=== JoshStrobl is now known as JoshStrobl|AFK [01:31] any around around by chance? [01:31] anyone* === JanC is now known as Guest55020 === JanC_ is now known as JanC [04:22] PR snapd#3806 opened: packaging/fedora: Merge changes from Fedora Dist-Git === Justin` is now known as Justin`afk [05:53] PR snapd#3693 closed: snapstate: improve the error message when classic confinement is not supported [07:04] PR snapd#3806 closed: packaging/fedora: Merge changes from Fedora Dist-Git [07:11] ++ dnf -q -y --refresh install 'pkgconfig(systemd)' [07:11] Error: Failed to synchronize cache for repo 'updates' [07:33] kyrofa: re [07:33] kyrofa: not sure about bug tracking, I think keeping it all in snapd is easier though [07:33] good morning everyone :-) [07:39] zyga-suse_: hey, I think I addressed your points in 3260, let me know. I will ponder a bit more over the cmd_userd_test.go in the meantime [07:39] hey [07:40] thank you, let me look [07:40] I'm working on a blanket in the grass [07:40] I wonder if people will wonder what I do here ^_^ [07:49] mvo: poor network connection, are you reading this? [07:50] zyga-suse_: yes [08:01] odd, I can chat on IRC but I cannot load any website [08:04] loaded [08:09] mvo: replied on the Ping comment [08:09] mvo: that's the only part I really cared about [08:09] let me read the diff [08:13] mvo: reviewed [08:15] mvo: I'm sorry about being strict about that ping method there, let me know if that works in the end [08:18] zyga-suse_: we are not implementing Ping() currently [08:18] zyga-suse_: we are not implementing the Dbus.Peer interface at all [08:18] mvo: go dbus doesn't do it for free? [08:18] it's typically implemented by libraries [08:18] zyga-suse_: unfortunately not [08:18] aww, bummer [08:18] that's not good [08:18] zyga-suse_: at least not as far as I can tell [08:18] zyga-suse_: d-feet is what i use [08:18] does it show up in introspection? [08:18] aha [08:18] zyga-suse_: yes, introspection shows up [08:19] I mean, does Peer show up? [08:19] zyga-suse_: no, peer does not show up at all, it does show up for the old safe launcher though (because dbus-gio) [08:19] right [08:20] good morning all! happy "all versions of go shipped in ubuntu are obsolete" day! [08:20] (1.9 is out) [08:20] oh, nice [08:20] anything shiny? [08:20] are generics in yet? [08:21] zyga-suse_: type aliases are a thing [08:21] zyga-suse_: and monotonic time [08:21] whee [08:21] so 2001 of us [08:21] zyga-suse_: I look into it, maybe we are just holding^Wusing it wrong [08:22] mvo: maybe we could just implement Peer [08:22] it's no-op Ping [08:22] and get machine id [08:22] zyga-suse_: the nicest change that would impact my work is that ./... no longer looks in vendor [08:23] yes, I saw that one [08:23] that's a nice change [08:23] so go test ./... would finally dtrt [08:23] zyga-suse_: yeah, otoh, if we can call OpenURL and get a reply like "cannot use ping-ping-ping:" we know the interface is up as well. but yeah, ping is nicer, I have a look [08:24] Chipaca: I'm still holding out for generics [08:24] mvo: agreed [08:24] mvo: if you want +1, just move to libexecdir [08:24] mvo: I may look at Ping over weekend [08:25] zyga-suse_: we are actually using the normal "snap" command for the userd subcommand, so it will be the regular bindir (unless I miss something) [08:25] mvo: ahh, sorry [08:25] I must have misread the diff [08:25] zyga-suse_: no worries [08:25] I assumed it would be a new binary [08:25] ok, I have two tests to fix [08:26] and a lot of goodies will be ready [08:26] but first [08:26] coffee!!! [08:27] zyga-suse_: heh, fun! if I export the ping