=== cpaelzer_ is now known as cpaelzer [06:31] morning... [06:32] Laney: I've been using latest version, but I agree it wasn't the case here to look at. But I tried that branch in kms and simply it will stop ubuntu running forever in kms, as CanGraphical never goes to 'yes'. [06:33] so indeed I think is a lower level thing, just speaking of the effect here [06:59] Morning Trevinho [07:02] Wait. Trevinho is first online. [07:02] * duflu suspects he hasn't gone to bed yet [07:02] xD [07:03] ahaha, I did, but I had to drive a friend to the airport, so I had an early alarm :( [07:18] good morning [07:18] hi didrocks [07:18] hey Trevinho [07:29] Hello didrocks [07:32] morning duflu [07:36] k, back! [07:36] hey didrocks Trevinho duflu [07:36] good morning desktopers [07:37] Hi seb128 [07:38] how is everyone today? [07:42] good morning desktoppers [07:44] salut seb128, hey oSoMoN [07:44] LP: #1798005 [07:44] Launchpad bug 1798005 in ubiquity (Ubuntu) "Installer crashed on the "Who Are You?" panel." [Undecided,New] https://launchpad.net/bugs/1798005 [07:44] can anyone check ubuntu main for that bug? [07:44] I have to walk out the door in a few mins [07:45] salut didrocks, seb128 [07:45] log: https://paste.ubuntu.com/p/kbzDFYpCBT/ [07:46] lut oSoMoN; en forme ? [07:46] seb128: on the "how is everyone" -> doing another yack shaving it seems :p [07:46] and you? [07:47] seb128, ça va, et toi? [07:50] Morning oSoMoN [07:50] oSoMoN, ça va bien! [07:50] didrocks, poor yack, winter is coming they need to keep warm :p the installer/screen reader one? [07:50] * duflu was briefly distracted by spider killing... Happy spring [07:51] seb128: yeah, the whole a11y stack basically [07:51] :( [07:51] do you have a clue what's the issue is? [07:51] yep [07:51] how do you ubiquity carry over the config to the installed system? [07:51] it's using… a11y-profile-manager [07:52] lol [07:52] which isn't seeded anymore [07:52] that explains :) [07:52] same, isolinux is using a11y-profile-manager [07:52] shrug, I did revert some of that when I fixed other a11y problems in previous cycle [07:52] which has a lot of vulneravities [07:52] I should have done a full revert to the codebase before that component was added :/ [07:52] like well-known names owned by root in /tmp [07:52] but a11y-profile-manager isn't used/on the iso anymore, is it? [07:53] also, this thingy doesn't really know about GNOME [07:53] it's not on the iso [07:53] but isolinux set the profile you select based on it [07:53] ah :/ [07:53] shrug, we should have reverted those changes [07:53] and ubiquity, at install time, runs a hook which triggers it [07:53] that component was added not so long ago [07:53] I like as well the usage of: [07:53] function_witout_parameter() { [07:54] … read global variables … [07:54] } [07:54] …set global variables… with "if profile" [07:54] function_witout_parameter() [07:54] the whole code is like that :/ [08:05] morning [08:06] hey willcooke [08:06] hey willcooke [08:07] at the office again today? [08:13] Morning willcooke [08:15] yo [08:16] hey Laney, how are you? enjoy the great London? [08:16] morning Laney [08:17] and the citizenM? [08:17] seb128, not yet, but might go in a little later on. Got some meetings this mirning [08:17] willcooke, you didn't send the traditional "who is where this week" and "meeting reminder" btw :p [08:17] darn [08:18] I'll do it no [08:18] now [08:20] seb128: it's good! nice cappuccino this morning :> [08:20] & hey didrocks willcooke [08:20] I forgot to do a status update [08:24] bad Laney :p [08:33] probably do it after we fix the installer that we broke :-) [08:34] haha [08:34] release traditions [08:34] with Didier we decided to stay away from doing late changes to fix a11y this cycle [08:35] the "onscreen reader is not enabled after installation if selected during install" is there since 17.10 :/ (casper still uses a11y-profile-manager than we killed from the