=== phoenix_firebrd_ is now known as phoenix_firebrd [04:30] Good morning [04:31] I did an experiment and the glob issue cannot explain the bug reported by our customer [04:31] I will focus on debugging that next [04:38] also the 2.32 branches cannot land because spread has issues on linode [04:38] many machine images don't pick up the ssh connection [04:38] must be Friday :) [05:07] morning [05:10] hey [05:10] mborzecki: there must be a 2nd bug [05:10] mborzecki: what I saw and fixed cannot explain the issue reported by the customer [05:10] mborzecki: I'd love to get to the bottom of 2.32 test failures [05:10] mborzecki: it seems we cannot reach our booted images anymore. [05:11] mborzecki: this is all on linode [05:11] mborzecki: for logs see #5110 [05:11] PR #5110: interfaces/apparmor: fix incorrect apparmor profile glob (2.32) [05:17] globs everywhere [05:20] had to double check what SecurityTagGlob looks like [05:24] zyga: travis build failure is the same as in #5073, my guess is that the images do not work, iirc those were -32 ones [05:24] PR #5073: set up journal streams in user session application autostart (2.32) [05:24] if we do .6 it woul dbe nice to get 5073 in as well [06:03] mborzecki: I asked mvo about what else to include but IIRC he wanted critical things only [06:03] ok [06:04] I wonder why the -32 bit images broke on us [06:45] hm 249 test files modified, if i split that into groups of 20, we'll get 12 PRs [06:46] Sounds good [06:47] Will the match list conflict? [06:51] i'm afraid so, but if we land at least one of these PRs a day, we'll be done in 3 weeks :P === pstolowski|afk is now known as pstolowski [07:03] morning [07:05] pstolowski: hey [07:06] hey pawel [07:06] mborzecki: maybe a .d directory for match could do it [07:06] mborzecki: but maybe it is better to just review [07:06] I'm a bit tired this morning, not sure why [07:07] zyga: we have to review it either way ;) [07:12] need to drop my dogs off at the groomer, bb in ~30 minutes [07:20] I'm sleepy [07:30] ok, coffee is ready [07:30] let's fight [07:30] mborzecki, pstolowski: can you guys please have a 2nd look on #4545 [07:30] PR #4545: interfaces/x11: allow X11 slot implementations [07:31] it's green and I'd like to merge it but a fresh set of eyes would help [07:33] re [07:37] pstolowski: can you take a look at https://github.com/snapcore/snapd/pull/5103 ? [07:37] PR #5103: tests: shellcheck spread tasks [07:37] once we land it i'll open a PR with the first barch of fixes [07:41] zyga: will do [07:41] thank you [07:41] pedronis: hey, do you have a moment for quick HO? i'd like to run an idea re hooks by you [07:46] pstolowski: finishing breakfast, can in a little bit [07:47] sure [08:07] zyga: is the bug re incorrect glob responsible for the issue you discussed yesterday and were trying to reproduce? [08:07] pstolowski: yes [08:07] pstolowski: but I think there's another bug too [08:07] zyga: yay! [08:07] but I'm still zombified and think slowly [08:08] I wrote a test that this bug fix from yesterday doesn't affect the case saw by our customer [08:08] I have a device to test now [08:10] pstolowski: I'm available now [08:11] pedronis: ok, standup HO [08:28] actually [08:28] I wonder now [08:28] what is coming up [08:30] zyga: hm? [08:30] c-release [08:32] cagey camel [08:32] cheery chiru [08:34] crispy civet [08:34] croissant cat [08:34] cuddly cockroach [08:35] morning all [08:35] PR snapd#5089 closed: tests: adding google-sru backend replacing linode-sur [08:35] good morning :) [08:35] cuddly cockroach [08:35] haha [08:35] instant +1 from me [08:36] that's a bad idea. email filters will turn that into cuddly roach [08:36] crispy courgette [08:36] http://goodyfeed.com/wp-content/uploads/2017/06/cockroach3.jpg [08:36] ha. that's the way to deal with filters :-D [08:36] and make an official cooking recipe with it [08:38] we should make a nsfwbuntu that was just ubuntu with a nsfw dev name [08:38] heh, reminds me of fedora with their 'beefy miracle' thing [08:39] ssh! don't say the b word [08:40] it sets neal off [08:40] mborzecki: what's your thoughts on the drop privs thing? [08:41] go for the mutex [08:41] mborzecki: option a) say "can't fix before we move to 1.10+" [08:41] mborzecki: option b) put a mutex, with a comment