[05:46] good morning desktoppers [05:57] Morning oSoMoN [06:01] hey duflu, how is it going? [06:01] ’lut seb128 [06:10] * oSoMoN → school [06:18] oSoMoN, sorry, too much going on.... Good, you? [06:30] morning [06:39] good morning [06:42] good morning desktopers [06:43] lut oSoMoN didrocks jibel [06:43] hey duflu [06:43] happy friday! [06:43] how is everyone today? [06:43] happy friday guys! [06:43] Morning seb128, happy Friday [06:43] good, will suffer from warm weather for a couple of days and looks better after that :) [06:43] + didrocks [06:44] hey duflu ;) [06:44] nice === maclin1 is now known as maclin [07:11] duflu, was it you who filed a bug because you cannot log into the unity session after logging into the gnome session? [07:11] jibel, yeah. Not sure if it's now fixed? [07:11] duflu, no it is not, do you have the bug #? [07:11] Actually can't log into any session (on the second attempt) [07:12] jibel, https://bugs.launchpad.net/ubuntu/+source/lightdm/+bug/1700485 [07:12] Ubuntu bug 1700485 in lightdm (Ubuntu) "Second login always fails. I have to reboot between logins" [High,New] [07:12] I can log into gnome several times but cannot open a unity session [07:12] thanks [07:12] jibel, I think that was a differnt bug [07:12] jbicha had a fix, I think? [07:12] * duflu looks [07:12] duflu, right, and I'm using gdm not lightdm [07:14] I can't find the second bug I mentioned [07:14] I think it was Xwayland failing to start [07:15] Might be (should be) closed already [07:16] On a slightly related note, this looks neglected: https://bugs.launchpad.net/ubuntu/+source/xorg/+bugs?orderby=-id&start=0 [07:16] (also server failures should be logged against xorg-server :) [07:19] Does anyone have any opinions on the preparation of this? https://git.launchpad.net/~ubuntu-audio-dev/pulseaudio/commit/?h=ubuntu [07:50] morning all, happy Friday! [07:52] hey willcooke, happy friday! [07:54] hey seb128! Late night or early morning? [07:54] :)_ [07:55] or both ;) [07:55] lol [07:56] lol but sort of true ... :p [07:56] late night was my doing, I decided to use a bit the time where everything is quiet to get some work and other things done (and watched some tennis on TV) [07:57] Wimbledon? [07:57] then also had an early morning, but that one was not my choice :p [07:57] yes [07:59] going to be through for Andy this year :p [07:59] ha, yeah [07:59] he's had his turn [08:05] hey willcooke [08:09] morning didrocks, thanks for the email - that's great - perfect solution I think. Sorry for being lazy :) [08:11] willcooke: no worry! I *think* that's the user story which makes sense regarding to our past upgrade story [08:13] It's a tricky one, right? Because we need people to move to the new shell, otherwise what's the point. But also - we don't want to freak people out and leave them stranded with something they can't (or won't use). So I think your solution is the right one [08:20] duflu, same here, too much going on… but I’m doing good otherwise :) [08:43] seb128: while I'm touching u-c-c, I noticed you didn't remove unity-control-center-faces, doing it as g-c-c-faces provides the exact same one + tweaking the dep, alright? [08:53] didrocks, k, it was on my list for today but if you are at it feel free to do that as well [08:54] seb128: yeah, I'm adding the additional recommends on deja-dup at the same time [08:56] great,thanks [08:57] let me know if you want me to approve the mp [08:57] unsure if we still go through landings for u-c-c? [08:57] hum, I was going to push directly as it's only small packaging change, wdyt? [08:57] wfm [08:57] 2 different commits for deja-dup and faces though please :p [08:57] (deja-dup building FYI) [08:57] yep [08:58] already that :) [08:58] great :-) [08:58] seb128: just pushed it if you want to have a look [09:02] didrocks, looks good! [09:08] seb128: all archs are built for deja-dup new package. So, it's in NEW when you get a chance… [09:09] didrocks, already done [09:09] do you have something that push notifications for you? [09:09] or did you hit refresh every 10s? [09:09] (I was doing that waiting