[00:24] maybe the Status update for version mails from the store should mention architecture? [05:00] robert_ancell_: ping [05:00] morphis_, hi [05:00] robert_ancell_: we're currently porting snapd-glib to Fedora and enabling support for it in gnome-software there too [05:01] morphis_, awesome, thanks! [05:01] robert_ancell_: however we saw something very interesting: http://imgur.com/a/CowmR [05:02] the default locale of the user is en_GB but he gets turkish messages; I figured out later yesterday that this is because policykit in Fedora doesn't have gettext support [05:02] oh [05:02] robert_ancell_: and found an old upstream bug you filed years ago https://bugs.freedesktop.org/show_bug.cgi?id=29639 [05:02] so policykit always falls back to the last entry in the .plolicy file [05:03] morphis_, ah, I probably forgot about that and should generate the PolicyKit file with the translations as per the spec [05:03] morphis_, I can fix that up tomorrow and make a 1.10 release [05:03] robert_ancell_: awesome! [05:03] morphis_, thanks for finding the problem! [05:04] robert_ancell_: so upstream polkit has gettext support but in a different way? [05:04] morphis_, I think it never got accepted upstream and I didn't notice because I'm testing on Ubuntu [05:04] robert_ancell_: np, Son_Goku and a gnome-software developer found it :-) [05:04] robert_ancell_: yeah.. :-) [05:05] robert_ancell_: as I am mainly focussing on cross-distro these days I am seeing a lot of these things [05:09] robert_ancell_: btw. would it make sense to move snapd-glib over to github.com/snapcore? [05:09] morphis_, possibly. I ended up putting on LP just to get going quickly. [05:10] morphis_, do you know who I'd have to ask to migrate there? [05:10] robert_ancell_: I think zyga or niemeyer can do that [05:11] robert_ancell_: if you don't overlap with them today I can talk with them or maybe better you create a forum topic on forum.snapcraft.io [05:11] morphis_, I'm past EOD, so I have to head off now. If you could raise it with them today that would be helpful. [05:12] bye === ricardokirkner is now known as pindonga === DedSec_ is now known as DedSec === andyrock_ is now known as andyrock === victorbjelkholm_ is now known as victorbjelkholm === eshlox_ is now known as eshlox === vigo is now known as vigo|afk [05:57] good morning [05:58] mvo: hey, could you please review https://github.com/snapcore/snapd/pull/3131 [05:58] PR snapd#3131: interfaces/mount: add OptsToFlags for converting arguments to syscall… [05:58] mvo: and perhaps https://github.com/snapcore/snapd/pull/3129 (just a struct) [05:58] PR snapd#3129: interfaces/mount: add InfoEntry type [05:58] zyga: sure [05:59] PR snapd#3039 closed: many: add support for partially static builds [05:59] thanks :) [05:59] mvo: ^^ I just landed partially static builds, let me know if something starts to misbehave [06:00] ok [06:01] PR snapd#3125 closed: tests: download and install additional dependencies when using prepackaged snapd === vigo|afk is now known as vigo [06:06] oho, a bug in regression test [06:06] /bin/bash: line 56: [: missing `]' [06:08] PR snapd#3134 opened: tests: fix incorrect shell expression [06:08] mvo: https://github.com/snapcore/snapd/pull/3134 this will fix some autopkgtest failures [06:08] PR snapd#3134: tests: fix incorrect shell expression [06:09] PR snapd#3106 closed: tests: enable docker test for more ubuntu-core systems [06:14] morphis_: hey, I wanted to get this on your radar https://bugzilla.novell.com/show_bug.cgi?id=1028568 [06:14] morphis_: I'll open a forum topic to discuss this [06:15] zyga: thanks, I'm going over all the one ones now [06:17] zyga: keen to see the tests, autopkgtest should be fixed [06:17] mvo: I'll merge master into https://github.com/snapcore/snapd/pull/3085 once the fix above lands and everything (maybe) goes green [06:17] PR snapd#3085: .travis.yml: remove travis matrix and do a single sequential run [06:21] PR snapd#3112 closed: interfaces: add a joystick interface [06:21] ha! #3115 has all tests green again, so nice to see that autopkgtest is now fixed again [06:22] zyga: hey! yeah I saw already that lxd is broken both on Fedora and openSUSE, but good that we have a dedicated bug for that [06:32] morphis_: I did a quick branch to fix that but we need coordination from ogra and the core snap [06:32] mvo: if you have a min, another review on https://github.com/snapcore/snapd/pull/3096 and a merge later today would be