[02:28] Trevinho, jbicha, nice work, thanks. You got out a lot more 3.28.1 than I even expected [02:28] duflu: :) [02:29] duflu: missing only your patch now, but there's a new iteration I pushed [02:29] Trevinho, performance improvements can wait. We'll still be adding to that list for months [02:30] Stability is more important [02:30] I've been knowingly working on 18.10 for a while now, not 18.04 :) [03:20] Trevinho, jbicha, this is odd: bug 1765261 [03:20] bug 1765261 in gdm3 (Ubuntu) "Ubuntu 18.04 login screen rejects a valid password on first attempt. Always works on the second attempt" [Undecided,New] https://launchpad.net/bugs/1765261 [03:47] RIP gksu [04:24] robert_ancell, did something replace it? Is that because of Wayland policy? [06:23] good morning [06:25] Morning didrocks [06:31] hey duflu [06:53] good morning desktopers [06:54] Hi seb128 [06:54] hey duflu, how are you? [06:54] seb128, going OK, you? [06:54] I'm good! [06:56] Good. And with that I will run away to check the mail and feed water the citrus [06:56] duflu, we managed to land the gjs changes as you probably saw :) [06:56] Yeah, thanks seb128 [06:56] thank you for getting those fixes ready for upload! [06:56] enjoy the watering :) === pstolowski|eod is now known as pstolowski [07:08] good morning desktoppers [07:09] salut oSoMoN, en forme ? [07:09] salut seb128, oui la forme, et toi? [07:11] oSoMoN, je pourrais dormir plus mais sinon ça va, il fait beau et chaud :) [07:12] yesterday was tennis night, was really nice to play again [07:13] I could play much since the beginning of the year, between travels, crappy weather and colds&co [07:13] couldn't* [07:15] yeah, sports is good, I should do more of it [07:18] Morning oSoMoN [07:19] hey seb128, oSoMoN [07:19] lut didrocks [07:19] hey duflu [07:19] salut didrocks [07:23] duflu: first results from the french users are really promising! [07:24] still some leaks ofc, but way lower memory usage [07:24] didrocks, seb128 upstream landed the same fix a couple of minutes ago [07:24] \o/ [07:24] duflu: it's hard to believe though, a lot of people says their RAM consumption even at boot are divided by 2 [07:24] too good to be real? just a placebo effect? [07:25] didrocks, yeah just log in and measure, and it's much better [07:25] nice! [07:25] good work again :) [07:25] duflu, https://gitlab.gnome.org/GNOME/gjs/commit/a6b6fc13 sounds like it could improve things as well, but that's for SRU at this point [07:26] seb128, yeah there are a couple more such fixes there [07:28] didrocks, when I booted my laptop this morning I was greeted by the initial setup GUI, which looks pretty good, and was mostly translated to French, so all good [07:29] oSoMoN: heh, the UI part is all Robert's :) [07:29] the last screen to choose to send the data or not, though, said something like "send data to Canonical occasionally", which made it sound like it's not a one-off, is that correct? [07:29] yeah, I think this is for popcon [07:29] ah, right, popcon [07:29] but I'm not even sure if this enables by g-i-s, I didn't follow that [07:30] I know it was discussed and robert said he will have a look [07:30] didrocks, oSoMoN, the "ocassionally" has been dropped in the version that is in the queue [07:30] ah ok :) [07:30] ok, problem solved then [07:30] replaced by "every 10 seconds", ofc? ;) [07:30] lol [07:30] ++ Yes, send system info to Canonical [07:30] right, the change is there [07:31] amazing that it's already mostly in French [07:31] we might add that back/enable popcon for .1 [07:31] yeah, we would need a g-c-c change then [07:31] to enable/disable it afterwards [07:31] right, that's why it was postponed to .1 [07:31] just enabling popcon was not difficult [07:31] yep [07:31] I guess same with whoopsie? [07:31] yes [07:32] you guys don't expect things to be calm again after next week :p [07:32] didrocks, I thought the French translations were your doing, now I'm disappointed :) [07:32] sounds like road to .1 is going to be busy [07:32] oSoMoN: oh, I can gladly take all the credits if that makes you feel better :) [07:32] oSoMoN, be nice to didrocks, you still need a +1 for your ubuntu-desktop application :) [07:32] yeah, I wonder if my past unity-2d experience is something