iso/main when we switched to GNOME) [08:35] hi guyz [08:35] whoops! [08:35] good that it's not new though [08:36] yeah, and we know what happens when we fix a11y in the installer, right? [08:36] we can create a new tradition, but I'm not in :p [08:38] you could try elif finished_step == 'ubiquity.components.partman_commit': [08:38] WHAT [08:38] https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/1797861 [08:38] that [08:38] Ubuntu bug 1797861 in ubiquity (Ubuntu) "Screen reader doesn't read installer's GUI during a live session installation" [Undecided,New] [08:38] guessing that's not covered by previous statements about a11y-thingy [08:39] No, it's not related on that one [08:39] (we are only talking about getting options from isolinux and setting them on the final installed system) [08:41] right, I'm going to give a try to that one [08:43] thanks! [08:48] Laney: stupid question, but how do you trigger maybe-ubiquity using Boxes/kvm? [08:48] Laney: I always have isolinux popping up [08:49] and so, no way to have maybe-ubiquity [08:51] you can add it to the kernel cmdline [08:51] do you have the exact syntax handy? [08:51] otherwise, I can look [08:52] just add maybe-ubiquity somewhere in there [08:52] I guess replacing only-ubiquity by maybe-ubiquity [08:52] k! thanks Laney [08:52] trying to reproduce the issue as well (can't when booting directly to a live session or install session) [08:53] confirmed with maybe-ubiquity -> try ubuntu -> enable a11y -> start ubiquity [08:54] indeed, no root apps are read by the screen reader under that config [08:54] so only when going through maybe ubiquity? [08:54] or "luck" on that boot? [08:54] 100% on one try for both right now ;) [08:54] doing more run to confirm or deny [08:57] ah, have reproduced in only-ubiquity [08:57] I wonder if orca gets confused by having the same apps register 2 times or something [08:57] k, guess not then [08:59] hum [08:59] it's weird [09:00] in that session, it "sometimes work" [09:00] like selecting the language list, I get one on fourth selection read [09:00] (even if I go very slowly) [09:01] I'm on the maybe-ubiquity screen and I've activated it but it fails to read anything e.g when I change language [09:01] try keyboard [09:02] oh, yes, I was just doing that [09:02] (I didn't try activating a11y on maybe-ubiquity itself) [09:02] ah ok [09:02] because on the fuzzy results before, I got 100% working with keyboard, but not mouse… [09:02] so yeah, I picked "try ubuntu" [09:02] right, same here [09:02] something is fishy [09:02] I was trying the mouse [09:03] well, keyboard is what makes sense [09:03] indeed [09:03] blind users can probably not use the mouse [09:03] but still, don't really like that :) [09:03] yeah [09:03] it's going to hit us back because we don't understand it fully [09:04] so the reader is working on the live session for me on first try going through maybe-ubiquity [09:04] doesn't for me on this second try [09:05] maybe-ubiquity -> enabled a11y -> live session -> (a11y enabled already) ubiquity -> nothing (nor mouse or keyboard) [09:05] k, try 2 [09:06] my first one was [09:06] "wait to boot to maybe-ubiquity -> alt-super-S, it reads something -> click "try" -> start ubiquity from the launcher -> it reads the ubiquity dialog buttons ("continuer") [09:07] I started it from the desktop icon [09:07] but apart from that, same… [09:07] you selected french on maybe-ubiquity? [09:07] yes [09:07] let's try [09:08] waow [09:08] it reads the list in maybe-ubiquity [09:08] but with an initial lag of at least 5s [09:09] but then, nothing in live session… === cpaelzer_ is now known as cpaelzer [09:10] we should call release week "look at a11y installer issue week" [09:12] second try, working as well in live [09:12] "continuer button" is spoken when I start ubiquity once in live [09:12] could be racy, but how come… [09:12] and if you sudo gedit? [09:12] does it work for you if