working one [15:22] yeah, not sure about nvidia ;-) [15:22] let me just try to open the display properties again [15:23] and restart g-s-d then [15:28] jbicha: yes, please mail me this kind of request, I don't process them without an e-mail traili [15:28] *trail [15:29] cjwatson: seb128: thanks [15:29] seb128: jcastro: so confirming, the gnome display thing is creating that file [15:29] as when you change the resolution it works [15:29] (with a lot of noise) [15:29] and let that invalid file then [15:30] maybe we should just blacklist the GNOME thing with nvidia blob driver? [15:30] I think so, the tool is useless with nvidia and it will just confuse people into thinking it works. [15:31] jasoncwarner, Sweetshark, bryceh, chrisccoulson, didrocks, tremolux, Riddell, kenvandine, cyphermox, mterry, rodrigo_, seb128, tkamppeter, pedro_: meeting time [15:31] o/ [15:31] o/ [15:31] didrocks, we used to do something like that, tseliot made patches by then to run the nvidia control dialog iirc [15:31] hello everyone [15:31] hey [15:31] hey [15:31] heyo [15:31] https://wiki.ubuntu.com/DesktopTeam/Meeting/2011-09-20 [15:31] * kenvandine waves [15:31] hello! [15:31] FYI, quick release status: the first smoketests of the current b2 candidates are coming in [15:32] we found a rather serious ubiquity bug and will need a respin [15:32] but as long as you are online, they actually install pretty well [15:32] hi :) [15:33] otherwise it's nothing earthshattering on our side [15:33] kenvandine: do you need to discuss anything partner-wise? [15:33] i gave the u1 guys the list of release bugs skaet is tracking for their packages [15:33] they are going to follow up on them all [15:34] hi [15:34] that is all i have [15:34] we also had a quick discussion with U1 team yesterday [15:34] kenvandine, pitti: will we get the whole u1 stack back on the CD after beta2? [15:34] looks like we'll put u1-client control-panel, and music store back on the default install [15:34] good [15:34] makes sense :-) [15:34] not the couchy bits, of course [15:34] that is a much better choice [15:34] cool! [15:34] yeah [15:34] * kenvandine never thought that was a good idea [15:34] we dropped them without need [15:34] :) [15:34] pitti, not ubuntuone-couch? (I don't think it pulls in anything couchy) [15:35] it was really quite pointless [15:35] mterry: I thought that's just a command line tool? [15:35] mterry: but I don't mind [15:35] pitti, no, it provides some python bindings that duplicity/deja-dup uses [15:35] "whatever we had in natty" [15:35] pitti, can't upload to u1 without it [15:35] does that pull in desktopcouch? [15:36] mterry: but then deja-dup should certainly have a dependency or at least a recommends to it? [15:36] kenvandine: no [15:36] good [15:36] pitti, it did before, but when we went down the installer route, i dropped it and added support to deja-dup to run the installer (which I made sure installs ubuntuone-couch) [15:36] mterry: it moved to universe yesterday, so better add a recommends [15:36] pitti, I can go back to before [15:37] * mterry wishes people would make up their minds [15:37] mterry: please add it back then [15:38] k [15:38] didrocks: thanks for the unity update on the wiki [15:38] yw [15:38] didrocks: anything to discuss? should the team test some newer compiz packages? (sorry, I lost track of them) [15:39] pitti: yeah, the ppa should contained all what is needed (as I claimed on the wiki ;)) [15:39] so go ahead, download, test, and ping me if anything is wrong [15:39] * didrocks looks at the stacking breaker, chrisccoulson! [15:39] chrisccoulson: ^ [15:39] :) [15:39] i've not tested it yet [15:39] chrisccoulson: and while you are at it, please make it fast, will you? [15:39] * pitti hugs chrisccoulson [15:39] that means switching back to 3d for a bit ;) [15:39] * chrisccoulson hugs pitti [15:40] chrisccoulson: come on, between two awesome fixes! :-) [15:40] :) [15:40] * pitti updates in the meantime [15:40] i'll try it in a bit. i want to get the gsd/gnome-desktop changes in to the PPA first [15:40] the PPA works great for me [15:40] I'll probably pushed the fixed Qt in the ppa as well waiting for the freeze [15:40] tremolux: anything to discuss for s-c? seems to be going well [15:41] oh, no new packages/ [15:41] ? [15:41] pitti: hey! nothing really to discuss, we are just fixing stuff [15:41] argh, seems add-apt-repository messed up retrieving the key or so [15:41] tremolux: that's good news, at this point :) [15:41] (if some people wants to test the oneconf staging server for syncing, please ping me, feedback would be welcomed ;)) [15:41] pitti: yes, I feel really good about it, it's getting into nicer shape every day [15:42] BTW, if you need to roll back, use "sudo ppa-purge ppa:ubuntu-desktop/ppa" [15:42] works really nicely [15:42] * kenvandine hugs ppa-purge [15:43] ok, nothing else from my side; AOB? [15:43] didrocks: I'll definitely test with the server, I've been using the local test data so far [15:43] Anyone have any MIRs that I don't know about? [15:43] tremolux: great, thanks! [15:44] didrocks: oneconf ftw! [15:44] hopefully not, pretty late for MIRs [15:45] mterry: I just got asked again about lo-menubar bing included in main ;) [15:45] veto [15:45] pitti: thanks [15:45] tremolux: thanks to believe in it! :) [15:46] Sweetshark, I thought we were holding off, for stability reasons? [15:46] didrocks: :) [15:46] mterry: right. [15:46] phew [15:46] :) [15:46] oh, the i386 and amd64 retracers have caught up again [15:46] they are running again now [15:47] \o/ [15:47] sorry for the 2 days downtime, needed to fix a couple of bugs [15:47] just so everyone is aware, i will be out again tomorrow, but returning again on thursday [15:47] so, then without further ado, let's fix this oneiric thingy! [15:47] oh, and I'm on vacation next week, FYI [15:47] thanks everyone [15:47] mterry, chrisccoulson: enjoy your holidays! [15:47] :) [15:48] heh, thanks :) [15:48] chrisccoulson: do you already have someone briefed for mozilla babysitting while you are away? [15:48] pitti - not yet. i can sort that out though [15:48] i guess i need a victim ;) [15:49] chrisccoulson, mterry: enjoy! [15:49] ok, g-s-d is in the ubuntu-desktop PPA [15:49] just doing gnome-desktop now [15:49] chrisccoulson, give it to the security team, they love firefox ;-) [15:49] lol [15:50] chrisccoulson, what changes does have that g-s-d on the ppa? [15:50] well, I actually had micahg in mind when I pondered people who know about firefox [15:50] rodrigo_, make it start up faster :) [15:50] ah [15:50] rodrigo_, https://bugs.launchpad.net/ubuntu/oneiric/+source/gnome-desktop3/+bug/854101/comments/7 [15:50] Ubuntu bug 854101 in ubuntu-boot-speed "gnome-settings-daemon takes waaaaaaaay too long to start" [Undecided,Confirmed] [15:55] chrisccoulson, cool! [15:55] ok, gnome-desktop uploaded too [15:56] rodrigo_, you probably want to review the changes and test it too, to make sure i'm not doing anything crazy ;) [15:56] chrisccoulson, do you have a link to the patch in g-s-d? [15:56] chrisccoulson, yeah, right :) [15:57] rodrigo_, there's 3 patches - you can grab them from the package in the PPA now [15:57] ok [15:57] i've not pushed them anywhere else yet [15:57] ignore the profiling patch ;) [15:57] that was just so i could get profiling information to stdout [15:59] pitti, can you score up the i386 builds for https://launchpad.net/~ubuntu-desktop/+archive/ppa/+sourcepub/1948415/+listing-archive-extra and https://launchpad.net/~ubuntu-desktop/+archive/ppa/+sourcepub/1948422/+listing-archive-extra ? === htorque_ is now known as htorque [15:59] seb128: done [15:59] pitti, danke [16:00] thanks :) [16:10] does zeitgeist really need to start as soon as the session loads? [16:11] chrisccoulson: it's dbus activated, but zeitgeist-datahub triggers it [16:11] chrisccoulson: but that's being fixed [16:11] chrisccoulson: seif told me that they rewrote this entire thing in C [16:11] pitti - ah, ok. that's