worth mentionning [07:33] it's been a long time I didn't sponsor you [07:33] but I'm definitively +1 with your past work [07:33] you worked with him on some gdm issues previous cycle? [07:33] yeah, that's one data point, but correct, that's good enough entries to vote [07:33] I'll do that later today [07:33] cheers! [07:33] * didrocks fixes… hem… ubuntu-report autopkgtests :p [07:34] :) [07:34] (readding vendoring, which I explicitely removed from autopkgtests because of previous strategy and didn't think to delete the rm -rf …) [07:34] now that my autopkgtests vm is ready, let me first get the failure [07:34] then ensure it fixes it [07:34] * oSoMoN will change his nick to olirocks as a token of admiration [07:34] ahah ;) [07:35] I hope there is no expiration period :p [07:35] duflu, your gdm/password rejected on first try issue sounds weird, did it start with gnome-shell 3.28.1? Maybe a regression? [07:35] seb128, yeah it started today [07:36] duflu, do you get also the issue if you wait a bit/type slowly? there was that issue about key repeating themself under wayland when the system is busy [07:36] seb128, that might be a factor. If I took the time to log in to ssh first to debug the problem it suddenly didn't happen [07:37] Although restarting graphical.target I was able to reproduce it again [07:37] seb128: not only on wayland, I triggered the repeating key on X as well [07:37] I wonder if you could type the password, ctrl-C it and select "log with another user" and ctrl-V to see if it's correct [07:37] Is that even possible? [07:37] Or would you be pasting dots? [07:38] I wonder [07:38] otherwise I guess that needs to patch the gdm code to not show * :p [07:38] didrocks, weird, I though that was a wayland specific issue [07:39] seb128: I thought as well, but I got it 3 times in the past 2 weeks on my X session [07:39] especially when my machine is swapping [07:39] is that a new issue as well? [07:39] I don't know if it's new, or just happens so rarely that we didn't notice it [07:39] (on X) [07:40] but not a high prio, it's really happened 3 times to me [07:41] https://bugzilla.gnome.org/show_bug.cgi?id=777693#c10 is the issue I was thinking about [07:41] Gnome bug 777693 in general "Input event (e. g. typing on keyboard) is sent repeatedly during high load" [Normal,Needinfo] [07:41] bugzilla.gnome.org bug 777693 in general "Input event (e. g. typing on keyboard) is sent repeatedly during high load" [Normal, Needinfo] - Assigned to gnome-shell-maint [07:41] bug 777693 in PernillaSYS "Crash-bug when trying to view registrations if none of the users conferences is locked" [Medium,Fix released] https://launchpad.net/bugs/777693 [07:41] which is specific to wayland [07:42] but there might be other issues with similar effects [07:43] seb128: yeah, basically the exact same effect [07:43] you press for a short while, under load, keys is repeated [07:43] (and double checked that I was on X) [07:44] didrocks, it's not on the lock screen/greeter that you see the pb? since the greeter/gdm is always under wayland [07:44] seb128: no, really the user session [07:44] :/ [07:45] yep :/ [07:45] I know… [07:45] I wonder if there is some regression in 3.28 [07:45] we didn't get reports about such issues in 17.10 [07:45] well, we were under wayland [07:45] this issue existed on 17.10 [07:46] k [07:46] (but, I was only testing it with wayland) [07:46] probably worth reporting upstream then [07:46] yeah, the thing is that it's quite vague [07:46] "under load…" [07:46] the only reports I find there are wayland specific [07:46] but it's something that Fedora has [07:46] e.g https://bugzilla.gnome.org/show_bug.cgi?id=776639 [07:46] Gnome bug 776639 in keyboard "Key Presses Repeated, Lost or Out of Order Wayland" [Major,New] [07:46] bugzilla.gnome.org bug 776639 in keyboard "Key Presses Repeated, Lost or Out of Order Wayland" [Major, New] - Assigned to gnome-shell-maint [07:46] bug 776639 in postgresql-8.4 (Ubuntu) "package postgresql-8.4 8.4.8-0ubuntu0.10.04 failed to install/upgrade: subprocess installed post-installation script returned error exit status 1" [Undecided,Invalid] https://launchpad.net/bugs/776639 [07:46] ah, you meant, the Xorg one? [07:47] yes [07:47] yeah, I'll