great! [06:32] PR snapd#3096: many: abstract path to /bin/{true,false} [06:32] morphis_: my plan was to essentially stop sharing /etc [06:33] zyga: is that easily possible? [06:33] except for hostname and resolv.conf [06:33] morphis_: yes [06:33] aww, brb [06:33] we may have a few more dependencies on it don't we? [06:33] like /etc/netplan [06:34] zyga: and grep'ing through interfaces/builtin shows a lot more [06:34] at least 12 which have rw permissions [06:34] zyga: https://paste.ubuntu.com/24311669/ [06:36] morphis_: sure, happy to [06:37] mvo: thanks [06:37] morphis_: looks great! thanks for your patience [06:37] mvo: np [06:38] morphis_: re :) [06:38] morphis_: so that will all work, but we need to tweak one symlink [06:38] zyga: you saw the list I've pasted above? [06:38] morphis_: as if we stop sharing all of /etc one of the writable things has to be adjusted [06:39] morphis_: no [06:39] https://paste.ubuntu.com/24311669/ [06:39] morphis_: sorry, I had to do IRL stuff [06:39] IRL? [06:39] morphis_: in-real-life [06:40] ah :-) [06:40] morphis_: not sure what I'm looking at [06:40] zyga: that is a grep through interfaces/builtin for things which refer to /etc [06:40] there are quite a few writable things in /etc [06:40] if we stop sharing those with the host things will break [06:40] morphis_: and they work with /writable and symlinks and such AFAIR [06:40] zyga: what is with classic? [06:41] morphis_: the whole point is that it already works in core/all-snap [06:41] zyga: ok, then I may don't get yet what you're planing to do [06:43] we have symlinks in /etc in the core snap, I get that, but wont those break too if we have the same core snap on multiple distributions as they would point to /writable which doesn't exist there [06:44] so if we stop sharing /etc with the host it looks to my like we have to repeat all the writable-paths handling the initramfs on Ubuntu Core currently does for classic [06:44] morphis_: 3084 has some more suggestions, but I must say I think its looking super nice, the extra bit in there will just make it even more nice :) great work there! [06:44] * mvo hugs zyga for the review too [06:46] mvo, zyga: will change that in a bit [06:46] zyga: 3096 needs a second review, if you have a moment, should be trivial then it can land [06:46] * zyga will review/read stuff in a sec [06:46] just sending kids to school [06:52] zyga: thank, no real rush [06:54] ok :) [06:54] all gone now [06:54] morphis_: of coruse writable would exist [06:54] morphis_: it is in the core snap after all [06:54] morphis_: and we look at the world _after_ the pivot_root is done [06:55] morphis_: so it is just a matter of putting the right host data files to /writable [06:55] morphis_: and setting everything up so that after pivot_root it's all resolving correctly [06:55] morphis_: writable would have to be a tmpfs [06:55] morphis_: and would need to be managed by either snapd or by snap-confine [06:56] mvo: I was already looking at 3096 :-) [06:57] PR snapd#3122 closed: packaging: do not compile spread for autopkgtests [07:00] zyga: right, and we need to bind mount things there from the real /etc [07:00] otherwise snaps wont be able to change timezone/locale/... [07:03] PR snapd#3131 closed: interfaces/mount: add OptsToFlags for converting arguments to syscall… [07:05] morphis_: interesting [07:05] morphis_: yes, perhaps [07:06] mvo: looks like adt will be green now :) https://github.com/snapcore/snapd/pull/3134 [07:06] PR snapd#3134: tests: fix incorrect shell expression [07:07] zyga: yeah, I like the level of greeness [07:08] once this passes I'll merge it and start merging it into pending brnaches [07:08] mvo: I'll start with those for 2.24 [07:08] oh, no more 2.24 :) [07:17] just one test yellow :) [07:17] zyga: but lets explore this a bit more, worth a forum topic :-) [07:20] morphis_: what? [07:20] all green, merging! [07:21] zyga: the unshared /etc on classic [07:21] morphis_: aha, yes, definitely [07:21] PR snapd#3134 closed: tests: fix incorrect shell expression [07:21] eveyone, let's not land failing tests now [07:21] everything should be green [07:23] PR snapd#3132 closed: overlord/state: make sure that setting to nil a state key is equivalent to deleting it [07:33] pstolowski: hey [07:33] zyga, o/ [07:34] pstolowski: I just resolved conflicts on https://github.com/snapcore/snapd/pull/3119 and I worry that it is too easy to add a wrong version of AddConnected{Plug,Slot} [07:34] PR