good. it currently seems to take up 0.5s CPU here before the rest of the unity-2d components [16:11] chrisccoulson: I mentioned that on https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis [16:11] ah, great. Im on omgubuntu. [16:12] pitti - i guess i could add my gsd findings to that too [16:12] seb128: bug #854448 is not really a duplicate, it's a multi-arch issue [16:12] Launchpad bug 854448 in unity "AMD64 Skype should depend on i386 sni-qt (multiarch)" [Undecided,New] https://launchpad.net/bugs/854448 [16:12] seb128: skype amd64 should pull i386 sni-qt [16:13] Sweetshark: doesn't actually read badly, though? [16:13] chrisccoulson: sure, please do [16:14] pitti: yeah, could have been much worse. I did not read the news there -- I first noticed my blog getting a lot more attention than usual ... [16:15] kirkland, hello! Did you ever make any progress on bug 845738? [16:15] Launchpad bug 845738 in ecryptfs-utils "Add ecryptfs-verify-private script" [High,Triaged] https://launchpad.net/bugs/845738 [16:18] didrocks, ok, it's not me who duplicated it, I just commented ;-) [16:18] seb128: ok, just pointing that we discussed lenghty about it this morning :-) [16:19] seb128: and as it will bring the i386 Qt stack on amd64, we can't install both by default on amd64 [16:19] apart from skype, not sure which other Qt apps is in this situation [16:19] didrocks, ok, thanks for letting me know ;-) [16:19] I'm glad I neither use amd64 nor skype :p [16:20] seb128: you want 2 Qt stack on your machine, I know it! :-) [16:26] chrisccoulson: tested your packages: http://img.xrmb2.net/images/696959.png → \o/ [16:27] wow, that's a massive win [16:27] i wasn't expecting that [16:27] and everything still works ok? [16:27] looks like it :) [16:27] those 6 blocks of Xorg CPU are g-s-d doing 6 probes [16:28] i've cut those right down [16:30] i added all my findings to https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis anyway [16:30] oh, rodrigo has gone? [16:30] ah [16:30] brb, dinner time [16:33] chrisccoulson: pitti: seb128: I was planning on babysitting firefox while chrisccoulson was gone (with the assumption that little will need to be done before release) [16:33] mterry: hrm, I haven't :-/ been pretty overloaded lately [16:34] mterry: tab's open in my browser, will try to get to Inbox-0 and Tab-0 by EoD today :-) [16:34] kirkland, heh, no worries [16:35] micahg, that's great thanks, and yeah chrisccoulson said there should be not a lot to do, maybe an updated tarball with no changes to upload [16:36] brb [16:38] good night everyone! [16:39] night pitti [16:40] pitti, did you see htorque's bootchart? [16:41] chrisccoulson: ignoring display ports 1-3 and non-existing hdmi ports 1-3 via xorg.conf shaved off another 14 seconds: http://img.xrmb2.net/images/141990.png - but that's likely a kernel regression as that works fine with natty's kernel. [16:41] 15s -> 3s for g-s-d [16:41] yeah, that looks better [16:45] chrisccoulson, that's awesome! [16:45] kenvandine, unity-2d session starts in less than 3s here now ;) [16:45] nice [16:46] i wonder how hard it would be to get compiz to start as fast as metacity :/ [16:47] chrisccoulson: dlopening, dlopening… [16:48] heh [16:48] we need one big fat binary [16:48] that's why chrome is one big binary [16:48] and why firefox probably will be at some point too [16:50] didrocks, what actually does take all the time when compiz starts? [16:50] it's not something i've looked at yet [16:50] perhaps i will do later though ;) [16:50] it's pretty sad compared to metacity [16:50] chrisccoulson: basically, initialising gl (composite) and dlopening all the plugins enabled [16:51] loading the configuration backend is slow [16:51] and the configuration backend backend (gconf) slower [16:52] heh, and i bet compiz takes the 0.5s hit of starting gconf as well now, seeing as nothing else is using it [16:52] oh, isn't metacity using it though? [16:53] yeah, metacity starts it [16:53] chrisccoulson: Libreoffice master has code for making everyting one big binary too on master. [16:53] metacity still using gconf [16:53] yeah [16:53] so it's a fair game there :) [16:53] chrisccoulson: but compiz reads all keys for all plugins [16:53] chrisccoulson: well, an experimental switch [16:53] didrocks, there's an easy fix for that. we could have gconf up and running whilst g-s-d is loading [16:53] chrisccoulson: 900 keys was the natty default profile [16:53] gnome-session used to start it really early anyway [16:53] indeed [16:54] i'm going to try that again, seeing as it won't block anything initially [16:54] it might help compiz a bit ;) [16:54] 0.5s is still someting :) [16:54] Sweetshark, yeah, big binaries seem to load quicker :) === zyga is now known as zyga-afk [16:55] jjardon, can we make the battery icon for indicator-power not turn red when i still have over 2 hours of battery life left? :) [16:56] it gets even faster when you optimize the binaries so everything needed is in order [17:00] oh, can I put some feelers out for how the testing of the stacking branch is going ? [17:01] chrisccoulson: are you volunteering to make compiz startup time faster ? [17:01] I've wanted to do this for a while but haven't had time to [17:02] smspillaz, possibly ;) [17:02] it would certainly be nice to make it start faster for P [17:02] chrisccoulson: of course [17:02] chrisccoulson: I don't plan to change compiz much for P though [17:03] yeah, probably a good idea :) [17:03] have a good evening everyone! [17:03] night didrocks [17:03] err [17:03] smspillaz: thanks, you too! [17:03] evening, rather [17:03] heh [17:03] "morning!" :) [17:03] yeah, time for doing some exercice :) [17:03] heh, indeed, morning [17:12] smspillaz, you asked about the testing of the stacking branch? I'm loving it [17:12] sweet [17:12] smspillaz, no regressions so far, fixes every wart I saw [17:12] I noticed one problem with libreoffice the other day [17:12] smspillaz, yeah, someone was saying that [17:12] i'm just upgrading now, then i will switch back to 3d and test :) [17:12] and the regression detection code I put in still notices some errors with more complex cases [17:12] smspillaz, but I haven't happened to use it [17:13] but it should be 99% fixed now [17:13] and considering I've been in lo for a bit now working on this uni assignment [17:13] and haven't seen it, guess we're good [17:13] smspillaz, I saw you had some branch for the minimized window case? [17:13] yes, that's been merged in [17:13] I forgot to tell you that [17:13] you can tell stbe there's no need to hack around that now [17:14] smspillaz, is it in a PPA for easy testing? I'd like to confirm that deja-dup will work like I expect now [17:14] mterry: no ppa yet [17:14] (it does some minimized craziness) [17:14] QQ [17:14] mterry: we're probably going to do another compiz upload later in the week with the stacking branch merged in. I'll cherry pick that fix in at the same time [17:14] err [17:14] next week [17:15] bigon, any idea why bug 828756 came back with empathy 3.1.92? [17:15] Launchpad bug 828756 in empathy "getting "connection is untrusted" warnings" [High,Confirmed] https://launchpad.net/bugs/828756 [17:15] smspillaz, yeah, but I'll be on vacation then, and I want to know whether I can revert my workaround in code before then. I'll test the patch [17:15] since it had been confirmed fixed in gnome-keyring [17:15] mterry: if you know how to build the compiz package I can give you a patch [17:15] mterry: the problem is in both unity and compiz though [17:16] mterry: or you can give me a testcase and I'll see if it does anything crazy here [17:16] smspillaz, guh [17:16] smspillaz, there's a testcase in the bug [17:16] mterry: # ? [17:16] kenvandine: dunno [17:16] smspillaz, bug 732997 [17:16] Launchpad bug 732997 in unity "Cannot open a window that starts iconified" [High,Confirmed] https://launchpad.net/bugs/732997 [17:16] mterry: oh, you mean the vala one ? [17:16] smspillaz, but there are some minor