[05:14] morning [06:45] zyga, mborzecki: good morning [06:45] mvo: hey [06:45] PR snapd#8600 closed: tests: run smoke test with different bases [06:45] mvo: pinged you in some prs [06:46] mborzecki: yeah, saw it in the first, will look for the others now :) [06:46] ok [06:46] maybe we should just make those pulseaudio tests manual [06:47] mvo: this one too https://github.com/snapcore/snapd/pull/8562 [06:47] PR #8562: store: implement DownloadAssertions [06:51] PR snapd#8562 closed: store: implement DownloadAssertions [07:02] morning [07:02] Good morning [07:04] pstolowski: hey, good morning [07:04] zyga: pstolowski: hey [07:11] PR snapd#8557 closed: c/snap-bootstrap: check mount states via initramfsMountStates [07:34] zyga: did you get a link to activate your suse account? [07:37] PR snapd#8436 closed: configcore,tests: use daemon-reexec to apply watchdog config [07:43] mborzecki: checking [07:44] hmm [07:44] no? [07:44] I got an email earlier [07:44] about the new account system [07:47] mborzecki: I _hope_ that with the current changes the pulseaudio tests will no longer fail this way [07:47] now that both tests kill PA reliably via systemd [07:50] remarkable little failures [07:50] >0 but usually just one [07:50] and a pretty short quuee [07:54] hmmm [07:54] upgrade/basic [07:54] - Download snap "core" (9066) from channel "stable" (read tcp 10.240.0.214:45966->91.189.88.179:443: read: connection reset by peer) [07:54] eh, just network woeas [07:54] *woes [07:55] https://github.com/snapcore/snapd/pull/8571 is super-simple and needs a 2nd review [07:55] PR #8571: overlord/snapstate: warn of refresh/postpone events [07:55] just warnings behind an experimental flag (though the flag is check in the code that is not in the diff) [08:04] PR snapd#8571 closed: overlord/snapstate: warn of refresh/postpone events [08:05] PR pc-amd64-gadget#46 closed: gadget.yaml: bump edition [08:07] \o/ [08:07] thank you mvo! [08:11] https://github.com/snapcore/snapd/pull/8563 needs a 2nd review, pstolowski maybe you cuold take a look, since we both reviewed the other bulk assertions PRs? :) [08:11] PR #8563: asserts/internal: introduce Grouping and Groupings [08:13] PR snapd#8603 opened: tests: pair of follow-ups from earlier reviews [08:13] mborzecki: yes absolutely, i meant to, thanks [08:14] ok :) [08:22] huh [08:22] gnome 3.36.2 has this changelog entry [08:22] - Reenable TLS 1.0/1.1 protocols due to COVID-19. [08:22] !??! [08:23] mvo: pstolowski: hi, #8602 will need to be cherry picked for 2.45, right? [08:23] PR #8602: configcore: only reload journald if systemd is new enough [08:24] pedronis: it looks like it, yes [08:25] mvo: I put the milestone on it [08:25] pedronis: I requested your review on https://github.com/snapcore/snapd/pull/8566 -- it is the principal part of the refresh-app-awareness that is yet to be designed [08:25] PR #8566: cmd/cmdutil: add run inhibition operations [08:25] this PR contains my take on the idea [08:28] ok, I probably need to look at older PRs first... hopefully I can go back doing some reviewing later today [08:29] ok [08:29] if you want to talk about it apart from the review and have some time just ing me [08:29] *ping me [08:29] I have enough to work on that it is not a problem [08:29] (not urgent) [08:30] having an idea for next week would be brilliant [08:31] pedronis, mvo: yes [08:34] * zyga reboots to unbreak his touchpad :/ [08:35] Those ThinkPads.... [08:36] re [08:37] Those touchpads :P [08:37] zyga: pedronis: when is the desktop sync meeting? can you invite me in case we discuss the fonts issue? [08:37] s/pedronis/mvo/ ^^ [08:37] IIRC tomorrow [08:38] let me check [08:39] note, the meeting tomorrow is the code review sync [08:39] invited you [08:40] zyga: thanks! [08:41] zyga: fwiw, pulseaudio failed again here: https://github.com/snapcore/snapd/pull/8563/checks?check_run_id=648645627 [08:42] PR #8563: asserts/internal: introduce Grouping and Groupings [08:42] with master from this morning [08:42] looking [08:43] hmmm, interesting [08:43] I was adding some precondition code [08:43] but perhaps something else is going on here: look that we check for the socket first [08:43] and initially it is not there [08:43] then it shows up [08:43] pulseaudio --check looks if the daemon is there [08:44] but does not connect to it IIRC [08:44] thanks, let