for the arm64 debs to hit the queue, build was done when I looked at it but the debs were not there yet) [09:09] seb128: I had the web page with all arch builds opened [09:09] :) [09:09] it auto refreshes? [09:10] https://launchpad.net/ubuntu/+source/deja-dup/34.4-0ubuntu3 [09:10] yeah, the arch part does [09:10] oh ok, nice [09:10] good to now [09:10] didrocks, thanks [09:10] yw, thanks for your review! [09:10] * didrocks pushes u-c-c thus [09:11] once publish, I'll demote u-c-c source and binaries [09:11] \o/ [09:11] (all the rest still on the page are already demoted if their are source + bins) [09:12] now if only we had working iso builds :p [09:14] i pinged the release team [09:15] sil2100, ^ any chance to have a look today or you prefer to wait for infinity ? [09:15] sil2100, today's isos fail in the same way than yesterday [09:18] jibel, can you remind me where are the logs for the iso build? [09:19] seb128, https://people.canonical.com/~ubuntu-archive/cd-build-logs/ubuntu/artful/ [09:19] and the error [09:19] E: No packages found [09:19] dpkg-deb: error: error reading archive magic version number from file /srv/cdimage.ubuntu.com/ftp/: Is a directory [09:19] duflu, bu 1702861 [09:20] bug 1702861 [09:20] bug 1702861 in gnome-session (Ubuntu) "unity session sometimes fail to start" [Undecided,New] https://launchpad.net/bugs/1702861 [09:20] I didn't find any useful data [09:21] where are gdm logs? (anything more verbose than the line in the journal) [09:22] jibel, you really meant "unity" in that report? [09:22] seb128, yes [09:22] no such issue with the GNOME session? [09:22] seb128, no it works fine [09:23] seb128, but frequently unity fails [09:23] unsure for the gdm logs [09:23] seb128, I've both session installed + gdm and all updates applied [09:23] I would expect them to be in the journal [09:24] yeah but the only line I've in the journal is [09:24] juil. 07 10:56:02 ubuntu gdm3[855]: GdmDisplay: display lasted 0,497889 seconds [09:24] not very useful [09:25] and /var/log/gdm3 is empty [09:27] jibel, I found some critical assertions in my logs of bug 1700485. Not sure if they are fatal though [09:27] bug 1700485 in lightdm (Ubuntu) "Second login always fails. I have to reboot between logins" [High,New] https://launchpad.net/bugs/1700485 [09:30] Probably not related, but just as problematic: gnome-shell's top crash seems to be caused by Xwayland dying (which is also fatal for gnome-shell on Wayland): https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1505409 [09:30] Ubuntu bug 1505409 in GNOME Shell "gnome-shell crashed with SIGTRAP in x_io_error() from _XIOError() from _XEventsQueued() from XPending() from gdk_check_xpending() ["Connection to xwayland lost"]" [Critical,Confirmed] [09:31] I've dbus errors in xsession-errors [09:31] dbus-update-activation-environment: error: unable to connect to D-Bus: Failed to connect to socket /run/user/1000/bus: Connection refused [09:31] Failed to connect to bus: Connexion refusée [09:33] is that after login in a session and out? [09:34] not sure. They don't change when a login fails so it might be unrelated [09:34] (they = the log files) [09:37] jibel, try maybe to edit /etc/gdm3/custom.conf and un-comment the debug enable=true [09:39] not sure if that's the same issue but I see something to duflu on my testing vm which is still on lightdm [09:39] login into GNOME works fine [09:39] log out and trying to log in again fails [09:39] ok [09:45] seb128, https://bugs.launchpad.net/ubuntu/+source/gnome-session/+bug/1702861/+attachment/4910785/+files/journal_debug.log [09:45] Ubuntu bug 1702861 in gnome-session (Ubuntu) "unity session sometimes fail to start" [Undecided,New] [09:45] there are dbus errors [09:45] and this [09:45] 1000 juil. 07 11:41:06 ubuntu gdm-password][10736]: GdmSessionWorker: jumping to VT 1 [09:45] 1001 juil. 07 11:41:06 ubuntu gdm-password][10736]: GdmSessionWorker: couldn't finalize jump to VT 1: Appel système interrompu [09:47] seems the debug option works :-) [09:48] it's not due to the login manager at least on my vm [09:49] I get the