corner cases I wanted to test (like a minimized window starting in an existing window group) and such [17:17] smspillaz, yeah, that was my testcase [17:17] mterry: ok, so the vala one works as you expect [17:17] smspillaz, sweet [17:17] it just appears in the launcher then when you click it it unminimizes [17:17] yay [17:17] as for deja-dup [17:17] smspillaz, deja-dup in 11.10 right now is patched to avoid that code path, so it's not a good test case right now [17:17] can I test things with what's currently in ubuntu ? [17:17] mterry: I can edit the code [17:18] re [17:18] smspillaz, you'd have to recompile. What you could do is grab 19.91 instead of 19.92 [17:18] smspillaz, the old package that is [17:18] mterry: I can recompile, it sfine [17:18] chrisccoulson, your g-s-d g-d updates seems to work fine [17:18] won almost 1 second login time there [17:18] nice :) [17:19] smspillaz, OK. Then the code is in widgets/WidgetUtils.vala, search for iconify. There are two code blocks commented out which should go back to like it was [17:19] mterry: I used gentoo and arch linux for a few years before I started working for canonical, I'm not afraid of compilers :) [17:19] smspillaz, :) [17:19] ok [17:19] smspillaz, but... are you going to edit bzr or the Ubuntu package? [17:19] package [17:19] if the Ubuntu package, you'll have to edit the C code, as it doesn't recompile the vala [17:20] *shrug* whatever [17:20] smspillaz, so edit widgets/WidgetUtils.c and do the same search :) [17:20] :) [17:20] vala is goofy sometimes [17:20] smspillaz, thanks! [17:20] vala is a nice concept [17:20] but it's just ... a pain sometimes [17:20] right, switching back to 3d now [17:20] at least in my experience [17:21] mterry: so what's the thing I should do to make it exhibit the behaviour you want [17:21] smspillaz, I think in the past year, it has at least become a stable platform/compiler. Before, not so much [17:21] smspillaz, hmm [17:21] I'll do it with and without your patch [17:21] mterry: valac-0.14 broke a bunch of unity code [17:21] it's not stable :) [17:21] smspillaz, oh really? I didn't have too much changes to make for 0.14 [17:21] smspillaz, unity is still unity vala? [17:22] still "using" [17:22] vala is cfront/glockenspiel 2.0. [17:22] smspillaz, so the simplest case is running "deja-dup --backup --auto" [17:22] * Sweetshark hides [17:22] seb128: I think there was some old stuff tha twas [17:22] chrisccoulson, http://people.canonical.com/~seb128/seb-e6410-oneiric-20110920-8.png [17:22] chrisccoulson, http://people.canonical.com/~seb128/seb-e6410-oneiric-20110920-9.png [17:22] chrisccoulson, that's before and after your updates on my laptop config [17:22] smspillaz, but the other thing to try is first running "deja-dup-preferences" and then separately running "deja-dup --backup --auto" and seeing how well that works (it should be two windows in the same launcher then) [17:22] chrisccoulson, btw do you also see plymouth staying there for almist 5 seconds? [17:23] mterry: just to double check [17:23] smspillaz, and I think also... "deja-dup --prompt" [17:23] seb128, oh, that looks better [17:23] deja-dup --backup --auto won't start thrashing my disk will it ? [17:23] seb128, no, i don't see plymouth at all on my laptop [17:23] slangasek, ^ what info would be useful about "plymouth sitting there for 5 seconds" on my bootcharts? [17:23] smspillaz, if you haven't set anything up, it should ask for your password, setting the urgency bit on the window [17:24] slangasek, it's like half the boot time ;-) [17:24] smspillaz, but not showing itself until you click on the launcher [17:24] ok [17:24] seb128: 'ubuntu-bug plymouth' should gather up the important details [17:24] (dmesg, /proc/fb, lspci) [17:25] slangasek, ok, is there any log that could point if it's really plymouth "blocking" the boot during that time? [17:25] slangasek, or any boot option I could try, like nosplash? [17:25] seb128: ... if only compiz was using gsettings in oneiric :( [17:25] seb128: "nosplash" is not a boot option, you just leave "splash" off :) What does the boot chart look like? [17:26] smspillaz, well, I will be happy if it uses it next cycle ;-) [17:26] slangasek, http://people.canonical.com/~seb128/seb-e6410-oneiric-20110920-9.png [17:26] seb128: /var/log/boot.log should show what upstart jobs are happening at the time [17:26] (but without timestamps) [17:26] seb128: +1 [17:26] seb128: I don't know why it wasn't put in this cycle actually [17:26] must have been some confusiopn [17:27] smspillaz, don't start on that when didrocks is around ;-) [17:27] smspillaz, basically you guys didn't manage to give us a tarball before feature freeze [17:27] ah yeah [17:27] smspillaz, that and we freaked out of other issues like the stacking and the unity dialog and speed [17:28] smspillaz, and we wanted to focus on getting those sorted before adding features [17:28] *shrug* [17:28] well, next cycle ;-) [17:28] prepare for a codedump next cycle then [17:28] with some luck it will be ready early in the cycle [17:28] the list is getting pretty long [17:28] oh literally [17:28] september 22 [17:28] what list? [17:28] I am merging like 6 different branches [17:29] seb128: the list of stuff that's done but wasn't merged into this cycle for whatever reason :) [17:29] smspillaz, you better not merge all those in once [17:29] seb128: don't really have a choice [17:29] we should rather prioritize and see what we want to land next cycle to avoid a disaster again and have to revert everything [17:30] give me time to fix bugs [17:30] its never perfect on the first run [17:30] that's all I ask :) [17:30] anyways, lets not discuss this now, it's a touchy subject [17:30] mterry: one small caveat with minimized windows [17:30] smspillaz, [17:30] k [17:31] mterry: windows that a start minimized are not accessible in the "per window" mode of the alt-tab [17:31] because they have not been mapped [17:31] and as such we are not able to access their pixmap data [17:31] so I've filed a design bug about this [17:31] because right now the launcher will unminimize the last used minimized window [17:31] if the window is part of a group [17:32] (ditto alt-tab) [17:32] smspillaz, if I manually call map on the window, does that workaround that? [17:32] mterry: err, it depends on how gtk works really [17:32] smspillaz, I forget how it handles that [17:32] mterry: how this works on an X level is .. interesting [17:32] basically in order to make a window initially minimized [17:33] smspillaz, well, I can live with it not being in alt-tab [17:33] you set the initial_state member to IconicState and then map the window [17:33] the window manager will intercept that and then transition the window from withdrawn to minimized [17:33] but not actually map the window [17:33] if you map the window again [17:33] it unminimizes [17:34] smspillaz, hrm [17:34] mterry: basically, the only way to access such windows right now is by using the spread which will temp unminimize windows [17:34] (that's only in the case of window groups though) [17:35] but lets see what design says about this [17:35] because I think that for initially unmapped / minimized windows [17:35] smspillaz, that's fine. The chances a user will have the preferences open when an automatic backup starts is small [17:35] activating the launcher icon should unminimize all windows [17:35] err [17:35] unminimize all initailly minimized windows and show any withdrawn windows [17:37] mterry: ok so valac doesn't include comments in the code, hurr [17:38] mterry: can you point me to the bzr version :) [17:38] smspillaz, :) [17:38] smspillaz, https://live.gnome.org/DejaDup/GettingInvolved/Coding [17:38] smspillaz, you can skip the bit about adding the ppa:deja-dup-team/dev repository [17:39] we should not let things skip alt-tab ever [17:39] this reminds me that we should also stack focus denied windows at the bottom of the tree [17:39] seb128: it's impossible to implement it unfortunately === zyga-afk is now known as zyga [17:39] seb128: without brekaing applications [17:40] seb128: the application will still be alt-tabbable [17:40] smspillaz, we shouldn't make applications skip alt-tab there [17:40] then [17:40] but you just won't see the initially minimized or withdrawn window [17:40] because to xcomposite, its pixmap doesn't exist [17:41] smspillaz, oh, my comment was a side one, I hate how apport dialog are "lost" if you alt-tab to something which is over them [17:41] like not listed in the launcher [17:41] not listed in alt-tab [17:41] yeah that sucks [17:41] the only way to go back to those is to reduce or close whatever is in front [17:41] I don't get why we just don't list them as normal dialogs [17:41] well that's rather an application design thing than a wm one [17:41] though I'm suprised that apport is not matched to an application by bamf [17:47] seb128: what's the video chipset on this, OOI? [17:47] slangasek, i5 [17:47] the intel chip integrated to the i5 [17:49] hi seb128 [17:49] I just set up a 'puter with i7 [17:49] I installed the nvidia driver [17:50] hey rickspencer3 [17:50] I guess there is some way I can choose between integrated graphics and the nvidia chip [17:50] ? [17:50] * rickspencer3 noticed seb128 had i5 [17:50] my i5 doesn't have dual videos [17:51] nice, so I restarted and got Unity 2e [17:51] 2d, even [17:51] I'm sure there is a way, but it would be a question before suited for Sarvatt or tjaalton or RAOF or bryceh ;-) [17:51] seb128: so I'm not quite sure what this points to, but plymouthd is stuck waiting for something... it coincides with the lifetime of the 'plymouth' command above it, which is *probably* 'plymouth show-splash'... so udev has signalled that the video is up, plymouth tries to display to it, and then it has to wait for some reason [17:52] * rickspencer3 removes binary blob [17:52] seb128: booting with plymouth:debug and attaching /var/log/plymouth-debug.log would be the next step [17:52] slangasek, ok thanks, I will do that next time I restart [17:52] seb128: I definitely want to see the dmesg for this. Have you filed a bug? [17:52] rickspencer3, no hybrid graphics switching is not supported yet [17:53] slangasek, not yet, I was trying to figure if I'm right to blame it on plymouth before [17:53] rickspencer3, there's sometimes manual ways to switch via bios or other hw-specific mechanisms [17:53] bryceh, so ... I was running Unity 3d, np, I presume from the integrated intel chip [17:53] seb128: plymouth is just the messenger ;) but ubuntu-bug plymouth gives *me* the information needed to further triage [17:53] I run jockey, it says I need the nvidia driver for Unity 3d, I install it, no unity 3d [17:53] bryceh, I suppose there is a bug here somewhere? [17:54] rickspencer3, yep [17:54] rickspencer3, well, wishlist feature actually [17:54] also, not it offers "nvidia_current" [17:54] oh, also, I think the first dotted line suggests that this is all being ascribed to the kernel budget, not the plumbing budget [17:54] which corroborates the idea that it's a kernel driver delay [17:54] slangasek, ok, I will open the bug and add the log and ping you with the bug number ;-) [17:54] sounds fine :) [17:54] dinner time for now, bbl [17:56] bryceh, so, I'm confused, why does jockey tell me to install the binary driver to get 3d, and then 3d doesn't work? [17:57] rickspencer3, I'd have to check logs to say for certain, but typically is something like this: jockey looks at all your video cards and notices that you have an nvidia card supported by the binary driver, so it helpfully suggests installing it [17:58] however when you reboot, the kernel starts loading up kernel drivers for graphics, notes you have an intel card so loads up that kernel driver [17:58] mterry: hm so [17:58] bryceh, uh [17:58] mterry: do you actually request for the window to start iconic and then show it later ? [17:58] now X comes along and sees that you have parts from one driver loaded in the kernel (intel), and parts from another (nvidia), and so can't do its job [17:58] so, basically, I can't use my brand new nvidia chip with 11.10? [17:59] smspillaz, yeah, iconify()/show()/iconify(). Then I rely on the launcher to show it when the user clicks the icon [17:59] rickspencer3, well, there's sometimes manual ways to switch via bios or something, but no hybrid graphics switching through software is not available yet [17:59] dang [17:59] mterry: that seems like wrong behaviour to me [18:00] * rickspencer3 disables driver [18:00] smspillaz, OK, why? [18:00] oh hang on [18:00] bryceh, what if I don't want to switch? [18:00] I just want to use the nvidia card? [18:01] mterry: ok, I misunderstood the way gtk works [18:01] the example does iconify () show () [18:01] rickspencer3, you could try blacklisting the i915 kernel driver, some people find that does the trick [18:01] I had show () iconify () [18:01] that wouldn't work until the window was realized [18:01] smspillaz, right, I think if you did that, you might flash on screen for a sec [18:02] smspillaz, which is a way to work around previous compiz behavior [18:02] no, it just mapped unfocused [18:02] huh wtf [18:02] smspillaz, oh, that might be because WidgetUtils.c sets focus_on_map = false [18:03] right [18:03] but I just changed it to do iconify () show () like the example and it still died [18:03] well [18:03] by died I mean [18:03] mapped [18:03] but not minimized [18:04] smspillaz, can you pastebin the vala code? [18:05] sure, hang on [18:05] just to verify you've turned the copier on [18:05] mterry: http://paste.ubuntu.com/693929/ [18:06] * smspillaz greps gdk for XWMH.initial_state [18:06] smspillaz, in hide_background_window_for_shell, don't set focus_on_map = true. That was part of the workaround [18:06] er, false [18:06] I mean, don't set focus_on_map = false [18:07] Not sure if that will affect anything, but... [18:08] it shouldn't [18:10] mterry: maybe the unity shell detection code isn't working here [18:10] smspillaz, add a: warning("hello!"); to the code block then [18:12] This is step 1 of my evil plan to make smspillaz a deja-dup developer [18:12] Then deja-dup maintainer, as I slowly fade away [18:13] hey hang on, I was planning to do that window compiz! [18:13] *with [18:14] ahhhh *what* [18:14] ok [18:14] so for some reason it is in my $PATH [18:14] but not running the right one [18:14] mterry: anyways, I ran it locally and it works as expected now [18:15] smspillaz, yay! [18:15] :) [18:15] smspillaz, for all three cases (alone, window group, prompt)? [18:16] mterry: ok test results [18:16] for deja-dup-preference [18:16] it shows a separte launcher icon [18:17] (dunno if that's intentional, seems to be a different desktop file) [18:17] when you have both deja-dup --promp and deja-dup --backup --auto open [18:17] smspillaz, it is, but I thought in such cases it got bundled in. If it doesn't, that's fine too [18:17] clicking on the launcher icon deiconifies the first one, then once you've dealt with it clicking it again will deiconify the other one [18:18] smspillaz, if that is what you expect, I'm fine with it [18:18] +1 [18:18] smspillaz, thank you so much for testing [18:18] np [18:18] smspillaz, I'll back out my workaround then before 20.0 release [18:18] +1 [18:22] smspillaz: wooo, my stacking issues are fixed with the desktop ppa [18:22] great job! [18:22] no guaruntees, but thanks for the compliment [18:23] smspillaz, it does seem good here too [18:23] looks promising :) [18:23] jcastro: kenvandine: if you run with --debug and open tricky applications like gimp and libreoffice you'll see that its not perfect [18:23] but its fixed about 99% of the problems [18:23] so that's good [18:24] i am very happy that the dash and switcher appear on top again [18:24] makes it much easier to use unity :) [18:24] yeah but something else changed, my dash performance took a hit last week on nvidia [18:24] jcastro: it is because the dash is constantly