that says "this can still end up with the wrong uid trying to do stuff but we couldn't reproduce it so it's probably very unlikely" [08:42] * Chipaca felt less and less confident about (b) as he wrote the comment for it [08:42] in fact, let me try something [08:58] mborzecki: https://pastebin.ubuntu.com/p/Mn8nF8VV62/ [09:01] zyga: cosmopolitan cockapoo [09:02] Chipaca: seems to work now [09:04] mborzecki: in which sense of the word 'work'? [09:05] mborzecki: lots of dots, no asterisks? [09:05] Chipaca: yes [09:05] mborzecki: go < 1.10? [09:05] * Chipaca hugs mwhudson for snapping go [09:07] pfff, got a couple of terminal-width lines of dots, switched back now and asterisks :/ it did take longer though [09:07] mborzecki: seems to vary, but yes [09:07] mborzecki: not surprised that a race is racy though :-) [09:07] not failed after 3 dots [09:08] s/not/now/ [09:08] i'm very intentionally tripping the bug, there [09:08] so in practice it'll be a lot less likely [09:08] even if we get someone to packport the patches to 1.9, the previous releases will stay unpatched [09:09] but with so many people using it, less likely is still quite likely [09:09] mborzecki: mwhudson has backported some cherries for us :-) but that won't help potato linux [09:10] an alternative that should work is to use cgo, clone a thread there, do the syscall there (we _don't_ want the libc wrapper) [09:11] but, ew [09:13] also the semantics of calling from cgo back to go are not nice [09:22] niemeyer: would appreciate your thoughts on https://github.com/snapcore/snapd/pull/4983#issuecomment-384915539 [09:22] PR #4983: osutil/sys, client: add sys.RunAsUidGid, use it for auth.json [09:23] niemeyer: (not on the PR, on the comment) [09:26] Chipaca: do you recall if it's also the case in older releasese that if a goroutine exits in a locked thread, the thread will be terminated? [09:35] zyga: conflicts in #5081 [09:35] PR #5081: interfaces/apparmor: add chopTree [09:35] mmm [09:36] Chipaca: ca you take another look at #5103 ? [09:36] PR #5103: tests: shellcheck spread tasks [09:36] fixed [09:39] mborzecki: sorry was afk getting coffee (and maybe cake) [09:39] mborzecki: it's very much _not_ terminated [09:39] Chipaca: nice combo [09:39] especially this cake :-) [09:39] my mum's chocolate fudge cake [09:40] Hi. What is the equivalent of "apt-file list " for displaying the contents of a snap package (e.g., "nextcloud") without having to install it first? (I'm using Ubuntu Xenial.) [09:40] uebera||: there isn't one [09:40] uebera||: what are you trying to do? [09:41] uebera||: (outside of some corner cases, AFAIK the files in a snap are not that interesting) [09:42] I want to see what's inside the package "nextcloud" w/o installing it. [09:42] uebera||: I meant one step back from that [09:43] uebera||: I mean: what do you need this information for? [09:43] I don't understand. This is the only step. I do not intend to install the package using snap, I just want to see its contents in order to see what is included as a dependency (b/c I doubt that this is in line with what the documentation lists). [09:45] uebera||: there isn't a way to do that without downloading the snap [09:45] uebera||: snap download && unsquasfs -l probably makes most sense then [09:45] or snap install and find :) [09:46] it's a 200MB download, so I can understand not wanting to do it [09:47] uebera||: is it really just this snap, or is it a general problem? [09:48] Well, this is the first snap I want to inspect (the third one I've ever come across), but I'd consider it a general problem, yes. Unless there are public catalogues like for apt... [09:50] uebera||: ok, as I said, there isn't a way (today) of doing this; we don't index that data [09:50] snaps are also much more fast moving than debs, so the information can be stale quickly [09:50] heck yeah :-) [09:50] Well, if the above "snap download && unsquasfs -l" works, that's good enough for a start atm. [09:51] why on earth is mysqld 200MB+ [09:51] uebera||: also -lls fwiw [09:51] However, having to start up a throw-away container just to inspect the contents of a snap in a safe environment is not convenient. [09:51] uebera||: why would you need a container for download+unsquash? [09:52] Not for download, but for