report it, unsure against what (and if it will be looked, as it's not wayland), but let's try [07:47] gnome-shell I guess? they can reassign if needed [07:47] thx [07:49] yeah [07:49] I would need to see if that happens in apps [07:49] or only in G-S UI [07:50] I don't remember TBH [07:50] seb128, could you re-paste the last version of your patch for orca? [07:50] jibel, that isn't working [07:51] jibel, but http://people.canonical.com/~seb128/ubiquity-screen-reader.patch (that one has extra print() statements for debugging) [07:51] seb128, I know it is not working, but I'd like to check something [07:51] thanks [07:51] yw [07:52] k, I'm moving back to my desktop, I had my morning coffee, trying to be back online before u.k joins :) [07:52] brb [07:52] too late sucker [07:53] Morning Frenchies [07:53] hey willcooke [07:53] and Aussie [07:53] s [07:53] how goes didrocks? [07:53] Thanks for mutter 3.28.1 :D [07:54] willcooke: it's sunny and not too warm yet! Good day for exercising abit [07:54] and you? [07:54] Morning willcooke [07:54] Sunny here too! [07:55] even there ;) [07:55] I hope it stays nice at the weekend and I can get out of the house and do some excercise [07:55] yes! even here on the island [07:55] People are a bit scared [07:55] https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/1757401 is fixed for me now! [07:55] Ubuntu bug 1757401 in mutter (Ubuntu) "hybrid GPU: one screen stays black in mutter 3.28" [High,Fix released] [07:55] especially for multiple continous days [07:55] dupondje: nice! [07:58] woot, thanks for the report dupondje [08:02] morning [08:02] hi Laney [08:02] It's summer again! [08:02] hey Mr Laney [08:02] (back) [08:02] you were a few seconds too late seb128, we beat you :) [08:02] DOH [08:02] 09:52:17 <-- | seb128 (~seb128@ubuntu/member/seb128) a quitté (Quit: Ex-Chat) [08:03] 09:52:40 willcooke | too late sucker [08:03] :) [08:03] so didn't run fast enough at *all* [08:03] FAIL [08:03] hey Laney! [08:04] half victory :p [08:04] hey willcooke didrocks seb128 [08:04] nope [08:04] well I can't see your join message [08:04] willcooke, the frenches will get you, don't worry! [08:04] but your first real message was after I said hi :P [08:04] :-( [08:04] * Laney was tying in sweet peas [08:04] AND [08:04] i'm wearing shorts [08:04] #summer [08:04] k, you win today [08:05] #summer feels good [08:05] * seb128 went to tennis yesterday, that was greeeaaaaatttt [08:05] * seb128 recommends it [08:05] too much time on the ground [08:07] kenvandine, ubuntu-desktop@ moderation queue got some build failure email from launchpad, in case you didn't see/receive it by some other way [08:07] https://launchpadlibrarian.net/366205059/buildlog_snap_ubuntu_bionic_amd64_gnome-3-28-1804_BUILDING.txt.gz [08:08] looks like a snapcraft issue though [08:08] " File "/usr/lib/python3/dist-packages/snapcraft/cli/env.py", line 17, in [08:08] from distutils import util [08:08] ImportError: cannot import name 'util'" [08:09] it's the remaining bug I didn't have time to file yet [08:09] got it with communitheme [08:10] I guess Ken fixed it already? [08:10] https://launchpad.net/ubuntu/+source/snapcraft/2.41+18.04.2 [08:11] oh nice! :) I never dare hijacking their upload/release process (but I think he didn't) [08:12] interesting files in the diff ;-) [08:12] I guess he didn't released it the way the snapcraft team is doing it :p [08:13] right [08:13] * didrocks notes Laney release team habits: clicking on the diff [08:14] looks like the fix worked, https://launchpad.net/~ubuntu-desktop/+snap/gnome-3-28-1804/ has a successful build [08:15] nice to see we finally got core 18 working/usable! [08:18] Morning willcooke, jibel, Laney [08:19] yeah! [08:19] * didrocks can't wait to see it moved, and have applications snaps connecting to the theme snap [08:20] hey duflu [08:45] jibel, did you figure out something new/more about the screen reader? how much do you think it's a blocker knowing that live session has screen reading working (assuming someone that needs the reader can easily boot to it, which might be not true if the isolinux menu isn't accessible and there is no screnereading in maybe-ubiquity) [08:49] seb128, nothing new yet. Without this, blind users