repainting itself [18:25] we are looking into it [18:25] oh, I thought it was my driver, because that got updated [18:25] could be that too [18:25] jcastro: bolt a few fans to your laptop and use nouveau [18:25] :3 [18:25] * smspillaz does that [18:25] no literally, I bolted fans to my laptop [18:25] my fan is already always maxed out [18:38] yeah, the stacking issues seem to be fixed here too [18:38] but it's time to switch back to 2d now, where the icons in the dash aren't blurry :) === eeejay_is_afk is now known as eeejay === eeejay is now known as eeejay_is_afk [19:15] RAOF, bug 854986 is for you ;) [19:15] Launchpad bug 854986 in xorg-server "5 second delay between Xorg starting and the greeter starting (Dell Latitude E6410)" [Undecided,New] https://launchpad.net/bugs/854986 [19:15] sorry :/ [19:17] chrisccoulson, it's weird how different your chart is compared to the one I did earlier [19:17] chrisccoulson, is that on your new dell? [19:17] seb128, yeah, i'm not sure why it's so different [19:17] yeah, this is the new machine [19:18] chrisccoulson, so quite similar to my dell [19:18] you have a 5s before the boot start, modprobe is quite hitting the disk for you, xorg is very slow as well [19:19] yeah, you don't seem to see any of that :/ [19:19] slangasek, is it normal for modprobe to take this long - https://launchpadlibrarian.net/80443468/farnsworth-oneiric-20110920-9.png? [19:20] actually, i'm not sure who i should ping about that ;) [19:20] buy seb128 doesn't see that, and we have fairly similar hardware [19:20] **but [19:20] your disk is a ssd as well right? [19:20] seb128, yeah [19:20] chrisccoulson, well maybe try the kernel team [19:21] we miss Keybuk to bring up such issues ;-) [19:21] heh [19:23] chrisccoulson: it's certainly not *normal*, it's a bug [19:23] it's not the first such bug I've seen this cycle however [19:23] slangasek, any idea how i could debug it? [19:24] yeah, i've been looking at the desktop budget today, and killed a few bugs there already [19:25] it seems my bootchart shows a couple of really long delays :( [19:25] anyone here knowledgeable about gnome-power-manager? seeing something strange/annoying... turn the brightness down on laptop monitor, reboot, and then brightness is back to the max [19:25] chrisccoulson: file a bug on the kernel; that should pull in dmesg and udev logs I think [19:26] slangasek, ok, will do that [19:26] thanks [19:26] chrisccoulson: and somewhere buried in those logs, you should be able to see some odd gaps in the timings that will tell you who's to blame [19:26] achiang, i'm sure there's a bug about that somewhere. i don't think it's ever saved the brightness across reboots unless you adjust it in the control center panel [19:26] in any case, that's controlled by gnome-settings-daemon now [19:27] brb [19:27] chrisccoulson: yeah, i've found LP: #771561 === eeejay_is_afk is now known as eeejay [19:42] chrisccoulson, hum, I wonder if your g-s-d update is buggy [19:43] seb128, oh, what's not working? [19:43] when I log in docked with the laptop lid close the screen is set to mirror [19:43] seb128, what did it do before? [19:43] where usually it's my external monitor 1920 resolution which is set and the laptop screen off [19:44] which is the configuration I defined with the screen capplet [19:44] so it's like g-s-d was not loading my config on login [19:44] hmmm, i'll take a look. perhaps that's the gnome-desktop change [19:45] do you want me to downgrade gnome-desktop to see? [19:46] mterry, nice catch on the gedit bug, just curious but how did you figure that? [19:46] I tried to look at it earlier in the cycle but neither gdb nor strace were really useful [19:46] seb128, ah, it doesn't load my config here too [19:46] chrisccoulson, do you want me to try with gnome-desktop downgraded? [19:47] seb128, GEDIT_DEBUG=1 and looking at different output [19:47] chrisccoulson, you should create a test user so you wouldn't have to log out and in ;-) [19:47] mterry, oh, GEDIT_DEBUG, learning every day... [19:47] thanks ;-) [19:48] seb128, yeah, can do (try with gnome-desktop downgraded) [19:48] brb [19:56] re [19:56] chrisccoulson, yeah, downgrading libgnome-desktop-3-2 fixes it [19:56] seb128, thanks. i'll take a look at that then [20:10] seb128, ok, found the issue [20:10] chrisccoulson, great! [20:10] g-s-d expects me to set the GError appropriately when the backup file isn't found, which i forgot to do ;) [20:11] so it just skips over the whole lot [20:11] easy fix :) [20:22] seb128, ok, fix uploaded now [20:22] chrisccoulson, thanks [20:24] it does mean that you'll suffer the cost of XRRGetScreenResourcesCurrent, as that is called if you have a configuration file [20:24] i wonder if i can avoid that :/ [20:24] it seems pointless to fetch the screen information, and then refresh it again immediately afterwards [20:25] i guess it's better than suffering from 3 of them though ;) [20:27] chrisccoulson, well, I personally don't care much, if you saw my charts the xrandr call have very low impact on xorg on my config [20:28] weird that it takes over 1s for some other intel users [20:31] yeah, i'm fairly sure the additional refresh is completely unnecessary there [20:31] it would make sense when applying a config manually later in the session [20:31] but it's triggering 2 reprobes in a row still [20:47] tremolux, I just got bug 848676 on up-to-date software-center when I was removing packages [20:47] Launchpad bug 848676 in software-center "[MASTER] software-center crashed with DBusException in _convert_dbus_exception(): org.freedesktop.DBus.Error.UnknownMethod: Method "GetAll" with signature "s" on interface "org.freedesktop.DBus.Properties" doesn't exist" [Undecided,Fix released] https://launchpad.net/bugs/848676 [20:49] mterry: hmm, well that's just sad [20:49] mterry: what's your version? [20:50] mterry: .3? [20:50] tremolux, yeah [20:50] mterry: did you apport it? [20:51] tremolux, I did, but it said "this has already been reported" and pointed me at that bug [20:51] mterry: yeah, I think that one's bugpatterned [20:51] tremolux, easy for me to trigger. I seem to get it by just removing packages [20:52] tremolux, hmm, no, but queuing up packages to remove [20:53] mterry: I guess we need to reopen, would you mind adding a comment to the bug? [20:53] tremolux, OK [20:53] mterry: thanks [20:53] mterry: so, for you it's quite reproducible then? [20:53] tremolux, I got it 3 times anyway [20:55] mterry: the key might be the multiple removes, this is one of those bugs that we haven't been able to repro ourselves [21:01] time to find out where Xorg is burning 5 seconds on startup here [21:02] Every time I do anything involving xrandr, everything stalls for a while. I think it's more than 5 seconds though, so possibly completely unrelated. [21:13] soren, no, i don't think that's unrelated ;) === m_conley is now known as m_conley_away [21:33] chrisccoulson: have you tried the latest natty kernel by any chance? my probing related cpu spiking (probably what soren sees) starts between ubuntu's 2.6.38 and 2.6.39 kernel. [21:34] not sure though if i wouldn't be chasing a red herring if i'd bisect the kernel. [21:35] Sarvatt, bryceh: ^ [21:35] on the topic [21:40] chrisccoulson, soren vga probing? [21:40] bryceh, yeah, see bug 854101 [21:40] Launchpad bug 854101 in ubuntu-boot-speed "gnome-settings-daemon takes waaaaaaaay too long to start" [Undecided,Confirmed] https://launchpad.net/bugs/854101 [21:40] I'd be interested to hear if the delay = ${NUM_VIDEO_OUTPUTS} * 1 sec [21:41] 'night [21:41] soren: 5 seconds! thats crazy! randomly curious, is that on a thinkpad? [21:42] chrisccoulson, tail -f /var/log/Xorg.0.log and then reproduce, and see if you see it printing RANDR stuff during that period [21:42] chrisccoulson, x220? [21:43] bryceh, no, i'm on a dell here. the probing isn't too bad for me (~0.5s). the main problem is the amount of times we do it :) [21:43] bryceh: nope he's on a dell, his numbers dont