you stop/start the screenreader? [09:12] any error in the journal? [09:13] sudo gedit speaks "document sans titre...3 [09:13] nothing failing looking for a11y in the journal [09:13] I tried to stop/start it again [09:13] not better [09:13] you have sound if you use aplay or something? [09:13] or the test audio from g-c-c? [09:13] oh a11y is working [09:13] for everything which !root [09:13] :( [09:14] orca has a stacktrace [09:14] on shutdown [09:14] but nothing when starting… [09:14] yeah, all windows which aren't root are working [09:15] let me confirm again (I'm sure I've done that) that booting a live session right away works [09:16] yeah [09:16] apps as root works in that case [09:16] nice sudo apt install pastebinit -> plymouth screen -> exit [09:17] it switched vt! [09:17] but the session is still running [09:17] http://paste.ubuntu.com/p/g97MHJdCYr/ <- working env [09:17] let's reboot in maybe-ubiquity + session switch and see [09:18] wfm on another try :/ [09:19] virtualbox, not kvm? [09:19] sooooooo (Laney), for what I think is also willcooke's issue I think that more than gdm itself, is the way gdm is started. So right now we start it once we've a tty. and for the daemon is probably fine. Problem is that that point there might be no drm card (yet). [09:19] A way to fix this is changing gdm3.service so that it wants a drm card / fb to be there [09:20] seb128: ah, even better, in that session, no a11y working at all… [09:20] :( [09:20] didrocks, right, virtualbox here [09:21] http://paste.ubuntu.com/p/X46nG3RPFj/ -> failing session (after maybe-ubiquity, enable a11y, try ubuntu) [09:21] but not sure if this is a general rule... As in pure fb scenarios I guess X would start without any FB device in... So maybe gdm should wait for a drm device if trying with wayland, while on X could go without.. .But again even X should wait if there's a fb but there's no X driver for it... So quite a mess :) [09:22] nothing fancy diffing the logs, apart from IDs changing… [09:22] :/ [09:22] ah! [09:22] oct. 16 09:22:28 ubuntu orca-autostart.desktop[8065]: Une autre instance du lecteur d’écran est déjà lancée pour cette session. [09:22] oct. 16 09:22:28 ubuntu orca-autostart.desktop[8065]: Lancez « orca --replace » pour remplacer ce processus par un nouveau. [09:22] in that session [09:22] ah! [09:22] if I disable/reenable screen reader [09:23] maybe there is a race + something else with sudo? [09:23] another bug I guess [09:23] the desactivate not working [09:23] yeah [09:23] orca is 4 minutes old [09:23] under ubuntu user [09:23] no log on shutdown request though :/ [09:23] works everytime here/on virtualbox [09:24] willcooke: your laptop is still with Andrea? [09:24] I need to step out for early lunch but I try with kvm once I'm back [09:24] bbl [09:24] seb128: enjoy! [09:24] thx [09:28] Trevinho, hey, yes [09:28] unless he sold it [09:28] ahaha [09:28] you know... you trusted an Italian.. [09:29] :) [09:29] With the money he got, he could have bought a milkshake [09:29] Trevinho, reading backscroll [09:32] seb128: second time I'm reproducing the "no a11y working at all", I have to kill orca, then I'm in the "orca works with all but root apps" [09:32] in general I think for making things working in any normal scenario, I think there are multiple options... 1) udev rules + systemd service updates for gdm waiting drm cards (what I've done locally, seems to work properly) | 2) making gdm daemon to monitor udev so that if a main drm card is added it tries to create again a display. As right now it fires once, then if it fails, it just doesn't try again. [09:32] Trevinho, yeah, so the problem with waiting for a drm device is, as you said, there might not be one in headless mode [09:33] Trevinho, from ISO testing some of the flavours I think that some of them do retry the greeter later on [09:33] If we could make the systemd service retry, say, three times before giving