cannot boot to the live session. [08:49] it doesn't boot after a timeout? [08:49] :/ [08:49] no [08:50] ubiquity-dm waits forever [08:50] I got sidetracked helped other work to land in the past days but I go back to screenreader in the afternoo [08:50] one heavy hammer would be to make ubiquity-dm just start orca [08:50] it's a bit more ram usage [08:50] but it's not going to speak unless the gsettings is on [08:51] then we just need to figure out why it's not outputting anything, but I think that might have to do with starting speech-dispatcher as well, I got it working the other day like that [08:51] I try that again with a fresh boot in a bit [08:51] my theory is that starting speech-dispatch and orca from ubiquity-dm is good [08:51] seb128, I am not even sure the a11y menus in syslinux are still working. There is lot of code that rely on a11y-profile-manager [08:52] as said it just creates a bit more memory usage, which is suboptimal but should be fine [08:52] I didn't even know we had menus for a11y in syslinux [08:52] I got it working by starting from a tty [08:53] just an export display + launch orca [08:53] what other features are missing/not working? [08:53] weird, I did that the other day and it would speak anything [08:53] I tried to -d and the log was empty [08:53] seb128, braille, osk, ... [08:53] but the most important is the screen reader IMHO [08:54] a11y-profile-manager was added in recent cycles [08:54] reverting those commits might not be too difficult [08:54] I don't remember why I disacarded that previous cycle when I looked in N-Y [08:54] I think that's because some of the old code was relying on indicators/unity anyway and we can't really go back to what it was doing [08:55] anyway I'm confident we can fix the screen reader case at least [08:55] me too [08:55] osk should be easy, that's handled by the shell itself in 3.28 [08:55] so the keybinding for that should just work? [09:02] wayland session doesn't work properly anymore with communitheme, unsure what GNOME update broke it (it worked when I released it) :/ [09:06] :/ [09:06] * didrocks tests some theory [09:06] it's even worst, first login doesn't work, second does [09:06] seb128, I confirm Orca is not started when selected from syslinux (or add access=v3 to the boot command line) [09:07] and XDG_DATA_DIRS are different between logins… [09:09] didrocks, :/ [09:10] what's the difference? [09:10] jibel, are those access= parameters documented somewhere? [09:11] seb128, I don't know any place where it's documented. Just press F5 in the boot menu [09:11] right [09:11] otherwise it's in casper [09:11] I'm going to review all those options while I'm at it I think [09:11] there is a casper-bottom script for a11y [09:12] thx for the pointer [09:12] seb128: ok, after some testing, it seems that in wayland, if you now set XDG_DATA_DIRS, it doesn't append the rest of the session anymore [09:12] which uses the profile manager [09:12] but [09:12] on the second login, you have everything added [09:12] really weird :/ [09:12] :/ [09:13] didrocks, bug #1764355 seems a bit similar [09:13] bug 1764355 in gnome-session (Ubuntu) "XDG_CONFIG_DIRS keeps getting expanded every login" [Undecided,Confirmed] https://launchpad.net/bugs/1764355 [09:13] seb128: yeah, I have repetitions as well… [09:13] I saw it reported recently but didn't have time to investigate [09:13] sounds like maybe something to investigate for release [09:13] need to bisect gnome-session maybe [09:13] you are debugging it? [09:13] yeah, I think I own it now :p [09:13] thx! [09:14] let's see first if I can find what/when it was working [09:14] * didrocks starts with gnome-session [09:16] gnome-session has a thing in /etc/profile.d to set up the XDG dirs [09:16] blame it on doookooo :) [09:16] ah? [09:16] I don't think our side of things changed much [09:17] but I'm just guessing, not based on fact [09:17] :p [09:17] yeah, let's starts with gnome-session [09:17] and see [09:17] k, I don't think it makes sense to start debugging those installer/a11y issues to stop in the middle for lunch so I'm going to step out for errands/early lunch now and then sit down looking at thise issues once I'm back and have a solid debugging afternoon [09:17] good