installation. [09:52] PR snapcraft#2107 closed: Release changelog for 2.42 [09:52] I.e., the "download/unsquash" method should be cited in a FAQ if it isn't already (I've missed it until now). [09:53] uebera||: right, for installation i could understand it, as it's mounting the squash and you'd have to trust us that it's ok to do so [09:53] uebera||: well, download is documented, as snaps are documented as being squashfs [09:53] uebera||: the rest is just joining the dots [09:54] s/as/and/ [09:55] zyga: so, due to my classic snap segfaulting as I can only build it on bionic due to golang version, I heard there was a way to declare which core snap you are using, I don't find the reference for it in the doc, any hint? [09:56] didrocks: "base: core18" [09:56] but I don't know if snapcraft supports this yet [09:56] it's very early [09:56] didrocks: if you build your snap on bionic you must use a container [09:56] uebera||: hm alternative would be to try https://uappexplorer.com/snap/ubuntu/nextcloud and download from there [09:56] didrocks: and for classic snap you must use a xenial-abi-compatible build of go [09:56] zyga: the issue is that snapcraft doesn't allow you to select the go version if I'm correct [09:57] didrocks: perhaps, kalikiana would know [09:57] sounds like bug #1616985 didn't really move [09:57] Bug #1616985: the go plugin doesn't use go build-snaps [09:57] so, basically, I'm between FTBFS and segfault :p [09:59] didrocks: even with core18? [09:59] Chipaca: well, I'm unsure how to inject that in my snapcraft.yaml definition [10:00] try "base: core18" [10:00] same level than grade: I guess? [10:00] right next to "name: doesrocks" [10:00] :-p [10:00] yes [10:00] ;) [10:01] * didrocks tries, thanks! [10:01] didrocks: if it shouts at you let us know [10:01] mborzecki: Thanks, this is an alternative w/o using snap (though "snap download" worked like a charm) [10:02] didrocks: That's not perfectly accurate. The go plugin insists on installing the deb, but it doesn't prevent you from using a remote part with your favorite version [10:02] Chipaca: at least, it downloads core18 when installing, good starting point :) [10:02] uebera||: 'snap download' also gives you the assertions which can be helpful [10:02] didrocks: huzzah [10:03] and it works! thanks Chipaca, zyga [10:03] uebera||: a word or warning, iirc the search by architecture is broken in uappexplorer [10:03] kalikiana: yeah, that's what I read, but I prefer to avoid relying on something else like another snap or a ppa with a newer build [10:04] didrocks: I can't parse that sentence. You want a specific go version that isn't in the archive, no? [10:05] kalikiana: exactly! It seems the only way right now is to have a part downloading go for you (from another snap or using adding a ppa as a build-dep) [10:07] didrocks: My suggestion above was a remote part. Not a PPA. As in, "after: [go] go: source-tag: go1.7.5" [10:08] kalikiana: ah ok, nice that there is a remote part! I'll see between that or using core18… [10:08] thanks :) [10:08] didrocks: i'd consider core18 'developer preview' quality, fwiw [10:08] didrocks: tested for some very narrow verticals right now afaik [10:09] so if you can get it to work without, you'll be happier [10:09] Chipaca: well, as it's for a classic snap, I think I only need an ABI compatible linker with my build [10:09] and as it's Go without Cgo, it's statically compiled (apart from accessing the network config) [10:09] didrocks: in particular I'm entirely unsure how well core18 will play with classic, today [10:10] didrocks: and 'base: core18' means installing your snap will download core18 [10:10] didrocks: so... let us know how it goes :-D [10:10] Chipaca: which we are going to do anyway soon AFAIK for 3.28 GNOME snaps [10:10] yeah, I'm happy to be the guinee pig here ;) [10:10] guinea* [10:10] didrocks: I'm happy for you to be, as well [10:10] as long as you know you are [10:11] heh, I know at least there is a fallback plan [10:11] I'll keep you posted in case I spot anything wrong [10:16] didrocks: to be clear if it's going to fail anywhere it'll be on 16.04 (or 14.04) [10:16] yeah, I'll try testing there [10:16] didrocks: if it's a desktop thing, don't try 14.04 :-) [10:18] * didrocks will happily refrain