snapd#3119: interfaces: API additions for interface hooks [07:34] pstolowski: I'd like to do a sanity test that looks at each interface and makes sure it doesn't implement the old APIs [07:34] pstolowski: no snippets, no attr-less connected plugs or slots [07:35] pstolowski: with the definer hack we essentially turned to duck typing on a security component [07:35] pstolowski: and this is a bit risky :/ [07:37] zyga, thanks for resolving conflicts in that branch; [07:38] fgimenez: conflicts on https://github.com/snapcore/snapd/pull/3105 [07:38] PR snapd#3105: tests: download previous snapd package from published versions instead of specific PPA [07:39] zyga: thanks on it [07:40] zyga, as for duck typing yes, I realized these risks too. I think from now only every interface needs to have full tests [07:40] zyga, good idea about sanity test, a simple shell script will do [07:41] pstolowski: I was thinking about a go based test actually, let me try something [07:41] everyone: I'm going through PRs and merging master into them, this should fix adt failurs [07:41] *failures [07:42] please don't merge things if they are not green anymore [07:45] hmm [07:45] FAIL: overlord_test.go:360: overlordSuite.TestEnsureLoopPrune [07:45] this failed on my laptop just now [07:45] was that known racy? [07:56] pstolowski: we need to plan a 2nd pass through all PRs to adjust them to the new interface APIs [07:56] pstolowski: but I'd like to see this new test merged first [07:56] pstolowski: I'll take a break, prepare for some meetings today and then start working on it [07:57] pstolowski: if you could do a trivial review on a struct: https://github.com/snapcore/snapd/pull/3129 [07:57] PR snapd#3129: interfaces/mount: add InfoEntry type [07:57] pstolowski: that would help me move update-ns effort forward :) [08:01] PR snapd#3135 opened: interfaces/mount: add high-level Profile functions [08:04] PR snapd#3108 closed: cmd: use libtool for the internal library [08:12] mvo: I'm done, nearly everyting is either yellow now or has been commented upon to have the author do something [08:12] mvo: probably we'll run out of machines but spread can be restarted [08:12] mvo: and adt will nicely queue [08:15] * zyga afk [08:24] PR snapd#2971 closed: data: ship "snap.mount" service that ensures /snap is MS_SHARED [08:28] zyga: can you have another look on https://github.com/snapcore/snapd/pull/3084 ? [08:28] PR snapd#3084: packaging: use templates for relevant systemd units [08:28] morphis_: yes, after my calls [08:28] thanks! [08:51] fgimenez: - autopkgtest:ubuntu-16.04-ppc64el:tests/main/classic-custom-device-reg failed on https://github.com/snapcore/snapd/pull/3129 [08:51] PR snapd#3129: interfaces/mount: add InfoEntry type [08:52] fgimenez: died on kill timeout there [08:52] mvo: it would be awesome if we had a way to re-try adt tests [08:53] we're starting to see green tests :) [08:53] not all but at least some [08:53] random failures on adt are annoying as they are harder to re-try [08:54] zyga: yeah, I have no idea how to retrigger those, we need help from the adt people on that. but its worth investigating the failures I think [08:54] mvo: if it is just slower infrastructure we should tweak knobs to kill them later [08:54] mvo: if it is real failure we want to investigate [08:54] mvo: so yeah, agreed! [08:57] aha, that looks like a real bug [08:57] [ 1602.570508] audit: type=1400 audit(1491294017.372:1090): apparmor="DENIED" operation="create" profile="snap.classic-gadget.hook.prepare-device" pid=27121 comm="snapctl" family="inet" sock_type="stream" protocol=6 requested_mask="create" denied_mask="create" [08:57] snapctl + ipv6 [08:58] mvo: did your branch with config that was addressing that landed? [08:58] s/landed/land/ [09:02] I added https://github.com/snapcore/snapd/pull/3129/commits/f82c78c07facd73f5ad1a31f26b0b337dc6ce4ce to try to figure out what is failing on ppc64 [09:02] PR snapd#3129: interfaces/mount: add InfoEntry type [09:05] zyga, i dont see a PR for the lxd bit [09:17] hi mvo: i'm looking into the expect errors on ubuntu-core http://paste.ubuntu.com/24305344/ and, trying manually tests/main/create-key on amd64 with edge core, it gets stuck after the "Confirm password" prompt http://paste.ubuntu.com/24312255/ if you could take a look when you have a moment that would be great [09:18] zyga: so for gnome-software we have a hardcoded /snap/bin in snapd-glib we need to workaround [09:18] err, a