[01:04] robert_ancell: yep [01:05] andyrock, was just checking your gnome-initial-setup MP (I wasn't subscribed so I hadn't seen it). Do you agree? [01:07] robert_ancell: seb answered before [01:08] robert_ancell: upstream does not want this, by design they want "quit" to show the wizard again [01:08] robert_ancell: we can avoid creating the stamp file in other places [01:09] robert_ancell: but is it worth doing that? [01:11] robert_ancell: so I'll ask again upstream tomorrow but I can already guess the answer [01:13] andyrock, I think if we drop the desktop check the patch will be smaller which is good if upstream doesn't want that. [01:13] there's no penalty to writing them twice that I can see [01:14] robert_ancell: not sure I'm getting what you're asking me to do :) [01:19] andyrock, I mean, why have the is_desktop ("ubuntu") part? [01:19] Just always write the files on shutdown [01:19] Then this change can be carried separate to our changes [01:19] our pages I mean [01:20] mmm are we using the same gnome-initial-setup for other desktops too in ubuntu? [01:20] robert_ancell: I can do that. Tomorrow morning [01:22] vanilla GNOME is the only other desktop that uses it [01:22] jbicha: same package? [01:22] Budgie has its own welcome app and it would need some work to be usable in Unity [01:23] yes, same package. It shows different pages in vanilla GNOME [01:23] robert_ancell: that's why I used the is_desktop [01:23] Ubuntu GNOME 17.04 included gnome-initial-setup by default [01:23] to not change the beavhior in other desktops [01:24] andyrock, sure, I just feel like this is more of a "we consider users who cancel the dialog to have completed it" rather than "this is specific to the Ubuntu desktop welcome" [01:24] And that leaves us with a smaller patch to carry. [01:24] jbicha: what do you think? [01:24] I don't have a strong opinion on this [01:26] it doesn't matter to me [01:26] jbicha: kk thanks [01:26] robert_ancell: ok I'll drop the is_desktop [01:26] andyrock, thanks [01:27] let me try to do that now [01:27] I think this patch has highlighted how annoying the Appmenu is :) [01:27] Or is it the dock that people use to close it? [01:28] yeah, you can do it from both.. [01:29] I didn't test the dock [01:29] but it should work [01:35] robert_ancell: I pushed it as a separate commit [01:35] if you want I can rebase the all thing [01:35] andyrock, thanks! [01:36] also I didn't build it (I'm on xenial with ubiquity running on a vm and I cannot restart) [01:36] but the change was trivial enough [01:36] andyrock, It looked fine to me, we can always fix it if not [03:49] hello [06:15] Morgen thumper [06:15] hey [06:16] I forgot I had irc open [06:16] biab [06:21] good morning desktoppers [06:34] good morning desktopers [06:34] good morning seb128! [06:35] hey didrocks [06:39] salut seb128, didrocks [06:40] lut oSoMoN, en forme ? [06:40] salut oSoMoN [06:40] oui, et vous? [06:40] nickel [06:40] tout se passe bien à Berlin? [06:41] oui, notre review est faite, pas de surprise [06:44] tant mieux [06:48] good morning to all :) [06:49] hey c-lobrano! [06:49] hey didrocks, how are you? [06:49] good good, yourself? :) [06:50] I doing good :) [06:50] *I'm [07:09] Morning seb128, didrocks, oSoMoN, c-lobrano [07:10] hey duflu, how are you? [07:10] good afternoon duflu [07:10] seb128, going well. I'm pleasantly surprised to see the bionic bug rush is over (or not started?) [07:11] Wie gehts Berlin, seb128? [07:14] sehr gut [07:15] quiet time until we enable upgrades for 16.04 users at .1? :) [07:16] hey duflu [07:16] ah, of course [07:18] morning [07:19] hey Nafallo [07:19] Hi Nafallo [07:25] duflu, did you made any progress on the ubiquity issue? [07:28] seb128, not yet. I'm reviewing and retesting yesterday's theories. Might have got it backwards but on the right track [07:29] backwards with uids? [07:29] I asked andyrock if he could help yesterday afternoon but I'm unsure if he was able to reproduce/figured out anything [07:30] seb128, yes, when the installer is working ps shows its uid is root. When it's not working, it's dropped to 'ubuntu' [07:33] But that may be an incorrect observation, which is why I am checking again [07:33] duflu, are you running the installer always from the live session? [07:33] jibel, yes. I like to wipe the partition table manually first [07:37] ALthough that may be correlation, not causation [07:39] Yeah. Seems to be correlated. The problem occurs around the point where the installer switches from uid 999 to 0 [07:40] And verified again also that