up, would that do it? [09:34] willcooke: the daemon itself starts properly. it's more a gnome-session thing of not retrying again I think. [09:34] ah, kk [09:34] so let's just put the sleep statement back in again then [09:34] willcooke: as gdm starts the daemon, then it launches various subdaemons [09:35] got ya [09:35] well it's better to use udev anyway... [09:35] like wait for a drm for time, if it doesn't work, then just try. [09:35] I was joking about the sleep statement btw :) Just in case... [09:35] this is the faster and easier fix [09:35] sounds good [09:36] well, like having a "max timeout" for headless versions where there might be weird setups with no dri cards, but... well for 99% of ubuntu targets things should actually work [09:36] we should get jibel to confirm that it will work for his automated testing. He'll be in London in a few hours [10:13] sooo... eventually I think our main issue is this one: loginctl starts as soon as we've a fb... but having a fb doesn't always mean we've a gfx (dri) card ready to go. So it starts and then fails... CanGraphical doesn't seem to reflect this properly even (as because CanGraphical if is connected to `master-of-seat`, that one is also not safe to use since might be there when fb is set, not when also dri card is there). So for a proper fix [10:13] we've multiple possibilities, for a fix for release I'd go with udev + systemd checks. [10:33] Trevinho, +1 [10:33] * willcooke -> train [10:51] Just did an install of cosmic daily 20181016 in virtualbox with 3d acceleration on and ubiquity died bug 1798065 [10:51] bug 1798065 in ubiquity (Ubuntu) "Installer crashed :(" [Undecided,New] https://launchpad.net/bugs/1798065 [10:52] I suspect this may be a dupe of bug 1798050 [10:52] bug 1798050 in ubiquity (Ubuntu) "Installation crashed after Continue clicked after presentation of the time Zone" [Undecided,New] https://launchpad.net/bugs/1798050 [10:52] I switched off 3d acceleration in virtualbox and the install seems to be progressing further. [11:11] popey, can you tell me the specs of the VM? Especially RAM and VRAM [11:11] Doing a quick test from the train of some old ISOs I have [11:11] see if I can easily find when it broke [11:11] willcooke: 4GB RAM, 128MB VRAM [11:11] thx [11:12] and did you do a "try" or go direct to install? [11:12] just install [11:12] ack [11:12] i just left it till i got the gui with two buttons [11:14] popey, and was it on a cosmic host? [11:15] i.e. cosmic on cosmic [11:15] yes, looks like it was [11:15] RTFB [11:15] no [11:15] oh [11:15] cosmic vm on bionic host [11:16] ah, ok, that data was from inside the VM [11:16] yeah, i filed the bug in the vm using the popup from ubiquity [11:16] popey, was it a one time thing? or can you reproduce with the same setup? [11:17] also ubiquity issues at this point of the cycle might be worth raising in #ubuntu-release [11:17] i will attempt to reproduce it now [11:17] thx [11:18] my ISO from Friday worked here [11:18] Oct 16 10:38:06 ubuntu ubiquity: AssertionError: Failed to mount the target: /target [11:19] it's similar to bug #1798050 reported earlier [11:19] bug 1798050 in ubiquity (Ubuntu) "Installation crashed after Continue clicked after presentation of the time Zone" [Undecided,New] https://launchpad.net/bugs/1798050 [11:19] I'll retry using the existing diks [11:19] or bug #1798005 [11:19] bug 1798005 in ubiquity (Ubuntu) "Installer crashed on the "Who Are You?" panel." [Critical,Triaged] https://launchpad.net/bugs/1798005 [11:19] ah [11:19] ^ [11:19] yeah [11:19] willcooke, popey, ^ that's the regression L_aney was mention earlier this morning [11:19] so "known issue/being worked on" it seems [11:20] woot [11:20] thanks seb128 popey [11:20] I'll be in the office in 30 mins, so will report back to the channel [11:20] didnt do it on this run [11:20] thanks seb128 [11:21] popey, it's racy so pretty sure it's that issue, looks like L_aney