hardcoded /snap/bin in gnome-software [09:37] morphis_: aha [09:37] morphis_: we can do a small patch or we could maybe have it ask snapd [09:37] morphis_: but I think we can do it easily with a small patch [09:38] ogra: hmm [09:38] zyga: yeah that is the plan [09:39] zyga: talked with Robert this morning and he wants to do a minor release of snapd-glib tomorrow anyway so I am asking him now if he can add a small API function returning the snap mount dir [09:39] we can than set it statically via a configure switch and later ask snapd [09:43] Good mornings [09:45] hey niemeyer! good morning [09:47] niemeyer: morning! [09:47] o/ [09:47] morphis_: sounds good! [09:47] niemeyer: good morning :) [09:48] zyga: you already had time to give snapd on Fedora a try or is that something for later this week? [09:48] pstolowski: error: cannot refresh "test-snapd-delta-refresh": cannot get refresh information for snap "test-snapd-delta-refresh": Post https://search.apps.ubuntu.com/api/v1/snaps/metadata: EOF [09:49] morphis_: I'll try that today, I have my 2nd machine standing by [09:49] pstolowski: this failed here: https://github.com/snapcore/snapd/pull/3111 [09:49] PR snapd#3111: snapd: initial implementation for systemd software watchdog for snapd [09:49] pstolowski: will that be fixed with https://github.com/snapcore/snapd/pull/3126 ? [09:49] PR snapd#3126: store: handle EOF via url.Error check [09:50] zyga: sounds great! we need 6 points to get the update done, so need to collect people :-) [09:52] morphis_: I'll boot F25 first [09:52] ok [09:53] fgimenez: tests/main/interfaces-network-observe failed on timeout (seems like test was stuck) [09:53] https://objectstorage.prodstack4-5.canonical.com/v1/AUTH_77e2ada1e7a84929a74ba3b87153c0ac/autopkgtest-yakkety-snappy-dev-image/yakkety/amd64/s/snapd/20170404_092050_bb1cd@/log.gz [09:53] zyga: thanks looking.. [09:55] fgimenez: your new dbus interface spread test seems to be geuinely failing https://github.com/snapcore/snapd/pull/3014 [09:55] PR snapd#3014: tests: add dbus interface spread test [09:57] had to reboot because my tests leaked out and borked the environ, and spread realised and restarted qemu without me having to meddle [09:57] * Chipaca hugs niemeyer [09:57] koza, next customer ... Bug #1679432 ... [09:57] Bug #1679432: Bluetooth SCO connections don't work on Dragonboard 410c [09:57] Chipaca: good morning :) [09:57] hmm, I really wish we could retry adt tests [09:58] error: RPC failed; curl 56 GnuTLS recv error (-110): The TLS connection was non-properly terminated. [09:58] * niemeyer feels the love and hugs Chipaca back [09:58] random error on git clone [09:58] zyga, niemeyer :-) [09:58] zyga, yes, i hope so.. [10:00] zyga: indeed, the test snaps weren't being built for ppc64el, doing that now [10:01] zyga: and thanks! :) [10:01] pstolowski: let's merge it then! [10:02] fgimenez: thank you! [10:02] fgimenez: I think we have a chance at all-green tests today [10:02] fgimenez: I'll do my best to help [10:02] zyga, i've just pushed the little fix for unneeded var error; [10:04] zyga: \o/ thanks a lot, spring is definitely coming to our poor test results :) [10:06] * zyga reviews 3126 [10:07] PR snapcraft#1230 opened: Refactor Cleanbuilder into Containerbuild and add Project [10:11] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1672819/comments/10 is exciting [10:11] Bug #1672819: exec'ing a setuid binary from a threaded program sometimes fails to setuid

[10:11] which reminds me [10:12] zyga— do you have a fedora system/vm up? [10:15] Guys I noticed that telegram puts downloads in /snap/.../Downloads/ instead of ~/Downloads. I know it's for the sake of confinement, but wouldn't it be a better idea to also mount this folder into ~/Downloads? [10:16] that way, the snap doesn't see other snaps downloads but I can go to ~/Downloads to get my stuff [10:16] /snap/Downloads is a long path [10:18] Chipaca: yes [10:18] zyga— can you check whether that bug happens in fedora also? (it should, but maybe they patch their kernel for this already) [10:18] brunch875— it's tricky [10:19] brunch875— ~/Downloads isn't guaranteed; in fact, it's localised [10:19] Chipaca: sure [10:19] brunch875— if you're in spanish I think it's ~/Descargas [10:19] Chipaca: we have $XDG_CONFIG_DIR/user-dirs.dirs [10:19] damn translations... how does firefox deal with this? [10:20] zyga— exactly, but people editing that is rarer than people speaking something