thus! [10:44] had to restart, master travis job again, and again it failed because not output was received [10:51] does anyone know why I default to the edge channel in Ubuntu Software? [10:51] snaps I install are apparently coming from edge by default [10:51] which is something I totally did not expect [10:52] or at least, it looks like they are [10:52] it says "stable" channel but gives me the version in the edge channel... [10:55] PR snapcraft#2109 opened: sources: clean up IncompatibleOptionsError [10:57] Son_Goku: that looks like a bug for sure [10:57] it should be the opposite [10:57] zyga: it's very obvious with the blender snap [10:57] Son_Goku, it knows who you are ! [10:57] meep [10:57] :) [10:57] and after installing snap info says you are tracking edge? [11:09] zyga, nope [11:09] it switches back to stable [11:09] it's very strange and confusing [11:09] yes [11:09] but is the snap you got from edge? [11:09] as in the revision matches what is in edge and not what it is in stable [11:10] no it matches stable [11:10] hmm? [11:10] it looks like something's confused in ubuntu software [11:10] so what is the bug [11:10] even when tracking stable, it's actually rendering info from edge [11:10] in ubuntu software [11:11] also, niemeyer, you really should fix the metadata for your blender snap [11:11] it's completely wrong :) [11:12] Son_Goku: showing info from edge? [11:12] Son_Goku: tell me moar [11:12] Chipaca, ubuntu software renders info from edge, even though it defaults to tracking stable [11:12] https://www.youtube.com/watch?v=DOYRLdisrTE [11:12] Son_Goku: in what case? [11:12] (wrong video though ;) [11:12] snap not installed [11:12] Son_Goku: i mean: what do you see as 'from edge'? [11:12] Ubuntu Software on 18.04, just open it up, and search for blender [11:13] well, the version [11:13] yeah, reproducible here - gnome software it showing the version from non-stable (hard to see if it's edge, beta or candidate) [11:13] but it's not in stable [11:14] also, Fedora 28 is GOLD [11:14] sparkiegeek: Son_Goku: any example that isn't also a deb? [11:15] Son_Goku: congrats! [11:15] Chipaca, not sure, I noticed it because the blender snap is missing so much data it shows up as separate from the deb [11:15] the vlc snap also shows the same issue [11:15] Chipaca: just trying to check the snapd API, do you have the curl invocation to hand? [11:16] though it renders the beta channel's number [11:16] Chipaca: reproducible for gitlptools (my snap) [11:16] sparkiegeek: snap install http && http snapd:///v2/find name==thesnapname [11:17] not seeing this in 16.04, fwiw [11:17] gitlptools showing 388db20 as is proper [11:17] Chipaca: http: error: InvalidSchema: No connection adapters were found for 'snapd://v2/find' [11:17] eh [11:17] sparkiegeek: forgot the extra / [11:17] snapd:///v2/... [11:18] wait, no i didn't [11:18] you did :-D [11:18] $ http snapd:///v2/find name==gitlptools [11:18] [11:18] http: error: InvalidSchema: No connection adapters were found for 'snapd:///v2/find' [11:19] sparkiegeek: which http [11:19] ah ha! [11:19] ¯\_(ツ)_/¯ [11:19] Chipaca: thanks, that was it [11:20] so what is broken? [11:21] completely in the dark guess? it's taking the first entry from the channel map instead of latest/stable [11:21] (or just the one outside the channel map) [11:21] Chipaca, zyga: it looks like version number outside of the channel map is reporting the wrong one [11:21] right, the latter [11:21] ah, ok [11:21] then, yes, that. [11:21] bug in store, already getting fixed [11:22] thanks! [11:22] :) [11:22] the snap client just happens to walk around the issue [11:23] the snap that's installed is tracking stable though, and then gnome-software is reporting the installed version (as tested with blender) [11:23] so I can't exactly reproduce what Son_Goku sees [11:24] sparkiegeek: i thought it wasn't the installed version [11:24] oh, after install, it shows the right thing [11:24] right [11:24] it was _before_ install it didn't [11:24] so it was just confusing [11:24] Son_Goku: agreed [11:24] easiest and quickest and most future-safe thing would be to tell snapd-glib (or was it glib-snapd) to grab the info from the channel map [11:24] although... [11:24] i should sit down with 'em and make it better [11:25] they're