as a candidate fix in https://git.launchpad.net/ubiquity/commit/?id=d784958 [11:21] \o/ [11:21] great success [11:25] yeah feel free to try it if you want [11:25] you can patch that from the live session [12:07] so... I'm going with the conditional-wait workaround for gdm now, I think is saner (wait drm a bit, otherwise continue how it is). For upstream i've various other ideas. Might be both systemd that would need to do a similar thing or gdm instead to be more robust. As Systemd right now advertize a CanGraphical as soon a there's a /dev/fb0 around, but this is not true if the drm device takes some time to initialize after the fb. And so when [12:07] the drmdevice has been finally added it can't advertize again a CanGraphical change. That's why the fix propsed upstream might help but doesn't fix the problem in all the cases. [12:07] probably lunch first though :P [12:21] hi [12:21] hi dgadomski [12:21] Trevinho: btw did you see my comment in LP: #1796822 that I get the corrupted screen when I enable autologin [12:21] could anybody please take a look at the merge request for bug #1755490? [12:21] Launchpad bug 1796822 in xorg-server (Ubuntu Cosmic) "Desktop live cd boots corrupted screen in Virtualbox on Bionic" [Critical,Confirmed] https://launchpad.net/bugs/1796822 [12:21] bug 1755490 in unity-settings-daemon (Ubuntu) "Incorrect information about display shown in unity-control-center" [Undecided,In progress] https://launchpad.net/bugs/1755490 [12:22] jbicha: mh didn't see no, but I think it all might be related to the same issue [12:22] dgadomski, is that especially urgent? [12:23] yes, I'm commenting because I think there were some indications we fixed the problem but that one still is easily reproducible for me [12:23] jbicha: run with this systemd unity (add a .service and enable it): [12:23] https://www.irccloud.com/pastebin/6gNvcnai/ [12:23] attach the log after that please [12:23] willcooke: nope, not super urgent, it's just there for some time [12:24] dgadomski, oki, I propose that we look at it after release then, if that's good for you [12:24] willcooke: yep, completely, thanks [12:24] cool [12:24] I'll add a trello card [12:26] I mean all these bugs are speed dependent, and because gdm starts too early basically :) [12:29] Trevinho: log attached to the bug [12:39] thanks [12:40] jbicha: was udevadm service running? [12:40] jbicha: did you systemctl enable it? [12:45] Trevinho: ok, new attachment, thanks! [12:59] ta [13:00] seems to be the same issue [13:29] hmm. remind me what URL the meeting updates lives on again? :-) [13:29] never mind. found it :-) [13:30] sorry folks [13:30] bear with me [13:30] unprepared [13:31] #startmeeting Desktop Team Weekly Meeting 16 October 2018 [13:31] Meeting started Tue Oct 16 13:31:07 2018 UTC. The chair is willcooke. Information about MeetBot at http://wiki.ubuntu.com/meetingology. [13:31] Available commands: action commands idea info link nick === meetingology changed the topic of #ubuntu-desktop to: Home of the Desktop Team, https://wiki.ubuntu.com/DesktopTeam | For help or questions, try #ubuntu | Work (read-only for non-developers): https://trello.com/b/3VYBPFaR/ubuntu-desktop-1810-cycle | Amaterasu watches over you benevolently | Desktop Team Weekly Meeting 16 October 2018 | Current topic: [13:31] Roll call: andyrock, dgadomski, didrocks, duflu (out), jbicha, jamesh (out), jibel/heber, kenvandine, laney, oSoMoN, seb128, tkamppeter, trevinho, robert_ancell (out) [13:31] hey! [13:31] o/ [13:31] o/ [13:31] hey [13:31] hi o/ [13:31] hey [13:31] hi there [13:31] Laney, andyrock and jibel are here in London [13:32] Everyone is very busy with release work, so let's keep this short [13:32] We haven't had chance to review the rls bugs, so we will have to do it live [13:33] Tracking first: http://reqorts.qa.ubuntu.com/reports/rls-mgr/rls-cc-tracking-bug-tasks.html [13:33] http://launchpad.net/bugs/1794280 [13:33] Ubuntu