=== chihchun_afk is now known as chihchun [05:03] morning [05:08] mvo: hey [05:08] mborzecki: good morning [05:08] mborzecki: how are things? anything important I missed last week? [05:09] mvo: no, don't think so, no fires or anything [05:11] mborzecki: great [05:11] * mvo dives into core18 and reviews then [05:11] mborzecki: anything exciting happened? how are parallel installs coming along, I'm quite excited about this one [05:13] mvo: i've bastardized snap-confine and snap-update-ns to accept the _ in snap names, then on friday i was looking into mounting foo_bar as foo [05:13] mvo: i should be opening some PRs today with some introductory changes in structs [05:13] mborzecki: sweet! [05:13] mborzecki: I was wondering how the internal changes will look like, was it a lot of fiddling? [05:14] let me show you a patch [05:15] mvo: https://paste.ubuntu.com/p/xNBgYhWcMp/ that's a work in progress version [05:17] mborzecki: aha, nice. StoreName() is a good helper [05:18] mborzecki: that is surprisingly short, nice [05:18] mvo: tried to make it least invasive :P [05:18] mborzecki: heh :) looks like a success to me [05:22] mvo: s-c bits are a bit fiddly, the mounts, i'm trying tmpfs over $SNAP_MOUNT_DIR, oh an the store part, you can't really install foo and foo_bar right now, because the store complains that you have the same snap appearing twice in the request [05:23] * mvo nods [05:33] https://www.bloomberg.com/news/articles/2018-06-03/microsoft-is-said-to-have-agreed-to-acquire-coding-site-github wat? [05:35] mborzecki: I heard rumors about this on the weekend [05:35] mborzecki: definitely in the waaat catgeory [05:35] mborzecki: source-safe-2018 [05:35] 5251 is an easy win if someone wants to review [05:35] hahah, well, at leas this one is usable :) [05:36] mborzecki: yeah :) [05:39] PR snapd#5251 closed: tests: reprioritise a few tests that are known to be slow [05:39] good morning! [05:39] zyga: hey, good morning [05:39] hey hey, how are things [05:39] zyga: hey [05:40] so Microsoft keeps snapd source safe ;) [05:40] PR snapd#5249 closed: interfaces/home: remove redundant common interface assignment [05:44] I will be here shortly, need to set the little one for a school trip [05:46] mborzecki: there is one open question by zyga in 5244 otherwise it looks good afaict [05:50] mvo: just pushed a fix [05:51] mborzecki: ta [05:51] PR snapd#5238 closed: tests: skip appstream-id test for core systems 32 bits [05:53] PR snapd#5232 closed: interfaces/tpm: Allow access to the kernel resource manager [05:53] https://www.youtube.com/watch?v=VYOXuOg9tQI haha look at the publish date [06:17] mborzecki: hahaha [06:32] tests are not happy :/ [06:34] unattended upgrades? https://www.irccloud.com/pastebin/Txy5hEt3/ [06:43] zyga: possible, yes === pstolowski|afk is now known as pstolowski [06:56] morning [06:58] ok [06:58] * zyga got stressed because of sudden call that the school trip rule sheet must be brought by all children [06:58] school is so poorly organised [06:59] I signed that paper last week [06:59] and suddenly they tell me I need to keep it (I got a blank copy) and -bring it-, as if they couldn't find their own copy [06:59] sigh [06:59] but all is good now [06:59] * zyga tries to forget about non coding stuff === chihchun is now known as chihchun_afk [07:14] could anyone please review https://github.com/snapcore/snapd/pull/5245 [07:14] PR #5245: cmd/snap-update-ns: add PathIterator === chihchun_afk is now known as chihchun [07:17] zyga: i'll do it in a while [07:17] thanks [07:59] any clues what happened to lxd? restoring google:ubuntu-16.04-32:tests/main/lxd seems to consistently fail across a number of jobs [08:00] the job hits a kill-timeout when issuing `lxd.lxc stop my-ubuntu` === chihchun is now known as chihchun_afk [08:00] mborzecki: I saw this as well, I have not looked yet but probably worthwhile to do a local spread run with -debug [08:01] Chipaca: hey, thanks for pushing the fixes for shellchecks :) [08:03] mborzecki: I was very bored :-) === chihchun_afk is now known as chihchun [08:20] PR snapd#5252 opened: store, jsonutil: move store.getStructFields to jsonutil.StructFields [08:24] damn, that lxc issue does not reproduce when running locally :/ [08:26] mborzecki: the one where it doesn't stop? [08:26] Chipaca: yes [08:26] mborzecki: it's not 100% on spread either [08:26] that is, it's racy [08:27] heh, i must be super lucky then :) [08:27] i've seen it sporadically since friday [08:27] (while iterating on those shellcheck tests) [08:28] mvo: extra bonus idea for --format=help: use tags to add brief explanation to fields [08:28] mvo: I can push a tweaked helper to grab you that if we agree on a format for 'em [08:29] (or, you can, it's not brain rocketry) [08:34] 2018-06-04 08:29:45 Failed task restore: 1 [08:34] - google:ubuntu-16.04-32:tests/main/lxd [08:34] i saw that over the weekend too [08:35] mborzecki: ah, it seems like what you mentoned above [08:40] Chipaca: feel free to do that, I'm wrestling firstboot to skip looking for core [08:41] mvo: aw [08:41] mvo: wrestl wrestl [08:58] * zyga -> tea [09:03] PR snapd#5253 opened: snap: introduce new fields for parallel snap installation [09:17] and now i got a failure on journal cursor in the google:ubuntu-14.04-64:tests/main/snap-confine-from-core test [09:22] mborzecki: hi, could you write in a forum topic how you plan to refactor things on the way to parallel installs, small PRs are good but is hard to guess the full picture. [09:22] pedronis: sure [09:23] pedronis: btw. https://paste.ubuntu.com/p/xNBgYhWcMp/ if you'd like a quick peek at the current changes [09:25] pedronis: as you commented in the PR, SnapSetup.Name() has been updated too, just that it's not included int the patchset [09:25] mborzecki: there are no changes to tests there, bit hard to judge if those changes are sufficient, also the change in store is a bit strange [09:26] mborzecki: any place dealing with info.Name needs a new test [09:27] in theory not, but in practice it sort of does [09:29] pedronis: hmm i think i'm more afraid of the places that poke the struct directly rather than calling its methods, iirc there was one place where SuggestedName was accessed this way (store?) [09:29] mborzecki: wrappers I think [09:29] pstolowski: I think the journal change is racy and it somehow doesn't always work [09:30] mborzecki: anyway new tests in managers_test and spread are probably welcome [09:30] zyga: yes.. and somehow it's only on 14.04 (afair) [09:30] pedronis: sure, will be adding those [09:30] I don't know, think I saw that across the board [09:31] hmm ack [09:31] offtopic, so the GitHub news seems true, [09:31] I wonder what this means for us [09:32] zyga: what is it about github? [09:32] pstolowski: ms bought it [09:33] woot [09:35] zyga, they did ? i thought they only think aloud about it yet ... was there an official statement ? [09:36] (they also said they might only invest in shares and not buy the whole company afaik) [09:37] mborzecki: also store changes are a bit buggy , that code was sometimes assuming that instanceKey == snapID which is not true anymore and doesn't seem fully updated [09:41] ogra_: I don't know for sure, there's plenty of news but also rumour around this topic [09:41] I think we'll just wait and see [09:41] yeah [09:42] afaik most is speculation atm [09:44] (what is clear is that MS has "some interest" in GH ... but perhaps thats only to the level like they had interest in imagination tech. where they hold a third to have influence on the graphics driver development) [09:44] http://www.bloomberg.com/news/articles/2018-06-03/microsoft-is-said-to-have-agreed-to-acquire-coding-site-github [09:45] ogra_: heh, how imagination tech. handled being up for sale was a fiasco :) at least for imagination... [09:46] yeah, totally [09:46] haha "Bill Gates and Paul Allen co-founded the company to give hobbyists a way to program a new micro-computer kit, the MITS Altair." [09:46] thats soo ... LOL ... [09:47] s/to give hobbyists a way/to lock in hobbyists that already had ways to/ [09:49] heh, this one is also great: [09:49] "GitHub preferred selling the company to going public and chose Microsoft partially because it was impressed by Nadella ..." [09:50] ... "he looks so good" ... :P [09:50] you look good, are you exercising? [09:50] haha [10:00] pedronis: good morning! I updated 5217 based on your feedback [10:02] mvo: hi, looking [10:03] pedronis: thank you. I'm looking at the firstboot changes right now, I think I will need to reuse bits from 5217 here :) [10:05] PR snapd#5162 closed: overlord/hookstate: support undo for hooks [10:10] mvo: I think snapd should come first in the snaps installed sequence [10:11] pedronis: ok, let me fix this [10:11] we then take care again of that in first boot [10:11] but we should keep being consistent I think [10:12] pedronis: yeah, makes sense [10:15] pedronis: updated === Wimpress_ is now known as Wimpress [11:00] pstolowski: is your -1 because of the last comment, it wasn't super clear (the other one is wondering and one is a nitpick so I suppose so) [11:02] pedronis: yes just