luck didrocks [09:18] seb128: are you thinking of something else as you mention foundation? [09:18] didrocks, no, just profile.d is a lower level afaik [09:18] it's coming from -common [09:19] at least, the xdg- one [09:19] I don't mean the script, but profile.d handling by the system [09:19] but no, I don't have real reasons to believe it's not our issue at this point [09:19] I was half trolling [09:19] yeah, let's see :) [09:19] could be systemd/bash/whatever handles those though [09:19] (only half- ;)) [09:20] good luck with the debugging :) [09:20] (yeah ;) [09:20] thx! [09:20] if only we had days of isos… [09:24] kenvandine, seb128 hey, you guys asked for a stable core18, we did some work on this in hte last few days and we are pushing new versions into edge, if you could test with your snaps if this is all working we can promote to beta/candidate/stable [09:26] nope, reverting gnome-session isn't it… [09:27] ah, better, now I constantly have login not working [09:34] ok, so confirming that first time in the session script, XDG_DATA_DIRS is empty [09:34] second time, it has /usr/share… [09:34] (before starting gnome-session) [09:34] mvo, thanks! Ken mentioned yesterday that he had snaps built on it [09:34] (on wayland only) [09:35] so I guess the /etc/X11/Xsession.d hides that bug forus [09:35] for us* [09:35] willcooke: do you know which snaps those are? I am writing a mail now too so that this does not get lost in irc but if you know I can test the new core18 with those right away [09:36] ah, 3rd login: XDG_DATA_DIRS empty again [09:36] mvo: [09:36] i have a build of gnome-calculator that uses gnome-3-28-1804 and core18 [09:36] all working [09:36] that was yesterday at 1515 my time [09:36] seb128: Laney: do you know if we have anything lazying tearing down our session and maybe gdm picks it up? [09:36] willcooke: ok, I try that one [09:37] willcooke: thank you [09:37] mvo, looks like it's in edge [09:37] and beta [09:37] the edge one will be the CI build from last night [09:38] willcooke: ta [09:44] ha, I was replying to a forum post about clicking and just as I was about to hit reply duflu beat me to it :) [09:44] Heh. [09:44] * duflu closes mail for the day [09:44] thx duflu [09:45] duflu, what's your reading on the feedback so far? Seems that it's not worse, but there are still a few leaks, which we expected [09:45] willcooke, yeah it's exactly as expected. Measurably better but we still need to get the other fixes as they land [09:46] hm, having said that, just rebooted my 18.04 box and it wont log in. Prolly unrealted, will debug [09:47] and I cant get to a vty either. grrr [09:48] willcooke, bug 1765261 ? [09:48] bug 1765261 in gdm3 (Ubuntu) "Ubuntu 18.04 login screen rejects a valid password on first attempt. Always works on the second attempt" [Undecided,New] https://launchpad.net/bugs/1765261 [09:49] duflu, hm, I think this is a different one. It's just locked up now [09:49] willcooke, hit Cancel? [09:50] I cant, mouse wont move [09:50] maybe I can ssh in [09:50] seb128: Laney: so, I think what happens in the "working" case is that the previous session isn't closed and envs are picked up from it [09:51] seb128: Laney: I did multiple tests, but noticed that if I logout/login right away, sometimes I am in the working state (old vars picked up), sometimes not. [09:52] with more testing, I noticed that if I wait for more than a minute, logged in, logout and log in again, I'm in a clean session (no older env vars picked up) [09:52] so, my bet is that the lazy loading of things, like gnome-software is the impact for delaying session close [09:53] and so, repeated env vars (picked up by old session) is due to that session not being closed yet (because you just logged off and in again after a short session) [09:53] so, this is one bug [09:53] duflu, killed gnome-shell & gdm and the mouse started responding again [09:53] and it logged in fine this time [09:53] the second one, which I'm looking, is why now the env vars don't complete anymore under wayland in the "no previous session case" [09:54] duflu, meh, probably just one of those things [09:54] maybe actually that never worked before in the wayland case and I always had a session hanging due to the