me try something and I'll send a PR shortly [08:45] it reads /run/user/UID/pulse/pid [08:45] so perhaps we are socket activated but not really running [08:50] pedronis: I have the logs now, feel free to restart or override [08:51] mvo: did you squash merge the watchdog one? it had the label [08:51] pedronis: yeah, I overlooked it and realized when pressing merge :( [08:51] * mvo is a muppet [08:52] * zyga hugs mvo [08:52] PR snapd#8604 opened: interfaces/builtin/desktop: do not mount fonts cache on distros with quirks [08:52] mvo: we need to ping Taipei when it's in edge [09:01] pedronis: I added a comment to the bugreport asking for testing [09:01] pedronis: I can also add it to the sync doc [09:01] thx [09:02] pedronis: added to the sync doc as well, if we need to revert I will take care of this (to make up for my blunder to not squash it) [09:03] let's see how it goes [09:18] fascinating :) [09:31] some seriously weird stuff is happening [09:36] so I ran the pulseaudio test and tweaked the prepare section a little [09:37] no other tests, no other stuff [09:37] on *some* runs pulse starts and works [09:37] on some runs pulse starts, dies and cannot recover [09:37] starting it again fails [09:37] I have a debug shell, can strace and stuff [09:37] just wondering what the real issue is [09:40] zyga: i have a fascinating story to share, https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1876583 i can reproduce it in a terminal that has 80 colums, with that specific locale, because the unicode messages add up to fewer bytes than expected [09:40] Bug #1876583: "panic: runtime error: slice bounds out of range" while installing any snap package

[09:40] heheh [09:41] are we slicing utf-8 by bytes? [09:42] PR snapd#8603 closed: tests: pair of follow-ups from earlier reviews [09:43] zyga: i think the translation is wrong, not what we expected [09:44] mvo: zyga: where are the non-packed translations for snapd? [09:44] dunno [09:44] what do you mean by non-packed? [09:44] probably in rosetta [09:44] but just guessing [09:46] * zyga tests a fix for pulseaudio tests [09:47] and I suspect I know why it passed sometimes too [09:47] the long story short is that pulse quits when idle [09:47] it starts up on specific request (socket activated) [09:47] and just quits [09:48] the two sanity chceks up front are checking the socket and the service [09:48] and if fast enough, pulse is still running after starting the session [09:48] (it takes a moment to start and quick) [09:48] *quit [09:48] so the solution is just not to run pulseaudio --check [09:49] #. TRANSLATORS: this needs to be a single rune that is understood to mean "minutes" in e.g. 1m30s [09:49] msgid "m" [09:49] msgstr "хв" [09:50] heh [09:50] so this breaks our message/column calculation [09:50] and the translator probably said "what on earth is a rune?" [09:51] nevertheless, we shouldn't panic :P [09:53] mborzecki: nice [09:53] mborzecki: also this requirement that it needs to be a single rune is silly, I mean, do we know that it always is in all the languages :) ? [09:54] mborzecki: (not your fault of course, just saying) [09:54] mvo: well, yeah agreed ;) [09:54] mborzecki: thanks for looking into this! [09:54] jamesh: lol, indeed [10:09] pedronis: https://github.com/snapcore/snapd/pull/8605 fix for what you reported [10:09] mborzecki: ^ fun [10:09] PR #8605: tests: fix racy check for pulseaudio activity [10:10] PR snapd#8605 opened: tests: fix racy check for pulseaudio activity [10:26] mborzecki: you need to adjust spread test in the fix for fonts https://github.com/snapcore/snapd/pull/8604/checks?check_run_id=648902472 [10:26] PR #8604: interfaces/builtin/desktop: do not mount fonts cache on distros with quirks [10:26] aah right [10:31] zyga: I've left a comment on your PA PR [10:31] thanks, looking [10:33] jamesh: replied [10:34] zyga: waiting for default.target would probably be worthwhile, yeah. [10:35] yeah, let's do it [10:35] I will keep the check for "active" as a sanity check [10:36] and add an extra patch that waits in prepare [10:37] zyga: you should be able to rely on the return value of "systemctl is-active" rather than needing the extra MATCH call [10:38] jamesh: yeah, just more obvious what went wrong when it does fial [10:38] though I guess the log would contain that as well [10:39] hmm, default.target