same issue if I "startx -- :1" [09:49] first time works [09:49] if I close and retry session fails to open [09:50] seems dbus is failing for some reason [09:55] OK, I give up. Birthday presents to wrap, and dinner to cook [09:55] o/ [09:57] * jibel relocates biab [10:04] ah [10:04] my intuition was good [10:04] downgrading gnome-session to 3.24.1-0ubuntu4 fixes it [10:05] it's a side effect of https://launchpad.net/ubuntu/+source/gnome-session/3.24.1-0ubuntu5 [10:05] " * Backport patches from 3.25.3 to kill D-Bus clients on log out." [10:05] jbicha, ^ [10:06] jibel, ^ unsure if that's your issue as well but might well be [10:20] jibel, would be interesting to know if your unity issue is also happening on a fresh boot [11:39] seb128, let me try [11:45] seb128: could you talk to halfline about that issue then, he was thinking about pushing the gnome-session workaround to Fedora 25; it's already part of Fedora 26 which will release in a few days [11:46] woot, indicators are off the iso [11:46] not indicator-application though [11:47] hey jbicha [11:47] jbicha, I commented on the bugzilla saying that [11:49] I guess libappindicator3-1 shouldn't recommend indicator-application then [11:50] seb128: your comment says "unable to log into GNOME again", but I thought it was Unity that had the problem? [11:51] jbicha, I didn't try unity, I'm saying that issue with a stock artful install and GNOME [11:51] saying->seeing [11:53] I can definitely log out and log back in here (that is part of the test case for the GOA bug) [11:54] we should do those "suggests" for indicator-application and indicator-messages if we can't swap off the deps, indeed [11:54] jbicha, well, I can't, so there is a bug that doesn't impact all users that doesn't mean it's not a bug :p [11:55] didrocks, yeah [11:55] hum, something has been pulling kerneloops-applet on the iso recently [11:55] do we want that on? [11:56] I don't think we want multiple crash report experience, do we? [11:56] seb128: yes, we just have to figure out what's different about your system than mine ;) [11:56] also, i wonder if the recommends from gdm3 to xserver-xephyr has any use [11:57] didrocks, right, I'!u [11:57] (I thought that would have been analyzed for the MIR) [11:57] didrocks, right, I'm unsure what's the status today of kerneloops though, are those handled by apport? [11:57] or just not handled? [11:57] did the MIR review got done? [11:58] security basically waived it in [11:58] well, when writing it, the reporter should look at depends/recommends [11:58] and ken wanted to bounce back to Mathieu but I don't think that was done at the end? [11:58] seb128: unsure about kerneloops [11:58] no, because for me he did +1 and push that to security [11:58] (the MIR team does a pass first, to ensure that security only has to review the code) [11:59] seb128, new image http://cdimage.ubuntu.com/daily-live/20170707.1/ [12:00] kenvandine: you didn't handle the Ubuntu GNOME -> ubuntu transition if I'm correct, right? [12:00] jibel, right, I saw, that's part of what we are discussing (indicators are off, nice!) [12:00] kenvandine: iirc, I added that to your card as the ubuntu GNOME team asked to transition to the ubuntu session [12:00] and you can blame didrocks for the iso not building after all :p [12:00] heh :) [12:01] only people doing things break things I guess :p [12:01] yeah, I was thinking that as typing :-) [12:01] true statement [12:01] * seb128 hugs didrocks [12:01] * didrocks hugs seb128 back [12:01] hum, I don't find interesting references to xephyr on gdm3's changelog [12:01] good work, quite some cleanup got done this week and things went without issues for the most parts [12:01] thx ;) [12:02] * didrocks looks at the debian package, just in case [12:02] I don't really see why xephyr would be useful there... [12:02] neither do I [12:02] didrocks: kenvandine: let's save the Ubuntu GNOME transition for next week, I want to discuss that with darkxst first [12:02] jbicha, that "system" is a daily artful install a bit less than a week old with no tweaking [12:02] jbicha: ok :) [12:03] jbicha, so pretty much as stock as you can get, it's in virtualbox if that makes a difference (but shouldn't for dbus) [12:03] ok [12:03] anyway, let's see if upstream has some debug hints [12:04] # Use Xephyr if it is available. It works better than Xnest since Xephyr [12:04] # supports the Xserver extensions, even if on a remote machine. [12:06] * didrocks reboots quickly [12:07] seb128, I cannot reproduce on a fresh boot, but as soon as I login/logout from gnome, I cannot log into unity anymore but can still log into gnome [12:26] didrocks, too much clean up bug 1702892 ? [12:26] bug 1702892 in ubiquity (Ubuntu) "Ubiquity does not use the Ubuntu font" [Undecided,New] https://launchpad.net/bugs/1702892 [12:28] jibel: interesting, I wonder if ubiquity doesn't use unity-settings-daemon and we don't have anything running in the ubiquity session anymore [12:28] jibel: that would my only explanation, nothing for fonts have been demoted [12:28] jibel: if you run the live session, and then, ubiquity, is it fine? [12:28] didrocks, and the installer fails to start [12:28] any logs? [12:29] didrocks, yes, 1 min [12:31] didrocks, http://paste.ubuntu.com/25039048/ and for bonus points a segfault o fupdate-notifier [12:32] hum, doesn't really help, mind trying live session & ubiquity? [12:32] I wonder if Gdk.Cursor is using some xsettings set by g-s-d/u-s-d and none of them are running, hence this no font/segfault [12:33] xnox: does it make sense to you? ^ [12:33] didrocks, that's what I did [12:34] ah, and the segfault is there? [12:34] yes [12:34] and non ubuntu font either? [12:34] didrocks, but that's 2 issues, 1 ubquity failing and 2. update-notifier crash [12:34] in the live session, like apps [12:35] hm apport-gtk crashes to [12:35] to* [12:35] too* [12:35] I guess all comes from the same issue [12:35] (at least, the crashes) [12:35] didrocks, ubiquity should start a settings-daemon and it has complex heuristics as to which daemons to start. let me check the code. [12:35] GDK can't get cursor [12:35] for whatever reason [12:36] I wonder why this happens though [12:36] oSoMoN: fyi for the LO i386 issue https://lwn.net/Articles/727206/ [12:38] didrocks, hm, so it should be activating gsd, but in the usd code path we have an extra signalwatcher and ubiquity-dm waits for the xsettings plugin to be activated before proceeding with the rest of the setup. [12:38] didrocks, syslog attached to bug 1702894 [12:38] bug 1702894 in ubiquity (Ubuntu) "artful desktop 20170707.1 - ubiquity fails to start" [Undecided,New] https://launchpad.net/bugs/1702894 [12:38] jibel: thx! do you mind making the manifest diff as you did the other day? [12:38] I'm starting a vm [12:38] imho all o fthis should go / shouldn't be necessory and simply normal gdm should auto-launch gnome-shell with ubiquity in it =/ [12:39] xnox: well, seems to be in the session as well from what jibel says, so could be unrelated [12:39] the "gdk can't get a cursor" sounds weird though [12:40] does gds still have plugins, or a way to wait for gds to fully initialise and e.g. create xsettings stuff /me wonders [12:41] didrocks, sure, will do [12:41] jibel: thx! [12:42] xnox: I guess in the live session, it is already fully started [12:42] maybe though we are in a wayland session by default… [12:42] now that gdm is default [12:43] * didrocks waits for a long build to finish and then spawn a vm [12:44] we aren't defaulting to Wayland yet anywhere [12:45] gdm has some heuristic I was told [12:45] but worth checking [12:45] oh, gdm3 does default to Wayland but the actual sessions should still be X by default, hmm [12:46] let's check on the live first [12:46] we haven't stopped the Ubuntu GNOME daily iso builds yet so you could check if things are working there [12:47] mpt: any news on the design? [12:47] didrocks, https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1702892/+attachment/4910862/+files/20170705-20170707.1-manifest.diff [12:47] Ubuntu bug 1702892 in ubiquity (Ubuntu) "Ubiquity does not use the Ubuntu font" [Undecided,New] [12:50] jibel: nothing too