first bug ^ [09:54] (which seems the most likely) [09:59] didrocks: you should be able to close a session and see what processes are still running [09:59] e.g. loginctl session-status $session [09:59] Laney: well, you need to be logged in for this [09:59] so, need to create another user :p [09:59] sure [10:00] I'll first take a lunch break, but I think this is the case [10:00] and regardless of that, XDG_CONFIG_DIRS shouldn't happen itself if already there, I'll modify this [10:01] enjoy your baguette didrocks ;P [10:01] willcooke: and *fromage* [10:01] lololol [10:05] wish my monitor didn't stop sending audio when it DPMS powers off :( [10:15] we need at least fixing /etc/profile.d/apps-bin-path.sh and /etc/profile.d/xdg_dirs_desktop_session.sh [10:15] * didrocks really goes now === pstolowski is now known as pstolowski|lunch === pstolowski|lunch is now known as pstolowski [11:42] mvo: I tested it with gnome-calculator and the new platform snap [11:43] morning kenvandine[m][m] [11:43] gnome-3-28-1804 is in edge but gnome-calculator is just a local build [11:44] willcooke: good morning [11:51] hey kenvandinemm [11:51] those [] are annoying to type :p [11:54] Laney, thanks for upstreaming that udisks fix, I had that on my list for today ... sorry for being too slow, but I'm not going to complain about one less thing to do, you win a free beer :) [11:55] seb128: it's OK, I just wanted to get that uploaded and liked to have that task squared off before then [11:56] also you sounded like you wanted to be in orca mode :-) [11:56] I do indeed, that's higher priority for the release [11:56] so thanks :) [11:56] the main reason for uploading is the volume-key thing [11:58] right, I guessed, thanks for sorting that one as well! [11:59] didrocks: Hello, I work for Entroware, a computer manufacturer that ships Ubuntu machines and I was wondering if you would accept a change to the OEM collection section of ubuntu-report to /sys/class/dmi/id/sys_vendor from the chassis_vendor? It seems like that would be more appropriate and this is how we set ours. I would appreciate any feedback you can give on this. [11:59] seems like our release issues list is getting in shape [12:03] didrocks: Hello, I work for Entroware, a computer manufacturer that ships Ubuntu machines and I was wondering if you would accept a change to the OEM collection section of ubuntu-report to /sys/class/dmi/id/sys_vendor from the chassis_vendor. It seems like that would be more appropriate and this is how we set ours. I would appreciate any feedback you can give on this. [12:04] mwilson-e, hey, we saw what you posted, didrocks is currently away but he's probably going to reply to you once he's back which is probably in less than an hour [12:05] just stay here for a bit if you can [12:05] otherwise I suggest to retry later or email [12:06] seb128: Ah, my irc client messed up before so I had to reconnect and did not know if it sent. Cheers for letting me know. [12:06] np [12:21] kenvandine[m][m]: you tested and things are ok? [12:24] mvo: yesterday I did and it seemed ok [12:24] mvo: I can test the latest in a few minutes [12:26] kenvandine[m][m]: yeah, please check hte latest version in edge of core18 [12:26] kenvandine[m][m]: but no rush [12:28] mwilson-e: hey! I referenced on https://wiki.debian.org/HowToIdentifyADevice/System, section "/sys" which list chassis_vendor, so I think more standard than sys_vendor [12:29] also, a lot of hw seems to only have chassis_* populated and not sys_* [12:29] for instance: https://natalian.org/2015/02/18/Archlinux_on_a_Lenovo_X1C3/ [12:29] this is why I preferred that one over the other one [12:30] I wonder if we shouldn't do a try/catch (like if chassis_vendor is empty, try sys_vendor) [12:31] (note that most stack exchange answers list chassis_vendor) [12:36] didrocks: Would it be possible to do a try catch if sys_vendor is empty? Our chassis_vendor value is set to "Notebook" which is a default from our partners and used by multiple OEMs so this may be confusing. [12:37] We set the OEM values for System Information in the type 1 section which is referenced here in the SMBIOS spec [12:37] https://www.dmtf.org/sites/default/files/standards/documents/DSP0134_2.7.1.pdf [12:37]