does not seem to be related to sockets [10:39] perhaps I should wait for sockets.target as well? [10:41] jamesh: something like this: [10:41] https://www.irccloud.com/pastebin/qZYsG5ek/ [10:43] zyga: probably. That's the one I try to start in e.g. tests/main/snap-session-agent-service-control/task.yaml [10:44] ah, indeed [10:45] zyga: actually, default.target should handle it [10:45] I _suspect_ so but I've started both [10:45] default.target requires basic.target, which requires sockets, timers, and paths.target [10:45] ah [10:45] I can drop sockets then [10:50] jamesh: please look again [10:51] heh, that fix isn't as simple as i thought [10:51] mborzecki: isn't that my line? :D [10:52] nothing is [10:53] zyga: looks good! [10:55] thanks! [10:56] quite a journey to test session-level user code [11:06] zyga: fwiw, I think the session-tool changes have made the tests in my user-daemons PR more reliable. I'm still seeing an occasional failure where connecting to the session agent socket fails, but that seems less common too [11:06] jamesh: do you have an idea why that could be the case? [11:07] zyga: sadly no. I've added some extra debug output, but I'm not sure what the cause was. [11:07] ok [11:07] well, little by little :) [11:07] I think we made a lot of progress [11:08] apart from gradual growth of reliability [11:08] writing tests is actually easier [11:08] less mucking with environment varibales, chown, quoting su -l ... test correctlyt [11:08] zyga: I thought it might have been a stray session from a previous test, but you fixed things to shut down the user slice [11:08] yeah [11:08] it's a surprise it doesn't shut down though [11:08] so many weird things across releases of systemd [11:12] * zyga -> afk [11:35] PR snapd#8606 opened: progress: fix progress bar with multibyte duration units [11:35] mvo: ^ [11:38] mborzecki: nice [11:52] PR snapcraft#3108 opened: tools: install wheel in snapcraft-dev [11:55] niemeyer: could you have a look at https://github.com/snapcore/spread/pull/104 when you have time? it's needed for the encrypted partition creation tests on gce [11:56] PR spread#104: Adding option for systems created on google to start with secure boot enabled [11:59] re [12:00] cachio: could you have a look at the error message mvo reported in https://github.com/snapcore/snapd/pull/8551? it seems that we have an unsupported system error there [12:00] PR #8551: snap-bootstrap: remove create-partitions and update tests [12:01] (is it caused by the still missing spread PR?) [12:07] cmatsuoka, I know the problem [12:08] kenvandine: hey, is there any plan to surface snap warnings to the user in a desktop session? [12:09] cmatsuoka, in nested.sh [12:09] in functinos get_google_image_url_for_nested_vm and get_ubuntu_image_url_for_nested_vm [12:09] plase add a * at the end of any system [12:09] cmatsuoka, I can fix it otherwise [12:10] as you prefer [12:10] cachio: the name would become ...-.img*? [12:11] cmatsuoka, no [12:11] or ...*.img? [12:11] ubuntu-16.04-64) -> ubuntu-16.04-64*) [12:11] ah, got it [12:11] thanks [12:11] for all the systems in both functinos [12:12] cmatsuoka, yaw [12:14] PR snapd#8607 opened: tests: remove user.sh [12:20] Bug #1876482 changed: There is no way to trace the upgrade history of a snap [12:46] degville: this is the line I use for non-encrypted installs (it's a stripped down version of the other one, possibly it can be further reduced): https://pastebin.ubuntu.com/p/8tfnS642yM/ [12:47] (smp 4 because I have a lot of cpus, the typical user will probably want smp 2) [12:59] cmatsuoka: thank you! perfect. [13:03] PR snapd#8608 opened: configcore: issue a warning on core16 when journal.persistent option is set [13:15] pstolowski: https://github.com/snapcore/snapd/pull/8608/files#r420780119 [13:15] PR #8608: configcore: issue a warning on core16 when journal.persistent option is set [13:15] zyga: thanks! [13:51] ijohnson: could you look at https://github.com/snapcore/snapd/pull/8605/commits/e3f01f69bb67376ac0f9065493b31d9e4efcd3f0 -- it's a part of thet fix but I pushed it after you approved [13:51] PR #8605: tests: fix racy check for pulseaudio activity [13:51] sure [13:51] the *essential* change is https://github.com/snapcore/snapd/pull/8605/commits/e3f01f69bb67376ac0f9065493b31d9e4efcd3f0#diff-a8fce2de6a6829d98f798117050b6c93R134 line [13:51] everything else is the consequence of that [13:56] zyga: we haven't really put any thought into that [13:56] hey! [13:56] zyga: got an example? [13:56] sure [13:57] one main example is the warnings we show when an app was not refreshed when it was open [13:57] and when it got refreshed despite being open [13:57] ah [13:57] but as we just stared to use warnings now, there will be many more [13:57] well that is something there are plans for [13:57] we *could* convert those two to special events that are handled differently as well [13:57] just wanted to start a conversation about it [13:57] the whole "firefox needs a refresh, do you want to restart to refresh now?" [13:58] not quite, that's not what this is [13:58] what you describe is more like an interacting application that is snap aware [13:58] let's say gimp is snapped [13:58] and it has a refresh [13:58] and we just tried but postponed it because you had gimp open [13:58] without gimp being snap aware, we emit a warning [13:59] now that warning shows up in command line [13:59] when you interact with "snap" in any way [13:59] that's the context [13:59] you *can* see this today on edge [13:59] just run an app, switch to a different channel and refresh [14:00] (with refresh app awareness enabled) [14:02] kenvandine: so in my naive thought [14:02] I had two ideas [14:02] 1) we could show a desktop notification that just goes away [14:02] it's a one off it flies off [14:02] but you see it (if you look at the screen) [14:02] 2) we could show them in gnome-software [14:02] bonus points for showing this in the context of a particular app [14:03] * roadmr awards zyga 💯 bonus points 👾 [14:03] woot :) [14:04] * zyga notices his ideas left people speechless ;-) [14:27] * zyga -> lunch [14:35] zyga: we're looking to monitor snapd for those refresh warnings and provide a notification with a button to restart the app. This way the app doesn't need to know how to talk to snapd [14:35] we could surface other warnings the same way, perhaps with a different action attached. [14:40] pedronis: mind if i merge master into #8564? [14:40] PR #8564: asserts: introduce Pool [14:40] pstolowski: no [14:54] PR snapd#8607 closed: tests: remove user.sh [14:54] thank you [14:54] mvo: after https://github.com/snapcore/snapd/pull/8605 we will have much less red PRs [14:54] PR #8605: tests: fix racy check for pulseaudio activity [14:55] just a little more :) [14:55] brb [14:56] zyga: nice one [14:56] PR snapd#8609 opened: Adds missing paths to desktop, solves lp:1876804 [14:58] PR snapd#8601 closed: Adds missing paths to apparmor, solves lp:1876804 [14:59] pedronis: 8563 has two +1, should I merge or do you want to push the tweaks to it (some style suggestions in there)? [15:00] mvo: please merge, I can do the tweaks in a follow up or the follow up, depending [15:01] ijohnson, hey, did you have this issue with pi image https://paste.ubuntu.com/p/f6C5BD9NK8/ ? [15:02] PR snapd#8563 closed: asserts/internal: introduce Grouping and Groupings [15:10] cachio: how big is the SD card ? [15:10] ijohnson, 32GB [15:11] cachio: how big is the image you are trying to write? [15:12] 8GB [15:12] hmm I have only see that issue with no space left on device when the SD card is physically too small [15:12] cachio: maybe try deleting all partitions on the SD card first before flashing? [15:12] ijohnson, I already did it [15:13] cachio: does it happen everytime for you that it fails to write with No space left on device ? [15:13] ijohnson, first time trying to write a core 20 image for a pi3 [15:13] cachio: I'd say just try again, I don't know why that would happen for you [15:14] I'll download another image [15:25] ijohnson, I see the problem [15:26] the systemd detects my sdcard as it'd have 3.3GB [15:26] instead of 32GB [15:26] not sure why [15:26] it is doing the same with all mys cards [15:29] there's one more race in that pulse test [15:29] pulse server writes the cookie [15:29] but we race trying to read it [15:29] I fixed that locally, will push in a while [15:30] cachio: ahh that's weird [15:31] pstolowski: I tried to answer your questions in the review [15:33] pedronis: thanks, checking [15:42] ijohnson: I pushed to https://github.com/snapcore/snapd/pull/8605 again, to fix one more race [15:42] PR #8605: tests: fix racy check for pulseaudio activity