crazy or unexpected at first look, same for you? [12:52] jbicha, thanks for the link [12:53] jibel, when did that issue start? [12:57] I guess today and didrocks is already looking at it [12:57] seb128, with latest image. and it's issues (1. font in ubiquity-dm 2. ubiquity fails to start from the live session) [12:57] let me know people if you need more eyes to look at that [12:57] seb128: I would like a second opinion (didn't start the vm yet) on the live diff and error reported [12:57] * xnox hasn't touched ubiquity in a long time, nor gdm3 ever =/ [12:57] seb128, second bug is bug 1702894 [12:57] bug 1702894 in ubiquity (Ubuntu) "artful desktop 20170707.1 - ubiquity fails to start" [Undecided,New] https://launchpad.net/bugs/1702894 [12:58] jibel: so you confirm the font is only ubiquity-dm related, correct? [12:58] like if you open gedit or so in live [12:58] you have the ubuntu font? [12:58] didrocks, I don't see anything obvious in the manifest diff indeed [12:58] seb128: the crash looks weird, can't get a GDK Cursor [12:59] my theory on the font is that g-s-d isn't started by ubiquity for $reasons [12:59] didrocks, I would guess "ubiquity doesn't work under wayland" :p [12:59] and for the crash, wayland yeah [12:59] (so, due to gdm being too smart and autologin in wayland) [13:00] I guess xprop doesn't report anything on wayland? [13:01] to know if you are under wayland you mean? [13:01] yep [13:01] (I wouldn't trust env variable as we can have some scripts settings them unconditionnaly) [13:02] loginctl show-session should tell you [13:02] didrocks, yes it's only in dm [13:02] didrocks, the live session is fine [13:02] or echo $XDG_SESSION_TYPE [13:03] ok, will give it a try once the current build is done [13:03] I'm downloading the daily, a few more minutes [13:09] confirmed [13:09] ubiquity crashes only under wayland [13:09] (tried on an old installation, switched session [13:09] ) [13:09] wayland -> crash, x -> works [13:09] jibel: seb128: xnox: ^ [13:10] on the font in ubiquity-dm, I bet g-s-d isn't started and we don't have u-s-d anymore… [13:11] confirmed that daily iso boot to wayland [13:11] so, not a cleaning issue, upstream code issue and gdm transition :p [13:11] anyway, I think we should force X for now [13:11] loginctrl show-session says type=wayland [13:12] seb128: what's the session name? [13:12] I guess we'll need to poke in gdm smartness [13:12] seb128: and do you mind looking/starting g-s-d in ubiquity-dm? [13:12] for the font issue [13:12] DESKTOP_SESSION=ubuntu-wayland [13:13] sure, can look to that [13:13] interesting, I guess it prefers /usr/share/wayland-sessions/ if it finds anything? [13:13] I guess [13:13] oh we didn't set any default in casper… [13:13] didrocks, g-s-d is started in ubiquity-dm. but in u-s-d codepath there is an extra wait for the xsettings plugin to be started before proceeding. [13:13] that should have been part of the gdm transition card [13:13] xnox: ah, that's the issue… [13:14] but [13:14] didrocks, imho it looks like we are not waiting for enough of g-s-d to be running before proceeding to create things. [13:14] IIRC, once it starts, it refreshes the windows? [13:14] or e.g. u-s-d != g-s-d in some other way. [13:14] like in gtk, it's the themes and such [13:14] but not text, i would not think. [13:14] I thought there were a signalling mecanism [13:14] hum [13:14] yeah [13:14] easy way to know your session is a wayland one, alt-f2 "r" [13:14] that gives you an error under wayland [13:14] there are a lot of hacks to change language on the fly in ubiquity. [13:15] seb128: "r"? [13:15] reload?! [13:15] yes [13:15] it's an hack telling gnome-shell to restart [13:15] interesting [13:15] nice trick seb128 :) [13:16] xnox: do you think you will have time to work on this ubiquity thingy? (I think your analyze is correct) [13:16] no [13:16] i do not work on ubiquity. [13:16] who does? [13:16] desktop team? [13:16] =) [13:16] good one :p [13:17] sounds like the same story than software-center, update-manager and all this :p [13:17] honestly none of us touched gsd or gdm