having to jump through hoops to get their stuff done, and it's nasty [11:25] i'll do that next week [11:25] Chipaca, snapd-glib [11:25] Chipaca: right, but then they'd have to encode the channel map priorities - e.g. no guarantee of there being latest/stable [11:25] Facu: remember the 'ha ha lol the store picks a revision at random to show the info if you say channel=""' thing we talked about? gnome-software in 18.04 hits it :-) [11:26] that explains why I get the beta version rendered for vlc [11:26] rather than edge, if it's "random" [11:26] Son_Goku: well, it's not random, it's most-recently-pushed afaik [11:26] it just seems random [11:27] now, if the store was open source... :P [11:27] ugh, snap store shows vlc edge to install by default? [11:28] thresh: no(ish) [11:28] thresh: it installs stable; the version number is wrong [11:28] thresh: in 18.04 [11:28] that's still bad [11:28] don't want the user to have an impression they'll install vlc 4.0 [11:29] right [11:29] we'll sort it [11:29] someday :P [11:30] thresh: to be clear, it's gnome-software (and presumably other stores using snapd-glib but i haven't checked) in 18.04 that expose the bug [11:31] good to know, fingers crossed on a swift fix :) [11:31] we can argue about whether it's the store, snapd, or snapd-glib that have the bug -- but it's fixable in any of those places :-) [11:31] also congrats on a release for everyone involved! [11:32] thresh: i think everyone directly involved is off today :-) [11:34] Chipaca, ugh [11:34] Facu: :-) [11:34] Facu: also, good morning [11:35] Facu: sorry to greet you with that news === JanC is now known as Guest28882 === JanC_ is now known as JanC [11:38] holy shit [11:38] VLC looks ugly [11:40] Son_Goku: sshh! you'll make thresh cry :-( [11:40] nah, I already cry daily [11:40] Son_Goku: try saying the same thing but with kinder words [11:40] * Chipaca hugs thresh [11:40] Son_Goku, what DE are you under? [11:40] GNOME [11:40] Fedora 28 [11:40] Son_Goku, can you please try beta? [11:40] sure, one sec [11:42] * Son_Goku sighs [11:42] I can't do it from g-s [11:42] down the CLI we go! [11:42] oh we've gotten fancy now [11:42] overlaying progress bars [11:42] Son_Goku: how long has it been since you used the CLI :-) [11:43] more months than it probably should be [11:43] I'd release the beta to stable any time, but the UI (e.g. open file dialog) is completely broken on KDE [11:44] thresh, not better [11:44] hmm, okay [11:45] can you drop the output of env | grep XDG_ to pastebin somewhere? or PM me? [11:45] and a screenshot of VLC / menus [11:45] thresh: also maybe 'snap run --shell vlc' and do the same in there [11:47] thresh: http://kinginuyasha.enanocms.org/downloads/screenshot-vlc-snap-about-20180427.png [11:48] Son_Goku, what about the icons / general UI fonts? [11:49] in the menu bar etc === sergiusens_ is now known as sergiusens [11:51] thresh: http://kinginuyasha.enanocms.org/downloads/screenshot-vlc-snap-with-env-20180427.png [11:51] it's pretty bad [11:51] ok, so it's an issue with the fonts [11:51] Son_Goku: thresh: aren't themes and fonts working? [11:52] the UI actually does not look like windows 95 [11:52] i thought they were [11:52] Chipaca, I'm pretty sure I need to add something to my snap to enable it? [11:52] I thought they were too [11:52] but apparently no? [11:52] Son_Goku: can you file a bug about the version info being wrong? i.e. ubuntu-bug gnome-software-plugin-snap [11:53] sparkiegeek, I don't have the Ubuntu 18.04 VM anymore :) [11:53] I only spun it up to check out g-s there ;) [11:53] because I heard there was funny business in g-s in Ubuntu [11:53] like the ability to select channels and permissions [11:53] neither of which I can do from g-s in Fedora [11:53] zyga: do you know about fonts? [11:56] Chipaca: is this a tricky question where I say yes and then go and dig through unholy C code that renders vectors? [11:56] popey: https://twitter.com/zygoon/status/989835561639841797 :-) [11:57] popey: it still needs some upstreaming on my part [11:57] and packaging love [11:57] and more [11:57] zyga: I don't want you dissing metafont now [11:57] but :-) [11:57] that skype snap isnt confined [11:57] popey: yes but it's not because of opensuse [11:57] neither is slack [11:57] ok [11:57]