it's not the scale, but timing [07:40] Which would also explain why some reports of similar crashes appear in 16.04 [07:41] Maybe [07:43] we had this bug several years ago and completely disabled the slideshow but I cannot find the bug # [07:45] Oh! Cairo uses that call in 3 places. And only this one place fails to sync and handle X errors. [07:55] seb128, I think I might have a fix today... [07:55] nice [07:55] by making that 3rd call use the same error handling that the other 2? [07:56] seb128, yes. The caller already has a fallback. So then it should fall back and keep working [07:56] We shouldn't need to fix the race, only handle it [07:56] great [08:04] moin [08:04] something weird happened to my router [08:04] but internets is happening now [08:05] hey Laney [08:05] back from London? [08:06] yeah [08:06] and it is sunny! [08:06] what's up in berlin? [08:06] nice! [08:06] currywurst! [08:07] week is going well, no surprise so far [08:07] good [08:08] Trevinho, good morning? [08:09] seb128: hey seb... Yeah, good morning to you and rest of the team too [08:10] oh, Marco is there :) [08:11] hey Trevinho [08:11] hey Trevinho [08:11] wb to the proper side of the ocean [08:11] and Laney! [08:11] thanks Laney [08:12] and hi didrocks [08:12] what up didrocks [08:12] nothing special, swapping with my vms ;) [08:12] you? [08:12] enjoying being home [08:14] going to fish that old inspiron out to see if that gdm bug happens [08:14] I wanted to ask Trevinho if he can have a look tomorrow but he was not around [08:14] that sentence feels time travelly [08:15] now unsure who wants to own/poke at it, either of you guy would do [08:15] yesterday tomorrow was today but now tomorrow is tomorrow [08:15] doh [08:15] ahah [08:15] yesterday :p [08:15] I'm tired :p [08:15] well I said I'd get some more logs [08:15] not sure if that is owning [08:16] seb128: which one? I didn't see mentions [08:16] Trevinho, can you look if you can reproduce https://gitlab.gnome.org/GNOME/gnome-shell/issues/227 & help upstream to fix it if that's the case? [08:16] GNOME bug 227 in gnome-shell "Login fails when preceded by incorrect password" (comments: 16) [Opened] [08:17] anyone know anything about https://bugs.launchpad.net/ubuntu/+source/grub-installer/+bug/1767703 too? [08:17] it is collecting a lot of dupes [08:17] some problem installing grub-efi [08:17] urg [08:18] indeed you did... fkng IRC client. I got no notifications on queries [08:18] Laney, let me escalade to our foundations friends [08:18] jibel: ^- maybe you know something about that one? [08:18] like if foundations know :P [08:18] thx [08:27] looks like people like the communitheme snap CI integration with per-PR branches: https://community.ubuntu.com/t/call-for-participation-an-ubuntu-default-theme-lead-by-the-community/1545/986 and https://community.ubuntu.com/t/call-for-participation-an-ubuntu-default-theme-lead-by-the-community/1545/987 :) [08:28] :) [08:35] Laney, seems like it could be the same as bug #1766945 [08:36] raised with Patricia/Steve now [08:37] cheers! [08:37] thank you for mentioning it [08:39] would be good to have a way to update the installer without needing a new iso [08:39] to be able to fix this kind of thing before the point release [08:39] +1 [08:39] even just for development… [08:40] (especially in ubiquity-dm mode) [08:42] yeah [08:43] though we might need a new libcairo for the hidpi/xerror bug [08:43] but yeah, having the installer auto-updating upfront would be nice [08:44] omfg [08:44] the inspiron has decided that its battery is fake [08:44] and it tells me this by making a deafening noise [08:50] andyrock, where should I log a bug for the software and updates livepatch UI? When the g-o-a u1 account expires it shows livepatch as not being enabled [08:51] Laney, this is a dupe of bug 1766945 [08:51] Trevinho, I'm suddenly seeing a lot of "I can see the desktop when I resume from suspend" bugs, anything I can do while its doing this to help debug? [08:52] willcooke: oh, are we again on that? :) [08:52] :)))) [08:52] jibel: ok, that's the same one seb pointed to [08:52] Laney, it's mainly because they removed the warning saying that installing a EFI system without an EIF system partition will fail. [08:52] mh... I think we should try to do the same we were doing in unity... andyrock also has a good background on that [08:53] so basically painting a dark frame before giving the ok to suspend [08:53] need to see how we can do that in a nice way though [08:54] Laney, and for reference due to fixing bug 1668148 [08:54] Ahh, I wonder if me install the gs connect shell extension has made this worse. [08:55] where is the bot? [08:55] willcooke: can you please add a card / bug on that? [08:55] e,g https://pastebin.canonical.com/p/G5P4JFG8S9/ [08:55] Trevinho, sure thing [08:55] willcooke, so basically it's your fault :p [08:56] yeah pretty much [08:56] makes a change, right? [08:56] lemme disable it and see what happens. [08:56] :) [08:56] Laney: as per https://gitlab.gnome.org/GNOME/gnome-shell/issues/227 when attaching such logs also do a `call gjs_dumpstack()` [08:56] GNOME bug 227 in gnome-shell "Login fails when preceded by incorrect password" (comments: 16) [Opened] [08:56] Laney: see what I've added to https://is.gd/wiki_gnome_shell_crash_debug [08:58] Trevinho: I did, see the top of the log [09:01] jibel: ok, sounds like this error was unexpected tho [09:01] good that they're going to look into it [09:02] Laney: also those Apr 23 15:45:26 adam-thinkpad-t430 gnome-shell[5986]: g_dbus_connection_signal_unsubscribe: assertion 'G_IS_DBUS_CONNECTION (connection)' failed seem like it's the cause of one of the bigger crasher we have [09:08] willcooke, there is https://bugzilla.gnome.org/show_bug.cgi?id=753678 [09:08] bugzilla.gnome.org bug 753678 in lock-screen "Desktop temporarily visible after wake up from suspend" [Normal, Assigned] - Assigned to gnome-shell-maint [09:08] but that's not specific to the case where you disable autolock [09:09] willcooke, https://bugs.launchpad.net/ubuntu/+source/unity/+bug/1532508 [09:10] willcooke, https://bugs.launchpad.net/ubuntu/+source/gdm3/+bug/1741248 was about gnome-shell [09:10] but duflu duplicated it from the unity one [09:10] I think we should undup [09:10] less confusing [09:12] those kinds of bug will never disappear… [09:13] Ok, I've logged a new bug since my test cases are very specific: [09:13] https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1768786 [09:15] Trevinho, ^ can you look at that one once you are done with 227? [09:15] or maybe andyrock? [09:20] seb128: yeah, sure [09:21] thx [09:21] * Trevinho adds cards or his mind won't work [09:23] Trevinho, are you adding them to the 18.04 board for now? I'll get a new one set up next week if we can last without it for that long [09:23] willcooke: yeah, we can move them once we've a new one [09:23] kk [09:23] willcooke: although not sure we can continue using that for 18.04.x's? [09:27] we have a .1 column, stack here for the moment [09:29] seb128: ah, i see it... However I'd prefer to have that most on the right as it feels more natural now to move things from left to right when they proceed [09:31] Trevinho, right, we do that in the new board next week [09:37] Trevinho, I assigned the bug from will to you, but feel free to trade with andy if he wants to take it, please reassign if needed [09:38] \o/ and I screwed the *last* step of ubuntu-release-upgrader! [09:38] glad to have tested it :p [09:38] wth communitheme? https://launchpadlibrarian.net/368384621/Zrzut%20ekranu%20z%202018-05-03%2009-21-19.png [09:39] now fix -> rebuild -> redo a release upgrade [09:39] :) [09:39] seb128: told you it's not yet fully ready :p [09:39] ah [09:39] the list of channels [09:39] yes [09:39] well, as told, there is a channel for each PR [09:39] (see my blog posts and the posts I've sent above) [09:40] they autoclose in 30 days, it's not configurable [09:40] is that a bug in g-s that they are all named the same? [09:40] yes [09:40] and there aren't as much [09:40] willcooke: why the goa account expires? mmm [09:40] i am trying to open geogebra in 18.04LTS. it isn't working. program closes after welcome screen... [09:41] hum, I wonder how I can get the list from snapcraft… [09:41] when I release something, I get it [09:41] but without releasing… [09:41] willcooke: target gnome-online-accounts and whatever software you used to enable it [09:41] ah status… [09:42] seb128: here is the real list: http://paste.ubuntu.com/p/s5sQQ8nKVT/ [09:42] seb128, Trevinho iirc they are already showing a black frame before suspend [09:42] you can notice snapcraft double them though :/ all and amd64 [09:42] didrocks, what option to dump the list? [09:42] seb128: Trevinho it can be the kernel/intel bug we had in unity as well [09:42] seb128: well, you need to be the snap author and use snapcraft status [09:42] andyrock, not when privacy->lock is disabled apparently [09:43] the lock on idle [09:43] didrocks, ah ok [09:43] only "collaborators/developers/who knows" can [09:43] seb128: I'll givet it a look [09:43] andyrock, thx [09:43] i was checking the permission thing in ubiquity [09:43] so yeah, bug in G-S [09:43] or snapd [09:44] andyrock, duflu made progress/has an idea about a cairo fix [09:44] mmm i found a problem when we deal with permissions [09:44] andyrock: I meant the thing we did painting a GL frame, more than highlevel thing [09:44] there's a thing from Jonas though on that bug that might help [09:44] Trevinho: I copied the idea from them :D [09:44] I didn't check it yet [09:45] didrocks, thx, https://bugs.launchpad.net/ubuntu/+source/gnome-software/+bug/1768779 is the bug, can you maybe put your pastebin list on it for Robert? [09:45] seb128: doing [09:45] thx [09:46] andyrock, oh, can you give details? [09:47] so before we call misc.drop_privileges_save() we have: uid=0 and euid=999 [09:47] after we have uid=999 and euid=999 [09:48] after we call misc.regain_..._save() we have uid=999 and euid=999 [09:49] the all point of those functions is to restore the previous beavhior [09:49] I'm checking what happens in 17.19 [09:49] 17.10 [09:49] k, thx [09:49] that sounds weird that it fails to restore [09:50] seb128: btw, I only reproduced repeated key in Xorg but using a VM… I'm not sure if the issue is clearly native though or just related to using the vm (and so, maybe qemu repeating the key…) [09:51] ah, good to know [09:51] I didn't see it mentioned by users/bugs reports yet [09:51] so doesn't seem to hit/annoy lot of users atm [09:51] yeah, maybe it's only VM related, and so, unrelated [09:51] yep [09:51] didrocks: repeated key a broken in mutter/gnome-shell even without vm [09:51] andyrock: Xorg? [09:51] andyrock, under xorg? [09:52] didrocks: xorg for sure [09:52] :/ [09:52] ah… so a real one [09:52] but when I tested on wayland is not better [09:52] yeah, we started seeing it on wayland here [09:52] I started to take a look before we started the work on gnome-initial-setup [09:52] was better in Xorg… but hemf :/ [09:53] didrocks: just keep f11 pressed while firefox is focused [09:53] save your session before ;) [09:53] * didrocks isn't going to do it then :p [09:54] in compiz we had a way to specify for a keybinding if autorepeat should work or not [09:54] with f11 it shouldn't [09:54] with alt+tab it should [09:55] there is not such a thing in mutter [10:04] andyrock, https://bugs.launchpad.net/ubuntu/+source/gnome-online-accounts/+bug/1768797 [10:04] andyrock, dont look at that now though :) [10:05] seb128, yes my cairo patch fixes the installer crash. I half expect upstream to not like it because of the performance implications. Although they already have a bug open for the same line of code, for similar reasons [10:05] duflu, nice, let's see what they say [10:06] we shouldn't hit that codepath in "normal use" right? [10:06] the xerror is hit only because ubiquity is doing weird uid handling? [10:07] seb128, it is a "normal" code path, but only "normal" for setup logic. Not central to animation performance [10:07] willcooke: I've a weird way to check if the credentials are still valid, it can be this the reason why it fails [10:07] I'll take a look later [10:24] seb128: ubiquity runs with uid=0 and euid=0 in 17.10 [10:24] when drop_privileges_save the uid and euid stay the same [10:24] what part/at what time? [10:24] seb128: in 17.10 from the start [10:24] you can notice just using ps [10:24] well that's right/supposed to be the same in 18.04 [10:25] nope in 18.04 is runs as 999 [10:25] at least the euid [10:25] let me check the last commits [10:25] well, ubiquity does those drop/restore dances [10:26] that don't work as they should :D [10:26] but the .desktop still do sudo ... [10:29] LUNCH! [10:31] andyrock, you are now assigned part of bug 1751252 :) [10:32] seb128, here's a fix: https://bugs.freedesktop.org/show_bug.cgi?id=98883#c6 I think I'm done for today [10:32] bugs.freedesktop.org bug 98883 in xlib backend "BadAccess errors in ShmAttach due to thread races with XNextRequest() usage in cairo-xlib-surface-shm.c" [Normal, New] - Assigned to chris [10:33] andyrock, if you have a fix for ubiquity too, please add it to bug 1751252. I'm heading off... [10:53] Trevinho: this has to be some refcounting thing [10:54] Laney: which one? The signal unsubscribe? [10:55] not sure where, these things are the worst [10:55] that's probably just the one that gets unlucky and makes the thing actually be disposed [10:55] side effect of the gjs leak fix? [10:55] dist-upgrading, and was almost done in the VM from 17.10 -> session closed, no crash file :/ [10:55] journal