code paths. [13:17] it was always gnome team, which i guess now is desktop team. [13:17] e.g. laney would often propose fixes for integration updates like that. [13:18] I'm having a look [13:18] it might be easy [13:18] ideally we should not have ubiquity-dm at all [13:18] I'm looking at the casper/set default gdm session [13:18] and just do casper override or some such to have default session started normally with gdm, in the installer mode. [13:18] yep [13:18] quite agree on that one [13:19] xnox: do you mind dumping the u-s-d hacks/xsettings sync notes on jibel's bug [13:19] which one? the does not start - or the font one? [13:19] font one [13:19] the "does not start" should be separated [13:19] like "get ubiquity to work under wayland" [13:20] and "casper set default session to X for now" [13:20] jibel: FYI ^ [13:22] reboots & back [13:22] done [13:23] xnox, is it normal that I can't go to a vt in ubiquity-dm mode? [13:25] seb128, yes. [13:25] seb128, that would also be fixed iff we switch to gdm3 launching ubiquity in installer mode. [13:25] basically systemd thinks one is not booted yet, and currently ttys are not started by systemd as they are ordered to be _after_ ubiquity [13:26] i send a patch for that by at the p itti did not take it =( [13:26] if you boot with debug-ubiquity option [13:26] ctlr-alt-t should start gnome terminal for you.... [13:26] thanks [13:26] or e.g. modify ubiquity-dm to launch gnome-terminal process for you in addition to others [13:26] might be quicker to do that. [13:26] (just like it e.g. launches g-s-d et al) [13:27] ah [13:27] xnox, didrocks, that's not the ubiquity-dm issue [13:27] hum? [13:28] g-s-d got split in different binaries [13:28] so there is gnome-settings-daemon binary anymore [13:28] oh, wrong path in ubiquity-dm? [13:28] since https://launchpad.net/ubuntu/+source/gnome-settings-daemon/3.23.3-0ubuntu1 [13:29] ahhhhhhh, so it's working due to dbus call on the session [13:29] instead of repeating ourselves in ubiquity-dm.... we really should look into launching normal session of gdm3 in installer mode without ubiquity-dm [13:29] but not ubiquity which expects the binary? [13:29] no [13:29] each binary has an autostart desktop [13:29] so they start in the session [13:30] but yeah, same result [13:30] (btw, confirmed we don't set a default session in gdm casper script) [13:30] ubiquity-dm does [13:30] gsd = '/usr/lib/gnome-settings-daemon/gnome-settings-daemon' [13:30] elif osextras.find_on_path(gsd): [13:30] making sense, so just reintroducing that new package split on live would work [13:31] jbicha, did you look at updating ubiquity for the new g-s-d? [13:31] well, obviously it worked somehow for Ubuntu GNOME 17.04 [13:31] no [13:32] I guess you don't use ubuntu fonts for ubuntu GNOME? [13:33] no [13:33] jbicha, it worked fully or you didn't get the right font used in install only mode and didn't notice? [13:34] seb128: I think for casper, instead of setting a default session, I could use WaylandEnable=false, which "should" pick the X session [13:34] sounds like a more flexible approach to me, but needs testing [13:34] right [13:34] (at worst, we fix it on monday if we notice we are still using wayland despite this) [13:34] ok, doing this [13:35] * didrocks opens a bug for refering that, as different from ubiquity [13:39] yeah, I guess it used the wrong font but it's difficult for me to notice a detail like that :) [13:40] hehe [13:44] jibel: seb128: xnox: casper "fix" uploaded to force a X session (I referenced bug #1702904). So next image should force-start the X session making ubiquity not crashing under live [13:44] bug 1702904 in casper (Ubuntu) "Ubiquity is crashing under wayland" [Undecided,New] https://launchpad.net/bugs/1702904 [13:44] great [13:45] however ubiquity not working with wayland is concerning long-term [13:45] seb128: are you handling the g-s-d binary so that we can check the font is loaded on ubiquity-dm [13:45] (and just because I had to say it: "NOT related to any iso cleanup I did" :p) [13:45] lol [13:46] I can have a look if you want [13:46]