seem that crazy [21:43] i'm more concerned about bug 854986 on my laptop though ;) [21:43] Launchpad bug 854986 in xorg-server "5 second delay between Xorg starting and the greeter starting (Dell Latitude E6410)" [Undecided,New] https://launchpad.net/bugs/854986 [21:43] Sarvatt: Yes, it is. [21:43] Xorg seems to take forever to start on my machine [21:44] bryceh: here it's a thinkpad t510. i see probes for lcd (internal), vga (nothing connected), hdmi1-3 (don't have those), dp1-3 (i have one - disconnected). [21:44] bryceh: ${NUM_VIDEO_OUTPUTS} as in possible outputs or active outputs? [21:44] chrisccoulson, yeah, I noticed that as well; crazy amount of repetitive repetition [21:44] soren, possible [21:45] I'll attempt to time it. Hang on. [21:45] Hm.. No. [21:45] soren, upstream (Intel) says they know of an issue where the VGA probing is unnecessarily slow, at least on certain hardware types, but it's uncertain if that's the case here [21:45] Just running xrandr takes slightly less than 4 seconds. [21:46] With 2 connected and 5 disconnected outputs. [21:46] Actually changing anything, though, is way slower. [21:47] I think there's a way to make it completely ignore specific outputs. dunno what the config syntax is though. but that would pinpoint if it really is the output probing [21:48] bryceh: http://paste.ubuntu.com/694052/ [21:49] bryceh: WEll, "xrandr --current" is nearly instantaneous. [21:49] soren: confirming. [21:50] --current [21:50] Return the current screen configuration, without polling for hardware changes. [21:50] ok, so that seems to support the hypothesis that it's the polling that's bugged [21:50] bryceh: This is kernel stuff these days, right? [21:51] htorque_, thanks. however I'm wondering if the configuration needs to happen kernel-side; do you find that xorg.conf change results in faster boot / randr ? [21:51] soren, correct [21:51] bryce: yes, a lot faster. [21:51] htorque_, ok thanks [21:52] bryce: it's basically the same picture as here if i put all my disconnected outputs on "ignore": https://launchpadlibrarian.net/80234431/oneiric-2.6.38-vs-3.0.0_crop.png [21:52] soren, would be interesting if you could add that same snippet and see if it also eliminates the lag (or a portion of it) for you [21:53] Sorry, which snippet was this? [21:53] bryceh: ^ [21:53] soren, bryceh: http://paste.ubuntu.com/694052/ [21:53] i haven't tested only VGA1 though - i did this for all disconnected ones. [21:53] htorque_, do you put that in /etc/X11/xorg.conf.d/vga.conf or similar? Or just in xorg.conf directly? [21:53] xorg.conf in /etc/X11 [21:53] bryceh: I'll give it a shot. [21:53] soren, thanks [21:54] bryceh: I'll repeat for all the disconnected ones, right? [21:54] htorque_, in those two images, is the ONLY difference the kernel version? (Same Xorg versions?) [21:54] fantastic, i've got some air crash investigation to watch [21:54] bryce: yes. i actually narrowed that down to 2.6.38-8.42 → good, 2.6.39-0.5 → bad. [21:54] soren, that would be quite interesting; I'm curious if it's just hte VGA that causes the slowdown, or the sum of all disconnected outputs [21:55] htorque_, sweet, excellent work. Has this been escalated for the kernel team yet? [21:56] bryce: no, i just finished all the rebooting fun. it's also happening with a daily upstream kernel from a couple days ago (from the ubuntu mainline ppa). [21:56] Oh, the mainline kernel builds are back in business? [21:58] htorque_, ok, no prob, would you mind filing a bug (ubuntu-bug linux) and I can handle getting it escalated to the kernel team [21:58] or actually [21:58] RAOF: oh, it was 201108310919 - so probably not :-( [21:58] htorque_, do `ubuntu-bug xorg`, which will get X logs that might be of use [21:58] I'll reassign to kernel after that [21:59] bryce: what about chrisccoulson's bug 854986? is this not the same? [21:59] Launchpad bug 854986 in xorg-server "5 second delay between Xorg starting and the greeter starting (Dell Latitude E6410)" [Undecided,New] https://launchpad.net/bugs/854986 [22:00] i don't think that's the same [22:01] if it is, then it's not any clients triggering the probing [22:01] well, just to be safe let's assume they're not. I can escalate both, and let upstream decide [22:02] bryceh: This is interesting. [22:02] bryceh: I tried disabling everything but LVDS1. xrandr was instantaneous. [22:02] chrisccoulson, lp #854101 is focusing totally on just gnome lag stuff? nothing I need to worry over? [22:02] Launchpad bug 854101 in ubuntu-boot-speed "gnome-settings-daemon takes waaaaaaaay too long to start" [Undecided,Confirmed] https://launchpad.net/bugs/854101 [22:02] Then I re-enabled VGA1 (I use an external monitor). Still instantaneous. [22:03] Then I re-enabled HDMI1. Now it's back to ~4 seconds to just run xrandr. [22:03] chrisccoulson: X does a probe during startup; if proping is taking ages, X startup will take ages. [22:03] bryceh, yeah, that's completely focused on what gnome-settings-daemon does before anything else starts [22:03] s/proping/probing/ [22:03] alrighty [22:03] RAOF, yeah, i'm not sure what it's doing. did you see my last comment on the bug? [22:03] lightdm also felt waaaay snappier when all the disconnected outputs were disabled. [22:03] morning RAOF [22:04] i found where it spends 2.5 seconds of its time [22:04] I'll see if HDMI1 is the culprit of it it's any disconnected output that makes it slow. [22:04] bryceh: should i undo the fix in that report? i'm also using this fixed g-s-d version (but found the kernel differences before that on a 'clean' system). [22:04] (i mean before reporting anything) [22:05] chrisccoulson: Does it actually do a modeset on X startup? [22:05] htorque_, hmm, might not matter, but to be completely scientific it would not be a bad idea [22:05] Wow. [22:05] For me, it's HDMI1. [22:06] I re-enabled everything but that, and now it's as fast as it was in NAtty. [22:06] RAOF, bug 854986 - i'm not sure if that has enough information to tell you that :) [22:06] Launchpad bug 854986 in xorg-server "5 second delay between Xorg starting and the greeter starting (Dell Latitude E6410)" [Undecided,New] https://launchpad.net/bugs/854986 [22:06] I don't have an HDMI port at all, though. It's an X200s. I think I'd have one if I had a docking station, but I don't. [22:06] soren: do you actually have a HDMI port? i don't... [22:06] ok :) [22:07] So it's no loss for me, but for someone with a dock, it'd suck to have it be disabled. [22:07] soren: No, you wouldn't have an HDMI port, even on the docking station. [22:08] * RAOF *has* a docking station; it's got exactly one video output: DP. [22:08] RAOF: No? [22:08] Oh. [22:08] The vbios is just being super-conservative. [22:08] soren, check your xrandr output; there are cases where you'll have "phantom outputs" that dont' actually have a physical connector hooked up [22:08] bryceh: Oh, the vast majority are fake. [22:08] Maybe lenovo will at some point create a docking station that has 3 DP outputs, a couple of HDMI... :) [22:09] bryceh: I only have two. One is the built-in display, and then I have a VGA port. [22:09] bryceh: but xrandr brags about DP1 DP2 DP3 HDMI1 HDMI2 [22:09] chrisccoulson, I'd also like to see you repeat the xorg.conf settings to narrow down if disabling HDMI or another output restores your speed [22:09] yeah [22:09] soren, there are ways to 'quirk' off these phantom outputs, but it requires kernel patching [22:09] soren, can you also file a bug report (ubuntu-bug xorg), so we can handle your case as well [22:10] bryceh: Sure thing. [22:10] I'm hoping that upstream can give us a general purpose patch, but it's possible they'll do quirking, which will be very hw-specific [22:10] bryceh, which settings are those? sorry, i haven't followed all of the discussion :) [22:11] chrisccoulson, soren, bryceh: http://paste.ubuntu.com/694052/ [22:11] chrisccoulson, stick that in your /etc/X11/xorg.conf (maybe expand to include other outputs), then restart X [22:12] ok, brb [22:15] bryceh: Bug #855124 [22:15] Launchpad bug 855124 in xorg "XRANDR operations very slow" [Undecided,New] https://launchpad.net/bugs/855124 [22:16] thanks [22:17] bryceh, probing seems quicker [22:17] i added all the disconnected outputs reported by xrandr -q to my xorg.conf [22:17] but it doesn't affect bug 854986 [22:17] Launchpad bug 854986 in xorg-server "5 second delay between Xorg starting and the greeter starting (Dell Latitude E6410)" [Undecided,New] https://launchpad.net/bugs/854986 [22:18] chrisccoulson, hmm [22:19] 1 second, brb [22:20] yeah, so probing went from ~0.5s to ~0.13s [22:21] chrisccoulson, interesting, but Xorg startup is exactly the same? [22:21] bryceh, yeah, it looks like it [22:21] bryceh, yeah, still ~5s [22:22] It seems like you've only got one monitor plugged in, right? [22:22] X shouldn't be setting a mode at all in that case. [22:22] RAOF, i've just got the laptop screen atm [22:24] ohhh [22:24] eDP1 [22:24] Which is why I'm surprised that setdesiredmodes is taking so long. [22:24] Yeah, eDP modesetting is not necessarily the fastest thing you could stick in your startup path :) [22:25] chrisccoulson, try connecting a monitor, and disabling everything but that (i.e. disable the eDP1 output) [22:25] RAOF, was just 1 week ago keithp was telling us the woe that is eDP [22:27] ok, i'll try that [22:27] And both nouveau and intel performed major surgery on their eDP stacks to make modesetting on the new MBAs work :) [22:27] need to go back upstairs for that though [22:27] bryceh: bug 855133 - i will see if it's only HDMI1 for me like for soren. [22:27] Launchpad bug 855133 in xorg "Xrandr probing slow" [Undecided,New] https://launchpad.net/bugs/855133 [22:27] htorque, ok thanks [22:27] soren, ok now on to your bug [22:30] ok, brb [22:31] * RAOF thinks chrisccoulson needs a second, dedicated IRC machine :) [22:33] bryceh, it was ~3.6s on that boot with eDP1 disabled [22:34] ok that sounds better [22:34] ideally I'd want to see ~1 sec [22:34] or 2 sec max [22:34] and it looks like all of that saving came from xf86SetDesiredModes [22:35] i patched my system here to log extra points :) [22:35] Yeah. [22:35] That would be because you're no longer modesetting the eDP. [22:36] bryceh: fwiw, it's also HDMI1 here. [22:36] the other big delay seems to be here: [22:36] [ 11.508] drmOpenDevice: open result is 12, (OK) [22:36] [ 13.231] drmOpenByBusid: Searching for BusID pci:0000:00:02.0 [22:36] ??? [22:37] Your machine is weird. [22:37] heh [22:37] chrisccoulson, final question - is this a regression? Did the boot speed work properly in natty or in oneiric with an earlier kernel version? [22:38] On my x200s the whole DRM open stuff takes less time than the resolution of the timer. [22:38] bryceh, good question. it's a new laptop [22:38] chrisccoulson, ok [22:38] i gave up on my old one because boot time was in the order of minutes ;) [22:38] ok that chalks up. I think eDP just sucks this way [22:38] can you return it? (kidding!!!) [22:38] heh [22:39] chrisccoulson, I've forwarded your bug upstream. Don't get hopes up this'll be fixed by release; from what keithp said, they're having serious troubles getting eDP to work right... the hardware design itself seems to be to blame [22:39] ah, ok. thanks [22:40] Yeah. There's lots of "set this thing and wait x hundreds of ms, then do this other thing, and wait again". [22:42] did I hear eDP? chrisccoulson, mind trying this kernel? http://kernel.ubuntu.com/~sarvatt/macbook-air/ [22:43] lots of huge delays peppered into the eDP modesetting code trying to make it work right because they didnt know what was going wrong, they're reworking it for 3.2 [22:43] ah, ok. i'll give that a try [22:46] I wonder what's happening in your 2 seconds of drm probing? [22:46] 1 second, brb [22:46] NOOOOOO! Unity's alt-tab switcher has regressed to being more annoying again. [22:47] GARGH! *Raise* the window you give focus to, please! [22:48] soren, ok, forwarded your bug report as well. Sounds like HDMI1 just needs quirked off and then it's good. Maybe they have an idea for a more general fix [22:48] I've added the ubuntu kernel team as well [22:49] Sarvatt, wow, that brings it down to around ~2s (with the eDP output enabled), and also fixes my ~5s modprobe issue too [22:49] chrisccoulson: argh [22:49] 15 seconds to a usable session now, including entering my password [22:49] that's impressive! [22:49] chrisccoulson: Does it also fix your crazy 2 second pause during drm loading in Xorg.0.log? [22:50] hang on a minute, i've got to reboot again, just to see it more than once [22:52] chrisccoulson: its about 100K worth of patches on top of oneiric's kernel, hearing that makes me think its really needed but way too much to SRU at this point so I dont know wth to do [22:52] RAOF, i still get a pause during drm loading (~1.2s) [22:53] yeah, i'm not too sure either [22:53] Working modesetting is worth an extra 10sec boot time. [22:53] it certainly would be nice to claw back some of the startup performance that we've lost over the last couple of cycles [22:54] http://ubuntuone.com/2zvoxdrCCAdn2K42yBHHdI [22:54] not sure if that works yet ;) [22:54] Working modesetting *without* the extra 10sec boot time could wait until P, I guess. [22:54] Sarvatt, is that a backport of upstream git or something? [22:55] so, that link i shared is my *new* bootchart :) [22:55] i'm really happy with that ;) [22:55] that's taken around ~7s off it [22:55] chrisccoulson, well in your case seems we've narrowed it to eDP, which is new HW that's just plain badly supported, so not the source of the lossages ;-) [22:55] bryceh: keithp got a macbook air which is one of the few eDP machines and is trying to make it work, and in the process rewrote eDP modesetting quite a lot but its still not 100% yet [22:56] And nouveau's rewritten their eDP support likewise. [22:58] chrisccoulson, upstream has a question on it- [22:58] Can we get an strace of X starting along with a perf report? Something like [22:58] strace -t -o x.strace X -ac [22:58] perf record -f -g -a X -ac && perf report | cat > x.perf [22:58] hey robert_ancell TheMuso RAOF and bryceh meeting time. [22:58] https://wiki.ubuntu.com/DesktopTeam/Meeting/2011-09-20 [22:59] jasoncwarner_, heya [22:59] hey bryceh , how are things? [22:59] bryceh, it doesn't need to be a cold start does it? [22:59] We interrupt your regularly scheduled bootperformance hackfest to bring you the Desktop meeting :) [22:59] heh [22:59] https://launchpadlibrarian.net/80459240/farnsworth-oneiric-20110920-16.png will make jasoncwarner_ happy ;) [23:00] jasoncwarner_, going good. raof and I had a good time at the X conference last week, lots of good convo's [23:00] chrisccoulson, nice! [23:00] Hey folks. [23:00] kenvandine, that's thanks to RAOF, bryceh and Sarvatt ;) [23:00] chrisccoulson: ~15s? i do like that ! [23:00] jasoncwarner_, yeah, it was >25s yesterday [23:00] now get that with unity-3d :) [23:01] chrisccoulson: that is exactly the trend we need ;) [23:01] just getting back into the swing of things today. Fixed (I think) one of the release bugs pitti posted about today, and trying to tackle some X boot speed issues (slow output probing), which seems HW-specific so far [23:01] bryceh: nice... [23:01] let's get the meeting cranking [23:01] chrisccoulson, very impressive [23:01] [TOPIC] X update [23:02] oh... as above. [23:02] RAOF, anything to add? [23:02] chrisccoulson: dont get too happy at that, the code was literally rewritten in the kernel yesterday and still isnt working properly so is some time off and is only fixing up your eDP machine (there are like 10 eDP i915 laptops total in the wild) [23:02] Did some triage of Xserver bugs; it looks like a bunch of them were accidentally fixed by the 1.10.4 upload. [23:02] Sarvatt, that's ok, that works for me ;) [23:02] bryceh: ok...thanks... [23:03] I'm pretty sure I've got bug #831867 nailed down, and just need to test. [23:03] my i915 LVDS machines