about Empty(); and i prefer to call it "requesting a change", not -1 because I'm +1 for you change overall :) [11:02] pstolowski: well, it really means "requesting a change and wanting to look again" [11:07] fair enough, it's just a minor thing, that's all i'm saying === pstolowski is now known as pstolowski|lunch [11:16] I'm under the impression that I can download any revision of a previously published store if I have the right developer access. Is that correct, or is there some kind of retention/expiry policy? [11:16] I'm trying to download r429 of git-ubuntu from the store, since r430 in edge broke what previously worked. [11:16] snap download --revision=430 git-ubuntu [11:16] worked [11:16] snap download --revision=429 git-ubuntu [11:16] says: error: cannot find snap "git-ubuntu": snap not found [11:17] I tried logging in with snap login, but after credentials that failed and suggested sudo [11:17] * cachio afk [11:17] sudo snap login works, with both credentials and a further prompt for 2fa [11:17] But even then, both snap download and sudo snap download still fail [11:20] hello [11:21] is there a "install" button widget or something I could add on my website to point to snapcraft.io/vlc ? [11:31] I found https://dashboard.snapcraft.io/snaps/git-ubuntu/revisions/429/ which has a download link which works. But that gives me only the snap, not the assertion. [11:33] rbasak: that's correct, it should be possible if you have developer access [11:34] rbasak: can you open a bug https://launchpad.net/snapstore/+bugs [11:38] pedronis: thanks. The dashboard link shows me " [11:38] Submitted by [11:38] nacc (nish.aravamudan@gmail.com) [11:38] " [11:39] Which isn't me. [11:39] Could that be the reason? [11:39] Some kind of team/individual confusion. [11:39] It should all be handed over to me, and this was done before that revision was pushed. [11:39] I remember having to re-auth Launchpad to the store. [11:39] rbasak: if you can push to it it shouldn't matter [11:42] PR snapd#5244 closed: tests: shellchecks part 3 === chihchun is now known as chihchun_afk [11:43] rbasak: what architecutre is the snap? [11:43] rbasak: if you are logged in, you can ask for any published revision, as long as it's for the current architecture [11:43] All amd64 [11:44] rbasak: 'snapcraft revisions git-ubuntu' agrees? [11:44] oh [11:44] wait [11:44] *snap download* [11:44] ah, download [11:45] me missed that [11:45] As opposed to what - publish? [11:45] rbasak: install [11:45] Ah [11:45] rbasak: download doesn't talk to snapd to get your tokens [11:46] Let me explain a little more what's going on [11:46] edge was working [11:46] I made an innocuous push to master in our codebase, which triggered a Launchpad build recipe to push to edge [11:47] That tripped some kind of non-determinism in snapcraft and pushed a broken snap to edge [11:47] rbasak: QOTM right there [11:47] I'd like to do two things: 1) restore edge; 2) examine the snap that Launchpad produced, which is broken, since that isn't what our CI did, which was OK [11:47] “I made an innocuous push to master in our codebase” :') [11:48] :) [11:48] rbasak: ok, so if edge is currently broken and that's the one you want to download, why not snap download it right now? [11:48] there's a bit of a dance to be done to get it to use the right tokens otherwise [11:48] https://forum.snapcraft.io/t/launchpad-post-build-pre-upload-testing/5545 has some broken [11:48] Chipaca: I've downloaded the broken one no problem. [11:48] ah ok [11:48] Chipaca: I'd like to download the previous good one too [11:49] I have managed to get the snap from the dashboard UI [11:49] rbasak: once you have the bad one, use snapcraft to re-pubish the good one, and download that one [11:49] But I was confused as to why "snap download" didn't work since I was under the impression that it would [11:49] Also that the dashboard web UI didn't give me the assertion file, but I presume I will just get a new one when I re-publish the good one? [11:50] sorry, I misunderstooed, download doesn't work in that case (unless you pass it tokens manually) [11:50] but that's known bug, not prioritized for a fix yet though [11:51] https://dashboard.snapcraft.io/snaps/git-ubuntu/revisions/429/ has a Release button [11:51] https://forum.snapcraft.io/t/improvements-in-snap-download/1422/8 [11:51] Can I stick it back in edge through there? [11:51] rbasak: yes [11:52] Great! I think I know how to do everything I need then. [11:52] Thank you for your help - it's much clearer to me now what is going