[00:03] Bug #1841001 opened: removing docker snap leaves apparmor misconfigured === mup_ is now known as mup === urluck_ is now known as urluck [05:16] morning === juergh_ is now known as juergh [06:24] mborzecki: hey, good morning [06:26] mvo: pedronis: morning guys [06:30] PR snapd#7304 closed: many: move channel parsing to snap/channel [06:30] PR snapd#7305 closed: check-pr-title.py: allow {} in pr prefix [06:31] PR snapd#7303 closed: interfaces/wayland,x11: allow reading an Xwayland Xauth file. Fixes LP: #1840925 [06:33] mvo: check this out https://i.imgur.com/4P10y8J.png [06:35] mborzecki: woooooooo [06:35] blender 2.80 33 stable blenderfoundation* classic [06:35] blender_279 2.79b 20 2.79 blenderfoundation* classic [06:35] mborzecki: thats really cool [06:38] mvo: didn't try any rendering though, hope they don't do some funny ipc with named sockets/semaphores that could conflict [06:42] mborzecki: right [06:43] PR snapd#7230 closed: tests: make interfaces-contacts-services cleanup more robust === jwheare_ is now known as jwheare [06:54] mborzecki: do you have a spread test (a very simple) one already for the parallel install on classic? [06:56] mvo: not yet, wanted to update zyga_'s mount-ns test, but doing some reviews first [06:56] mborzecki: ok, just curious [07:11] mornings [07:18] good morning pstolowski ! [07:22] hey [07:22] I got renamed to zyga_ and got banned/muted, not sure why [07:23] I didn't realize nothing got through [07:23] mborzecki: nice work [07:29] looking at ian's PR, got confused by task handlers again, iirc when a do callback fails, it should clean all the work it did internally right? so for instance if doLinkSnap fails on say, createSnapCookie it should run UnlinkSnap() (which isn't done really)? [07:31] mborzecki: not entirely [07:31] zyga: happend to me earlier as well! [07:31] mborzecki: when a handler fails it should clean up itself [07:31] mborzecki: when a handler finishes but another fails later [07:31] mborzecki: the undo handler for that task is called [07:31] mborzecki: that's the current approach [07:31] it's a bit hard to use TBH and hard to get right [07:32] it would be nice if we ran the "undo" one automatically but none of the code is ready for it [07:32] I spent way too long last night fighting cgroups [07:32] right, so if something inside doLinkSnap() fails after backend.LinkSnap() is called, it should ideally call backend.UnlinkSnap() before returning from the handler [07:32] there's something wonky going on that doesn't make sense to me yet [07:33] especially that it spans both ancient systemd and super recent systemd (on arch) [07:33] but not on other systems [07:33] mborzecki: I think it's well worth having a discussion about task handlers [07:33] mborzecki: I doubt we write them correctly today, it's very hard to do [07:37] mborzecki: in addition I would raise that the use of get/set is very pythonish [07:37] and doesn't fit go [07:37] it'd be nice if there were actual types [07:37] zyga: snap get/set ? [07:37] no [07:37] task handlers call get/set on task state [07:38] it's all a bit of wild west without typing [07:38] task state that is [07:51] pstolowski: about 7081 [07:51] I'm writing one quick test to check something and then if that fails I'm +1 it [07:51] brb [07:54] zyga: great, ty :) [07:57] * zyga writes actual hand-made test that could be changed to spread test later on [08:09] noticed some PRs form yday were restarted, anyone know what failed there? [08:09] pstolowski: Chipaca: hey [08:10] mborzecki: 👋 [08:20] PR snapd#7307 opened: mkversion.sh: fix version from git checkouts [08:24] hey Chipaca === Trevinho_ is now known as Trevinho [08:38] zyga: what exactly is needed for 7299? [08:38] mvo: let me push it [08:38] zyga: ta [08:38] zyga: you may need to pull first but there should be no conflicts [08:38] zyga: I just applied your suggestions [08:38] #7271 and #7135 need 2nd reviews [08:38] PR #7271: many: generalize assertstate.Batch to asserts.Batch, have assertstate.AddBatch [08:38] PR #7135: image: support prepare-image --classic for snapd snap only