with SSDs have <10 second boot times here the whole cycle [23:03] Launchpad bug 831867 in sane-backends "colord assert failure: colord: browser.c:701: avahi_service_browser_free: Assertion `*_head == _item' failed." [High,In progress] https://launchpad.net/bugs/831867 [23:03] http://www.bryceharrington.org/Arsenal/Reports/ubuntu-x-swat/totals-oneiric-workqueue.svg looks nasty, but I thinkj that's mainly a slew of False GPU lockup bugs we can ignore [23:03] RAOF: oh, nice! [23:03] (sorry for interrupting the meeting) [23:03] colord was causing some weird startup slowness, correct? [23:04] oh, I got a solid lead on the false gpu lockups chatting with Intel. but will probably need to wait until next cycle to experiment with [23:04] let's move to LightDM/Unity-Greeter... robert_ancell ? [23:04] No; the gnome-settings-daemon color plugin was doing xrandr probes on startup, and that's expensive. [23:05] jasoncwarner_, I'll try and get one more release in before beta2 to fix some minor issues. The important issue seems to be crashing in the greeter due to the indicators being threaded. I've changed them to load sequentially in a thread and it seems more stable now [23:05] robert_ancell: cool...ok... [23:06] robert_ancell: the greeter you had me test that seems to help with my laginess, is that released generally yet? [23:06] jasoncwarner_, yes, that should be the latest one. The most recent logs seemed to indicate it was round trip time to X that was the problem, and the stuff people have been finding in gnome-settings-daemon seems to confirm that [23:07] The very last one I sent you had g-s-d disabled, which indicated that is probably the cause [23:07] unfortunately we can't do that in general ;) [23:07] robert_ancell: yeah...I was just getting lag again so was wondering what was up...now i know! :) [23:08] robert_ancell, bryceh and RAOF you guys have clues on what is going on in U-G with X for a possible fix? [23:09] the logs showed a big initial delay, which is probably xrandr blocking the X server. [23:09] g-s-d causes boot slowdowns? What a surprise! [23:09] U-G? [23:10] bryceh: unity-greeter...I've taken to the custom of shortening everything ;) [23:10] The ongoing lagginess was shown in the mouse right, which indicates the X server not keeping up [23:10] well, we have been suspecting output-probing related issues, perhaps a regression with vga probing, but so far haven't pinpointed it to that [23:10] RAOF, you could reproduce it right? [23:11] Ok...sounds like we have some clues, but nothing too exact yet...thanks. [23:11] chrisccoulson has isolated some X-related delays to be due to g-s-d over-querying randr, thus exacerbating the probing issue [23:11] robert_ancell: Yeah. I sent you the traces. [23:11] TheMuso: care to update on a11y, sound and all that good stuff? [23:11] i.e., it's slow to probe to begin with, and then GNOME requests re-probing multiple times [23:11] Something was probing xrandr, causing the delays. [23:12] But I'm not sure what, because the xtrace of unity-greeter didn't show RANDR requests. [23:12] Ok, we should finally be over all the a11y related upgrade issues, pitti and myself spent some time on Monday afternoon sorting that out, so that should be good now. [23:12] bryceh, 6 times on the default install :) [23:12] and that's before anything else in the session begins to load [23:12] I've had reports from users saying that the stuff I recently added to make using a11y on the live CD is not working, but I suspect they are either not doing something right, or are not using a recent daily, or are using beta 1. Need to run through things myself once again to be sure, hope to do that today/tomorrow, so I can get any fixes in post beta 2 if they are needed. [23:13] i added my findings to https://wiki.ubuntu.com/DesktopTeam/11.10/BootSpeedAnalysis for anyone who is interested [23:13] Still need to fix accessibility for notifications, as they are broken. May file a bug against notify-osd to make sure its known by DX, in case someone like Alehandro has time to look at it... [23:13] But I am going to try and fix it myself if I can. [23:14] Audio wise, a few more bug fixes post beta 2, but nothing major. [23:14] So things are starting to wrap up on my end. [23:14] TheMuso: how is accessibility in ubuiquity? I believe you said you needed to do some more testing there last week ? [23:15] jasoncwarner_: Yeah its usable, needs more work cleaning up widget names, but thats one goal for 12.04. A user can still get through and work out what is what and get the system installed. [23:16] TheMuso: ok...cool...would you say that, using Unity-2d, we have a fully accessible desktop? [23:16] Not yet, still waiting on more bugfixes from the 2d guys [23:16] But the only critical thing that is missing is the quicklists and no access to the panel. [23:17] The dash is mostly usable, and hopefully will be more so after next week's upload. [23:17] TheMuso: ok...sounds florian is making 'em a priority. let me know if it looks like something si going to fall short so we can get it reprioritized. [23:17] thanks! [23:17] ok... [23:17] ok [23:17] [TOPIC] Any other Business? [23:18] anyone want to add anything ? [23:18] nope [23:19] alright...thank everyone [23:19] Nor here. [23:19] [END MEETING] [23:19] thanks [23:20] bryceh, hey, does failsafe X work now? LightDM now exits if it can't start an X server / greeter but I'm not sure if the code is there to trigger failsafe X from upstart [23:20] robert_ancell, the code should be there, but I haven't had a chance to test it again since before XDC [23:21] bryceh, ok [23:21] robert_ancell, at least the udev rule ought to be firing [23:22] robert_ancell, I plan to play with it some more this week once I'm though some of these release bugs [23:23] ok, let me know if lightdm behaves badly [23:23] robert_ancell, right will do [23:27] htorque, ok tackled your bug report too - I've sent it upstream with your comments [23:29] bryceh: thanks! :) [23:29] htorque, yours and soren's look very similar, possibly dupes. but will let upstream decide. [23:31] bryceh: yeah, thought so. also, we are both using thinkpads (though not the same models). [23:36] robert_ancell: planning to go for a beer on some evening during hackfest? :) [23:37] robert_ancell: hi! fyi, with bug 855133 worked around, bug 828112 is fixed for me. no lagging whatsoever. [23:37] Launchpad bug 855133 in linux "Xrandr probing slow" [Undecided,Confirmed] https://launchpad.net/bugs/855133 [23:37] Launchpad bug 828112 in unity-greeter "Password field feedback slow at times" [High,Triaged] https://launchpad.net/bugs/828112 [23:40] cyphermox, which hackfest (though the answer is almost certainly yes) [23:40] htorque, yes! Can we get jasoncwarner_ to test that too? [23:41] robert_ancell: beginning of October in Montreal [23:41] robert_ancell: update and test? I rebooted yesterday and still had it (will reboot again here shortly) [23:41] cyphermox, you'll be there? then of course! [23:42] robert_ancell: I live a few minutes away, and work a few days a week from the Montreal office :) [23:43] sweet [23:47] jasoncwarner_: are you using a thinkpad by any chance? [23:47] htorque: I am...x220 [23:48] htorque: and I've heard of others using the 220 that see the same thing... [23:48] htorque: are you using thinkpad? [23:49] jasoncwarner_: great! can you follow the description in bug 855124 and create or add that to your /etc/X11/xorg.conf? - yes, i'm using a thinkpad too (T510). [23:49] Launchpad bug 855124 in linux "XRANDR operations very slow unless (phantom) HDMI1 disabled" [High,Confirmed] https://launchpad.net/bugs/855124 [23:50] after that you hopefully shouldn't get any more lagging at the unity-greeter. [23:53] htorque: where is the xorg.conf? I don't have one in /etc/X11 [23:53] in that case create one [23:53] (at /etc/X11/xorg.conf, yes) [23:54] ok [23:54] that is kind of what i was wondering :) thanks [23:55] htorque robert_ancell be back in a few...testing it...