on. [12:00] pstolowski|lunch, mborzecki: updated 5245 [12:00] and I have a working follow-up that uses it extensively [12:01] and drops all segments from the code [12:01] zyga: ok, looking [12:10] skomorokh: it isn't currently possible to do what you want. I suggest creating a forum topic at https://forum.snapcraft.io [12:13] pstolowski|lunch, mborzecki: I pushed the other PR to https://github.com/snapcore/snapd/pull/5254 [12:13] PR #5254: cmd/snap-update-ns: discard the concept of segments [12:13] PR snapd#5254 opened: cmd/snap-update-ns: discard the concept of segments [12:13] jdstrand: hey, good morning :) [12:14] hey zyga :) [12:14] jdstrand: I'm working on trespassing, went on a small tangent to clean up validation interface, ended up killing segments [12:16] cool [12:16] I have one last cleanup on top, to make error messages less duplicated, shorter and to the point [12:16] then the stack unwinds and validation is back, also with fuse validation [12:16] I didn't see fuse as a problem but now I do :/ [12:17] anything fuse has one filesystem type number [12:17] so we no longer know read-only squashfs from writable fuse-exfat or whatever [12:17] I'd love if there was a way to see if a filesystem is read-only without trying to write to it === pstolowski|lunch is now known as pstolowski === chihchun_afk is now known as chihchun [12:29] zyga: doesn't statfs's ST_RDONLY work? [12:30] mmmm :D [12:30] maybe, I didn't think about that ! [12:30] Thank you John, I'll check that [12:30] wouldn't be surprised if it didn't, because it being fuse means somebody would have to care :-) [12:34] PR snapd#5255 opened: cmd/snap-confine: applied make fmt [12:36] PR snapd#5256 opened: cmd/libsnap-confine-private: helper for extracting store snap name from local-name [12:45] mborzecki: can we not use the word base there [12:45] it's not about the code but the word would become overloaded and confusing [12:45] what's the terminology for the instances/parallel installation etc [12:46] haha ok ;) [12:47] if we had a list of forbidden words, i'd nominate classic [12:48] * zyga should take the dog for a walk [12:52] mborzecki: you should call it 'root', that way it's unambiguous [13:07] Chipaca: sc_snap_root_name sounds good [13:08] mborzecki: I was being facetious [13:09] Chipaca: heh, given the limited possibilities root actually make sense :) [13:27] mborzecki: sc_snap_uninstanciate_name [13:27] mborzecki: sc_snap_deinstance_name [13:28] mborzecki: sc_snap_get_rid_of_this_instance_nonsense_in_the_name [13:28] sc_snap_drop_instance_name [13:28] mborzecki: sc_snap_what_is_this_instance_stuff_just_i_dont_even [13:28] sc_snap_why_u_name [13:29] mborzecki: sc_snap_wat [13:29] mborzecki: putting my serious hat on for a moment, sc_snap_drop_instance_name might be the best one so far [13:31] pstolowski: can you look at the iterator PR again please [13:31] zyga: will do [13:31] feel free to open more points there, I changed the code much after really using it [13:33] niemeyer: $7.5B in stock [13:33] niemeyer: the PR I mentioned is here: https://github.com/snapcore/snapd/pull/5221 [13:33] PR #5221: [RFC] snap: parse connect instructions in gadget.yaml [13:33] zyga: How much is that divided by the number of developers? [13:33] Just trying to figure how much I'm worth :P [13:34] niemeyer: ~100 AFAIK [13:34] ah sorry [13:34] 745 exactly [13:34] man, that's even nice to divide :P [13:35] 10M each, even for the guy mending the coffee machine [13:41] pedronis: Looking [13:46] cachio: is gce cleaner using timer services? [13:46] mborzecki, yes [13:46] nice [13:46] mborzecki, :) [13:47] working very well so far [13:57] noticed the banner at the top of the github page? [14:02] mborzecki, yes [14:04] pedronis: Reviewed.. probably needs a quick sync, on that last question.. going into a call now, and we can sync after it === ErichEickmeyer_ is now known as ErichEickmeyer === chihchun is now known as chihchun_afk [14:16] PR snapd#5257 opened: snapstate: ensure fakestore returns TypeOS for the core snap [14:16] back from the walk [14:16] so so so hot [14:22] * zyga made some ice coffee and gets back to work === andyrock_ is now known as andyrock [14:25] @niemeyer eh [14:25] https://gitlab.com/Zyga [14:25] :-( [14:25] who is pawel z? [14:29] zyga: uh...new face of cybersquatting [14:38] * zyga enjoys some brief breeze of wind [14:52] niemeyer: oops, I got confused and thought your meeting was 1h [14:53] pedronis: It wasn't too far from it :) [14:54] PR snapd#5258 opened: tests: fix lxd test which hangs on restore [15:00] PR snapd#5217 closed: asserts,image: add support for models with bases [15:03] robert_ancell: o/ === chihchun_afk is now known as chihchun [15:23] pstolowski: updated [15:24] pstolowski: the error from the constructor fits the rest very nicely, I will now update the remaining PR [15:27] cachio: hey, can you take a look at https://github.com/snapcore/snapd/pull/5248? The only issue with travis is that the lxd fix hasn't made it to 2.33 [15:27] PR #5248: tests: add test to ensure /dev/input/event* for non-joysticks is denied - 2.33 [15:27] }jdsure [15:27] zyga: great, thanks [15:27] at least I think I'm remembering that right [15:27] cachio: thanks [15:47] mvo: let me know if I'm bikeshedding too much on your RFC PR [15:49] oops [15:50] PR snapd#5259 opened: devicestate: support seeding from a base snap instead of core [15:53] Chipaca: just read it, its great [15:53] Chipaca: thank you! [15:59] pstolowski: can you re-review that PR please, it would get me unblocked [16:04] niemeyer: I'm available for the next hour for that chat, otherwise tomorrow [16:10] zyga: doing [16:10] Thanks :) [16:12] zyga: looks good, although you made it strict in all cases now, that's intended? i thought you would have to variants? [16:12] *two [16:12] ah, cleanpath should succeed generally [16:14] pstolowski: no, I thought so too but the usage down the line is fine as-is [16:17] +1 [16:17] \o/ thanks! [16:18] let's hope tests are lucky now [16:26] PR snapcraft#2153 opened: project_loader: allow loading null parts === chiluk_ is now known as chiluk [16:32] pedronis: Hey, give me a couple and will be with you [16:33] PR snapd#5245 closed: cmd/snap-update-ns: add PathIterator [16:36] pedronis: https://meet.google.com/ftn-sqmv-kwy [16:37] pstolowski: follow-up in #5254 [16:37] PR #5254: cmd/snap-update-ns: discard the concept of segments [16:37] now shorter and just to the point [16:38] PR snapcraft#2151 closed: Debian [16:39] zyga: i'll take a look tomorrow morning [16:39] pstolowski: sure [16:41] pedronis, hey, any idea why it could be happening? https://paste.ubuntu.com/p/F8fwwrYHVt/ [16:41] it is not retrying the assertions === pstolowski is now known as pstolowski|afk [16:44] I'm having some problems understanding how to create a snap. I have a binary, a few depencencies, and a wrapper script. Nothing is being compiled. I'm not sure how to go about doing this. I'm looking at the site, but not having much luck understanding it. [16:44] jdstrand: hey, do you have some time for a small refactor review? [16:45] zyga: which PR? [16:45] Trel: see what snapcraft does, it allows you to bundle your binary (using the copy plugin AFAIK but kalikiana can correct me), you can also define your dependencies which will be added to your snap [16:45] Trel: with some iteration you should get it going [16:45] jdstrand: #5254 please [16:45] PR #5254: cmd/snap-update-ns: discard the concept of segments [16:45] jdstrand: this is related to trespassing, the segment verification was unable to verify rootfs and was super confusing with off-by-ones [16:45] zyga: I've been trying to use snapcraft the entire time. I'm not getting it [16:46] Trel: snapcraft defines how various parts should be combined (perhaps built, perhaps not) into a snap and what the snap itself should export [16:46] Trel: I'd suggest installing hello-world [16:46] looking at what the snap actually contains (in /snap/hello-world/current) [16:47] specifically focus on meta/snap.yaml [16:47] Trel: then run snap run --shell hello-world [16:47] and explore how the filesystem looks like, it's not the same as before [16:47] this should give you some feeling of how snaps execute [16:47] after that the question is "how to build one for myself" [16:47] there are many paths, snapcraft is the most complete and automated one [16:47] but for experimentation and learning you can hand-make one with a few commands [16:48] all you need is meta/snap.yaml [16:48] some files you want to have in your snap [16:48] then you can "snap try" from the directory containing the "meta" directory (this becomes the top level directory in your snap) [16:48] and see what happens [16:48] "snap try" is fantastic for iteration IMO [16:48] you can even get started by copying hello-world snap into a directory of your own [16:48] and tweaking some things [16:49] and "snap try" that [16:49] once you are more comfortable research snapcraft and see how you can take your binary and ship it as a snap [16:49] and people here will no doubt help you out with specific questions [16:49] See, I'm not following any of what you said. [16:49]