images [08:39] 7287 also needs a second review, should be an easy win [08:40] PR snapd#7297 closed: cmd/snap-mgmt: set +x on startup [08:42] mvo: ah, I see it is only running on 18.04 [08:42] zyga: I will work on your suggestions for 7288 now to land it I just saw another failure. or maybe we land as is and I do a followup? [08:42] zyga: it should unbreak some missing cleanups [08:43] 7300 was failing iirc [08:43] mvo: +1 [08:43] mvo: ty for pushing the naming fix to #7306! [08:43] PR #7306: overlord/configstate: sort patch keys to have deterministic order with snap set [08:44] pstolowski: no worries, looked so trivial I figured it would do no harm [08:44] mvo: https://github.com/snapcore/snapd/pull/7299 +1 [08:44] PR #7299: sanity: report proper errror when fuse is needed but not available <⛔ Blocked> [08:44] https://github.com/snapcore/snapd/pull/7299#discussion_r316560942 [08:45] PR snapd#7288 closed: tests: cleanup "snap_daemon" user in system-usernames-install-twice [08:51] zyga: mvo: https://github.com/systemd/systemd/issues/10872#issuecomment-523809771 [08:52] mborzecki: interesting! [08:52] mborzecki: I need to change our stuff [08:52] thanks! [08:53] zyga: btw. make fmt keeps changing cmd/libsnap-confine-private/error.c cmd/snap-confine/udev-support.c :/ [08:53] mborzecki: indent suuucks [08:53] mborzecki: just don't use it [08:54] mborzecki: run it once and commit the changes to the new code [08:54] mborzecki: discard the rest [08:54] zyga: yeah, too small for a seaprate PR though [08:55] zyga: maybe i'll add those in a separate patch in the parallel installs PR [08:55] PR snapd#7308 opened: tests: add new "user-tool" helper and use in system-user tests [08:55] mborzecki: I'm confused, what do you mean [08:55] mvo: oh :D [08:55] mvo: I was writing user tool just now [08:55] though my version is in python :D [08:55] mvo: let's get your version, I will only replace the implementation on top, ok? [08:56] zyga: https://paste.ubuntu.com/p/pQN7XFWP3V/ that's the diff, since it's fairly small, too small for a PR, so i'll add it to 7302 [08:57] mborzecki: don't change that please [08:57] mborzecki: my suggestion is to respect indent only on new code [08:57] zyga: hm ok [08:57] and not change existing files exactly because indent is unstable [08:57] mborzecki: and new files, if possible, should be formatted with clang-format [08:59] zyga: can you also re-review #7287? [08:59] PR #7287: hookstate/ctlcmd: snapctl unset command [08:59] pstolowski: enqueued [09:00] zyga: fifo or lifo? ;) [09:02] pstolowski: life-o [09:04] lipo? last-in, pooped-out? :-p [09:06] lipo is the opposite of pooped-out tho [09:07] hoovered-out more like [09:08] mvo: did you see #1841001 ? [09:08] Bug #1841001: removing docker snap leaves apparmor misconfigured [09:09] mvo: I saw that [09:09] mvo: it's not easy to fix [09:09] mvo: we discussed that at the last sprint [09:09] we can only remove apparmor profiles after all processes using them are gone [09:10] but that's hard to measure now [09:10] Chipaca: I did not see it :/ [09:12] mvo: https://github.com/snapcore/snapd/pull/7309 [09:12] zyga: that's removing them from the kernel though, isn't it? [09:12] PR #7309: tests: re-implement user tool in python [09:12] Chipaca: can you re-state your question? [09:12] Chipaca: we are not removing the profiles [09:12] zyga: as opposed to removing them from disk? Or is this bug about in-kernel? [09:12] PR snapd#7309 opened: tests: re-implement user tool in python [09:12] Chipaca: because removing them would remove them from running processes [09:12] Chipaca: it's about kernel [09:12] Chipaca: they are gone from disk [09:12] ah [09:12] Chipaca: they persist in the kernel until you reboot [09:13] yeah :/ [09:14] zyga: what's "any-python"? [09:14] Chipaca: tests/lib/bin/any-python [09:14] ah [09:14] I wish it wasn't needed [09:14] I wish unix had py [09:14] but ... [09:14] well :) [09:16] mvo: https://twitter.com/rbtcollins/status/1164327713269620737 [09:21] zyga: uh [09:21] zyga: thanks [09:22] zyga: we had this discussion before, we suck and we need to fix it :( [09:22] there are three bugs there [09:22] I'm going through them now [09:23] zyga: thank you [09:23] zyga: nice picture of robert, haven't seen a picture of him in years, I'm quite impressed :) [09:23] https://bugs.launchpad.net/snappy/+bug/1841001 [09:23] https://bugs.launchpad.net/snappy/+bug/1840244 [09:23] https://bugs.launchpad.net/snappy/+bug/1656976 [09:23] Bug #1841001: removing docker snap leaves apparmor misconfigured [09:23] Bug #1840244: docker snap cannot bind mount ssh sockets correctly [09:23] Bug #1656976: classic mode cannot start services [09:23] all triaged, the last one is easy-ish [09:23] the other two are not [09:24] the middle one is an UX usability bug [09:24] maybe we need to make twitter our bts? [09:24] the first one are two bugs in one [09:24] one needs to be handled by docker snap [09:24] the other by us eventually [09:24] I'll raise this at the standup [09:25] mvo: who wants to triage the remaining NEW bugs? [09:26] I need that coffee [09:26] zyga: I think we should talk in the standup but to make this work we will need something like bug triage duty on specific days [09:26] zyga: I think [09:26] zyga: like the security team is doing [09:26] yeah [09:26] it cannot be knee-jerk reactionary like that [09:26] zyga: or we need to put someone on it (like sergio) and make it his reposonbility [09:30] coffee sounds like a plan, zyga [09:30] yeah [09:30] * zyga goes to make some [09:39] PR snapd#7310 opened: devicestate: add missing test for remodeling possibly removing required flag [09:39] mvo: ^ there's a bit of doSetModel that was not unit tested that I need to tweak in the core20 model work [09:39] mvo: is #1579268 really an issue with snapd? I thought it was an issue with the snaps themselves [09:40] Bug #1579268: Mouse cursor is different inside graphical windows of snaps (snaps not using system theme) [09:42] hah, we've just had that mentioned in the Ubuntu Podcast Telegram channel :-) [09:43] Chipaca: I think its a issue with the snaps but iirc zyga was looking into this a while ago, not sure if we need to tweak anything on our side [09:44] mvo: there's also the issue that if people use themes that aren't packaged they're not usable from the snap [09:44] mvo: that aspect i'd say is wontfix? or did we have plans for that? or is that what zyga was looking at [09:46] Chipaca: we had plans for this, the idea was to have theme-snaps that would be auto-downloaded [09:46] Chipaca: this was a colaboration with the desktop team, there should be a forum post somewhere [09:46] * mvo looks [09:47] Chipaca: jamesh is a good POC I think [09:48] only a POC? I think james might claim to be GA/RTM [09:48] at least MVP [09:49] diddledan: I don't think jamesh is an airport in Colombia [09:49] really? [09:49] diddledan: 87% sure [09:50] I think this needs more investigation [09:55] Bug #1579268 changed: Mouse cursor is different inside graphical windows of snaps (snaps not using system theme) [09:58] Bug #1581188 changed: How to run snappy from inside Docker - systemd issues? [10:00] I'm doing `snap set snap-store-proxy.internal.store.id=''` but the value does not appear to change. Is this likely to be a problem with the snap-store-proxy configure hook, or am I doing something wrong? [10:01] ideally, I'd like to set it to null/no value [10:02] ogra: you around? [10:02] yes [10:02] tomwardill: doesn't =null do that? [10:03] and no, i have no idea about that nextcloud box :P [10:03] ogra: #1587453, i'm not sure if that's an Invalid or a Won'tFix or a Confirmed Wishlist [10:03] Bug #1587453: Reduce write to SD cards [10:03] Chipaca: `=null` will work on another key, but the value doesn't change on the key I care about [10:03] so I guess it's my problem somewhere :) [10:04] tomwardill: sounds like a validation bug yeah [10:04] tomwardill: also note there's going to be 'snap unset' soon [10:05] Chipaca: ah, that sounds like what I'm after :) [10:05] pstolowski: what release is snap unset scheduled for? [10:05] is it even meant to be user changable considering it's marked internal? [10:05] Chipaca, wontfix unless