different [10:20] not saying it's impossible; it's tricky [10:20] Chipaca: yeah [10:20] Chipaca: just sligthly trickier than fixed path [10:20] Chipaca: it's tricky if it changes underneath [10:20] Chipaca, i think we stopped localizing it ... your original install must be old [10:20] it's three indirections to get the thing [10:20] but that is rare I hope [10:20] yep [10:21] is it bad that i made tests pass by removing the failing ones :-D [10:21] Chipaca: I bet there's a bible rerefence that fits this but I don't think we want that ;) [10:22] zyga— “And the beast shall come forth surrounded by a roiling cloud of vengeance. The house of the unbelievers shall be razed and they shall be scorched to the earth. Their tags shall blink until the end of days.” (here "beast" could be "snapd") [10:24] ogra, haha thanks, reading now [10:24] Chipaca: the church of snapd [10:24] Chipaca: apply for tax deductions [10:25] in civilization i always name my religion "worms" [10:25] because i spread worms to other civilization and that makes me chuckle [10:25] but "the church of snapd" could work also [10:27] reviews for https://github.com/snapcore/snapd/pulls?q=is%3Aopen+is%3Apr+milestone%3A2.24 would be great - so that we can get 2.24 ready [10:28] Chipaca: thu shal not have other packages before me ;-) [10:28] thu shall refresh on weekends ;) [10:28] * zyga gets back to being useful [10:30] PR snapd#3136 opened: snap-confine: add code o ensure that / or /snap is mounted "shared" [10:32] mvo: I'm about to jump into it shortly.. just want to get a discourse snap with SSO support building meanwhile [10:33] Chipaca: So it was indeed a bug in the kernel.. [10:35] niemeyer: thank you! [10:37] Chipaca: fedora booting, sorry, vmware decided it's time for upgrade [10:39] fgimenez: DEBUG: restarting into "/snap/core/current/usr/bin/snap" [10:40] fgimenez: this seems to clober and affect tests on https://github.com/snapcore/snapd/pull/3010 [10:40] PR snapd#3010: snap: skip /dev/ram from auto-import assertions to make it less noisy [10:40] fgimenez: I wonder why it only happens here. it looks like either we reexec where we previously did not [10:40] fgimenez: or we log where we previously did not [10:40] fgimenez: I have no other ideas [10:41] oddly [10:41] fgimenez: same spread run has this message: [10:41] 2017/04/04 09:49:49.300407 cmd.go:114: DEBUG: not restarting into "/snap/core/current/usr/bin/snap" ([VERSION=2.23.6 2.23.6]): older than "/usr/bin/snap" (1337.2.23.6) [10:41] so it restarted once [10:41] but not another time [10:41] during one run [10:41] mvo: does this make any sense to you? [10:42] it seems that all the failures there are caused by the extra DEBUG message [10:45] mvo: aha, so that branch sets + SNAPD_DEBUG: 1 [10:45] mvo: I think we need something more flexible [10:45] zyga: indeed - maybe going back to snappy_testing? [10:46] mvo: I commented on the PR [10:46] mvo: snappy_testing? what is that [10:47] zyga: its a flag we set when spread runs, it tells that the system is being tested [10:47] zyga: let me try that [10:47] * mvo really lunch now [11:05] Chipaca: ok, updated my f25 to latest kernel and trying the suid bug now [11:06] fgimenez: I have a question about https://github.com/snapcore/snapd/pull/3085/files [11:06] PR snapd#3085: .travis.yml: remove travis matrix and do a single sequential run [11:07] fgimenez: what happened to the prepare_all_snap vs prepare_classic code? [11:07] aha I see you excluded core [11:08] fgimenez: so do we run any tests on core with that PR now? [11:10] zyga: of course, the core system filtering is a refactor only for the unit suite, previously the tasks in that suite were excluding the core systems, with this PR the exclusion is done at the suite level [11:11] aha, makes sense [11:11] thanks! [11:11] let's merge it when it goes green [11:12] zyga: np thank you [11:15] Chipaca: yep, it also affects fedora [11:15] Chipaca: but curiously the go version is ok [11:15] Chipaca: but the c+pthread version is not [11:16] Chipaca: maybe fedora has different golang build defaults? [11:36] zyga: https://github.com/snapcore/snapd/pull/3096 has some strange commits, did you maybe push the wrong branc there? [11:36] PR snapd#3096: many: abstract path to /bin/{true,false} [11:39] zyga— the go version might be ok if you have only one (real) cpu on your vm [11:40] mvo: checking [11:41] oh, I must have [11:41] sorry, [11:41] shall I push --force to remove them? [11:41]