bug 1794280 in xorg-server (Ubuntu Cosmic) "gdm doesn't start on a fresh installation of Cosmic Desktop" [Critical,Confirmed] [13:33] We're working on that one, happy its in hand. Could have a fix/work around soon [13:33] http://launchpad.net/bugs/1796822 [13:33] Ubuntu bug 1796822 in xorg-server (Ubuntu Cosmic) "Desktop live cd boots corrupted screen in Virtualbox on Bionic" [Critical,Confirmed] [13:34] Initial indications are that this is the same bug [13:34] Should know for sure soon [13:34] http://launchpad.net/bugs/1794951 [13:34] Ubuntu bug 1794951 in gnome-calculator (Ubuntu Cosmic) "Calculator (snap) is slow to start on a freshly installed Cosmic machine" [High,Confirmed] [13:34] Ken will look at that tomorrow. Can be fixed with out affecting release [13:34] http://launchpad.net/bugs/1797543 [13:34] Ubuntu bug 1797543 in gnome-software (Ubuntu Cosmic) "Unable to update software - Button Restart & Update, restarts no updates applied" [High,Fix released] [13:34] fixed [13:34] http://launchpad.net/bugs/1796056 [13:34] Ubuntu bug 1796056 in xorg-server (Ubuntu Cosmic) "Xorg crashed with SIGSEGV in __strstr_sse2_unaligned() from glamor_egl_init() from try_enable_glamor()" [Medium,Confirmed] [13:35] I think we've worked around that mostly. tjaalton anything more needed there ^ ? [13:36] EOL [13:36] willcooke: nope [13:36] there is an upstream patch that has been merged to error out "properly", unsure if that's good enough though if the erroring out still means screwed systems? [13:37] upstream released 1.20.2 with that, too late for cosmic anyway [13:37] tjaalton, will it get an SRU? [13:37] willcooke: maybe [13:37] FWIW I havent seen that specific bug in recent testing, so I think all the other fixes we've done are masking it [13:38] yeah, probably not worth targetting at this point [13:38] ok, lets untarget [13:38] tjaalton ok with you? [13:38] sure [13:38] oki, I will do that in a mo [13:38] Incoming: [13:38] http://reqorts.qa.ubuntu.com/reports/rls-mgr/rls-cc-incoming-bug-tasks.html [13:39] http://launchpad.net/bugs/1797543 [13:39] fixed [13:39] Ubuntu bug 1797543 in gnome-software (Ubuntu Cosmic) "Unable to update software - Button Restart & Update, restarts no updates applied" [High,Fix released] [13:39] http://launchpad.net/bugs/1796548 [13:39] Ubuntu bug 1796548 in gnome-shell (Ubuntu) "[regression] New windows are all centred" [Medium,New] [13:39] Feature not a bug [13:39] +1s to reject? [13:39] yep [13:39] yep, changed on purpose [13:40] https://bugs.launchpad.net/ubuntu/+source/upower/+bug/1796550 [13:40] Ubuntu bug 1796550 in upower (Ubuntu) "[regression] Keyboard brightness control keys (down/up) don't work in cosmic (upower 0.99.8)" [Medium,Triaged] [13:41] not a release blocker IMO. There is a simple fix, but in the spirit of this meeting, notfixing? [13:41] that sounds like a good SRU [13:41] right [13:41] +1 [13:41] I'm going to add it to my SRU list [13:41] (and don't impact everyone btw, just tried) [13:42] https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1791574 [13:42] Ubuntu bug 1791574 in gnome-shell (Ubuntu) "gnome-shell crashed with SIGABRT: assertion failed "window->display->focus_window != window" in meta_window_unmanage" [Medium,In progress] [13:42] sru too [13:42] ack [13:42] seb128: list that too plz [13:42] ? [13:42] https://bugs.launchpad.net/ubuntu/+source/gnome-shell/+bug/1791574 [13:42] (on the SRU list, I guess) [13:42] [not if it's "yours"] [13:43] not a release blocker for 1791574 [13:43] ah, SRU seems fine [13:43] if that's the question [13:43] dunno what you mean by "list that too" [13:44] https://bugs.launchpad.net/ubuntu/+source/gstreamer-vaapi/+bug/1770725 [13:44] Ubuntu bug 1770725 in gstreamer-vaapi (Ubuntu) "Totem window is all black in Wayland sessions on Sandy Bridge CPUs" [Medium,Confirmed] [13:44] Since wayland is not the default, I say -1 for rls [13:44] rls-notfix since it's waylabd [13:44]