error? [10:56] nothing either [10:56] apart from the session closing part [10:56] I didn't get that starting in the exact same state… [10:56] :/ [10:56] (and it went further down, as I got the issue on saving the upgade telemetry file) [10:56] I guess that's another try now, reverting to the snapshot [11:21] seb128: not sure if that... But indeed it's the garbage colector. However I don't like how it works the glib side of things either, so I was looking into it [13:03] seb128: so the crash has been introduce by https://git.launchpad.net/ubiquity/commit/?h=bionic&id=dcb538ded2159439e3d9c4c988a6d9d2c4138622 [13:03] seb128: the commit is correct, but it's triggering the crash [13:04] ahah [13:04] Laaaaannnnneeeyyy [13:04] it's not Laney's fault [13:04] :D [13:05] what [13:05] redoing an upgrade test. It seems the issue is with screen locking… [13:05] at least, want to confirm telemetry first before even starting debugging this… [13:06] didrocks, what are you debugging/fixing? [13:06] Laney: so without that commit drop_privileges_* does not work and this does not trigger the crash [13:06] but having drop_privileges_* not working is not a good idea [13:06] the same one duflu fixed? [13:07] but at least we know why it started happening [13:07] yes [13:07] ok... [13:07] duflu is not sure his fix is not going to have a performances impact [13:07] so we are waiting to hear from upstream before considering it as the way to go [13:07] but yeah, those permissions issues in ubiquity are really annoying :/ [13:08] we need to make it an user app with proper backend/polkit use [13:08] seb128: I'm adding upgrading logs to our telemetry [13:08] not going to be for .1 though [13:08] yes that's well known [13:08] but unless some people get assigned to do this work it's not going to happen [13:08] right [13:09] yeah, we need to have an UI piloting a separate backend with clear communication and permissions between those 2 [13:09] however, I'm afraid that's almost a rewrite… [13:10] we raised up that installer needs more resources [13:10] ideally some extra headcount is needed for installer, n-m, etc [13:10] let's see if we get anything [13:10] yep [13:14] Laney, I've no idea what your backlog looks like atm, do you think you have space to look at bug #1768541? unsure how important that problem is, we got a few reports, would be nice to at least have someone trying to understand what's going on to see if that needs to be fixed or is low importance [13:14] bug 1768541 in gnome-keyring (Ubuntu) "package libp11-kit-gnome-keyring 3.20.1-1ubuntu1 failed to install/upgrade: libp11-kit-gnome-keyring:all 3.28.0.2-1ubuntu1 (Multi-Arch: no) is not co-installable with libp11-kit-gnome-keyring which has multiple installed instances" [Undecided,Confirmed] https://launchpad.net/bugs/1768541 [13:17] andyrock, so does it mean on 17.10 the webkit code is used as root? [13:18] seb128: that's what I was thinking [13:18] it should not work as root but it does [13:18] k, well at least it explains why it started recently [13:18] Can do later. [13:18] at least things make sense [13:19] Laney, thx! [13:19] andyrock: Try moving the widget creation to the constructor or something that is not running as root [13:19] andyrock, k, thanks for poking/figuring that out. Feel free to move back to the other things you were worked on, if cairo's upstream is happy with duflu's patch we should be set [13:19] Save it in the class and fetch it out when you need it later [13:19] Laney: thanks.. I'll try to do that [13:20] maybe try what Laney suggests [13:20] seb128: let me spend some more time [13:20] on this :D [13:20] kk [13:34] seb128: confirmed that the webkit process runs as root in 17.10 [13:34] the two processes [13:34] k [13:34] well that explains then [13:35] I'm not sure to understand why we hit the access Xerror when being "ubuntu" [13:35] I would have expeted to be the other way around [13:35] seb128: I guess it's the change of [e]uids [13:35] that confuses X [13:35] in 17.10 they don't change [13:36] in 18.04 they change and the change confuses X [13:36] hum, k [13:44] phew, first successfull dist-upgrade with do-release-upgrade done [13:44] same, it's a pain to test without publishing a version… [13:44] now, seems harder to test the gtk version [13:45] I think I'll as it needs to sudo with my download bypass switch to a Xorg session [13:45] (on 17.10) [14:08] didrocks, what sort of metrics do we add to the upgrader? [14:10] seb128: which version we upgrade From, and the various step + time they take [14:10]