you plan to hack up disk cache settings randomly [10:06] Chipaca: it partially landed for 2.41 (and what landed is fully usable), however 'snapctl unset' didn't make it [10:07] Bug #1587453 changed: Reduce write to SD cards [10:07] Bug #1593151 changed: Snapd should support custom servers [10:10] Bug #1593435 changed: snap install will go get store version instead of sideload [10:13] Bug #1593558 changed: sox not configured for pulseaudio when packaged in a snap [10:13] Bug #1593915 changed: Ad the "snap all refresh" command [10:16] Bug #1593958 changed: Downloading snaps takes too long [10:16] Bug #1593960 changed: snap refresh unfriendly error message [10:19] Bug #1593989 changed: snap installed .desktops collide with .deb installed .desktops in unity7 [10:20] all the bugs! [10:20] PR snapd#7311 opened: snap/naming: introduce SnapRef, Snap, and SnapSet [10:22] PR snapd#7312 opened: tests: use user-tool to remove test user in the non-home test [10:22] Bug #1594328 changed: snappy keeps several copies of the same snap [10:24] PR snapd#7313 opened: many: add the start of Core 20 extensions support to the model assertion === pedronis_ is now known as pedronis [10:25] mvo: ^ [10:26] pedronis: nice [10:28] did someone turn on the firehose? http://www.belch.com/img/firehose.jpg :-p [10:28] Bug #1580738 changed: exec in desktop file should be app name and not include the friendly (current) package name [10:28] Bug #1594842 changed: Docs have "revision" as an int [10:31] Bug #1600140 changed: App indicator adds an entry for each snap launch [10:31] Bug #1600489 changed: freecad after being installed with snap doesn't exists [10:33] mvo: #1606539 is fixed, yes? [10:33] Bug #1606539: handler errors from `snap create-user` gracefully [10:34] Bug #1601859 changed: Need a way to autostart applications (not services) [10:37] Bug #1607252 changed: User data is missing or lost when utilizing a snap instead of a traditional package [10:37] Bug #1607744 changed: "snap install " over an already installed snap doesn't copy the user data [10:37] Bug #1609757 changed: Enable ordering of services provided by a snap [10:39] pedronis: were you aware of the request in #1609864 ? [10:39] Bug #1609864: Enable a command to be run on machine shutdown [10:41] pstolowski: https://github.com/snapcore/snapd/pull/7306 [10:41] PR #7306: overlord/configstate: sort patch keys to have deterministic order with snap set [10:41] zyga: ty [10:42] PR snapd#7290 closed: tests: allow test user XDG_RUNTIME_DIR to phase out [10:43] zyga: otoh, i'm slightly scared of loosing 'green' spread status on that PR... [10:43] haha, sure [10:43] follow up +1 [10:44] tough times [10:44] pstolowski: https://github.com/snapcore/snapd/pull/7287#pullrequestreview-278333771 +1 [10:44] PR #7287: hookstate/ctlcmd: snapctl unset command [10:45] pstolowski: I'm mid-way writing that spread test for batch profiles but I went upstairs for a moment and doing reviews from a laptop now [10:45] zyga: sure, thanks for trying to break it [10:47] Bug #1611526 changed: temp directories not deleted when snap fails to start [10:47] Bug #1611530 changed: can't install devmode snap from store [10:47] Bug #1611706 changed: Test suite failures on Arch [10:47] ah, 7287 is actually rebuilding anyway for previous changes... [10:47] pstolowski: 7287 can land now [10:47] oh [10:47] bummer [10:47] nah it can't [10:47] Chipaca: no [10:48] let's see, if it fails for random reason, i'll make the spread test tweak as well [10:49] jdstrand: can you please enqueue https://github.com/snapcore/snapd/pull/7252 [10:49] PR #7252: interfaces/network-manager: allow using org.freedesktop.DBus.ObjectManager [10:50] zyga: ah, no, 7287 can land indeed, ty, i confused my two PRs [10:50] Bug #1611837 changed: all-snaps: Boot breaks on reboot [10:50] Bug #1612783 changed: snapd requires U1 account to install local packages [10:50] Bug #1614730 changed: dpkg: dependency problems prevent configuration of snapd - snapd depends on ubuntu-core-launcher (>= 1.0.23) [10:50] PR snapd#7287 closed: hookstate/ctlcmd: snapctl unset command