interface things work [08:27] zyga-suse_: oh well, I will update things [08:30] zyga-suse_: did you see the guy that had generics implemented via unicode and generators? [08:34] mvo: there was no joy with the fedora tests btw, still failed in prepare in strange ways, are we tuning ip6 on fedora as well? [08:35] mvo: things work? [08:35] Chipaca: I think I don't actually want to know anymore :D [08:35] pedronis: probably not, it's behind ubuntu spread test flag [08:37] anyway something is off and makes those even more unstable than usual [08:37] zyga-suse_: bottom line: it looked like “type ImmutableTreeListᐸElementTᐳ struct {” [08:37] or the fedora infra itself is flakier, no clue [08:37] pedronis: I have not looked at the details for fedora unfortunately [08:37] zyga-suse_: yeah, ping works once the interface is exported via xml, no code needed, I will update in a bit (working on tests right now) [08:38] woot! [08:38] nice :) [08:38] ah [08:38] right [08:38] because introspection is how dfeet knows about it [08:38] though odd that this is not done by gdbus itself [08:40] PR snapd#2837 closed: interfaces/apparmor: allow reading from ecryptfs [08:44] thank you :'-( [08:44] though we _may_ have a way to solve that with what I'm working on [08:45] PR snapd#3346 closed: [WIP] allow access to nm-dispatcher scripts [08:50] hmm, i've got to test stuff that involves rebooting. I'll be pestering you with in/out messages soon :-) [08:51] Chipaca: VMs/ [08:52] zyga-suse_: i'm too close to my memory limit :-( [08:52] (this is a test that needs a full desktop) [08:52] brb [08:53] aww [08:53] * zyga-suse_ has apparmor on opensuse :) [08:53] and we can now probably also enable reexec [08:53] there's still a few things missing but I think it's very very closenow [08:54] mvo: is the PR ban still in effect? [08:59] morphis_: zyga-suse_: so it looks like moving /etc/X11/Xsession.d/65snappy to e.g. /etc/profile.d/snappy.sh will DTRT in both X and Wayland (might need some tweaks, but basically it'll work) [09:00] morphis_: zyga-suse_: I know RH/fedora are slightly different wrt profile.d, but afaik it boils down to it getting loaded more, not less; can you confirm? [09:00] Chipaca: if you test it, +1 [09:00] Chipaca: I think we want to dedupe those files [09:00] this is me looking at snapd#3398 [09:00] PR snapd#3398: env: set XDG_DATA_DIRS for wayland et.al [09:00] Chipaca: there's one in every distro packaging [09:00] Chipaca: and then we want to call it sanely, say snapd.sh [09:01] Chipaca: and keep it in data/ [09:01] zyga-suse_: i tested it on ubuntu, but as i said i don't have the space to test it on other things [09:01] Chipaca: I see [09:01] bah. i might be able to get this fedora running [09:01] should work with 1G, right? [09:01] Chipaca: I can test on all of them but I'd prefer to do that in the afternoon with less sharp mind [09:01] yeah no probs [09:01] Chipaca: nah [09:01] try 2G [09:01] sigh, ok [09:01] how much ram do you have? [09:01] lots of swapping ahead for me :-) [09:01] I have a _physical_ fedora box with 2GB and it's a pain [09:02] I added 1GB stick to make it work [09:02] zyga-suse_: free right now? 1G [09:02] OMG [09:02] and in general? [09:02] zyga-suse_: 4G total [09:02] darn [09:02] no way to expand? [09:02] i had 8 but one of the sticks died [09:02] and the whole box is starting to die [09:02] so i'm just letting it [09:02] aha [09:02] thinkpad? [09:02] while saving up for its replacement [09:02] man, buy t430 for the price of a good meal [09:02] zyga-suse_: latitude e series [09:02] they are indestructible [09:03] and cost peanuts [09:03] it's ~8 years old [09:03] if i were to buy the same thing, it'd cost peanuts too [09:03] what are you saving for? [09:03] but these old cpus suck 45W when angry, i'll gladly pay to worry about that less [09:03] so what if you can get a battery with 11hrs of life? [09:04] zyga-suse_: probably a latitude e series :-) [09:04] and still buy three for the price of one new box [09:04] zyga-suse_: 15 [09:04] zyga-suse_: this box had 15 hours battery life when it and the batteries were new [09:04] * zyga-suse_ chooses not to buy new laptops again [09:04] zyga-suse_: ah, it'd not be new, it'd be refurb [09:04] ah [09:04] +100 on that [09:04] it's worse than with used cars [09:05] i can't in good conscience justify the price nor environmental cost of new [09:05] you open the door and 30% is gone [09:05] yep [09:05] especially when new CPUs suck so badly lately [09:05] with less and less reliablity and features [09:06] I wonder when amd will show their mobile ryzen/threadripper units [09:08] * zyga-suse_ sees ants crawling over his keys [09:08] not ogod [09:09] they probably just want to help [09:10] * zyga-suse_ plays bug squashing [09:11] Chipaca: buy this and run !ubuntu distro on it http://www.ebay.es/itm/Lenovo-ThinkPad-T530-i5-3210M-2x2-5GHz-8GB-128GB-SSD-UMTS-NVS5400-Webcam-WIN10-/253112700548?hash=item3aeeb14e84:g:KsAAAOSw01dZnrBX [09:11] Chipaca: and work on it 2/5 days a week [09:11] snapd must feel great everywhere [09:12] well, if you want low-end, get a beaglebone :P [09:12] ogra_: ahem [09:12] ogra_: that's way more than sufficient for work [09:12] ogra_: and it has twice the RAM that Chipaca has now [09:13] it would be good if we had someone running centos [09:13] or if feeling less bold, just run opensuse or fedora where there's very good support already [09:13] ogra_: what are you using daily? [09:13] (h/w wise) [09:14] well, depends where i sit [09:15] in my office i have a "Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz" with 32G, 3TB raid, 500G SSD, triple head nvidia 970 [09:15] (+3x 22" LG HD monitors ) [09:15] in my living room my first gen XPS13 [09:16] (some early i7, i forgot which and 8GB) [09:16] zyga-suse_: http://www.morgancomputers.co.uk/c/512/IBM-Lenovo/ [09:16] nice desktop setup! [09:17] OTOH i coould set up my bbb [09:17] i'd need bluetooth and wifi to be working on it though [09:17] punch-card speed IO [09:17] hmm, then a pi3 rather ... [09:17] i thought io on the pi3 was slower than the bbb [09:17] (though thats pretty powerful for an arm board) [09:18] no, nothing is slower than bbb [09:18] well, pi3 is quad core 1.2GHz ... [09:18] Chipaca: pi3 when using HDD is very much usable [09:18] bbb is single 800MHz [09:18] Chipaca: just get that fancy Pi drive and boot away [09:18] no SD cards needed [09:18] as long as you dont use any other USB devices on it, yeah [09:18] (or any HDD with a usb cable) [09:19] it's good enough for coding IMO [09:19] wired anetwork and HDD at the same time can already get nasty [09:19] the USB hub gets saturated easily [09:20] it's limited but for the price it's really good [09:21] well ... for the price, yeah [09:32] mvo: trivia PR pre-reviewed by jdstrand https://github.com/snapcore/snapd/pull/3807 [09:32] PR snapd#3807: cmd/snap-confine,packaging: import snapd-generated policy [09:32] I stated doing internal reviews with jamie so that we can start with a +1 and land things faster [09:33] PR snapd#3807 opened: cmd/snap-confine,packaging: import snapd-generated policy [09:33] hmm ... [09:34] PR snapd#3791 closed: cmd/snap-confine: allow using additional libraries required by openSUSE [09:34] PR snapd#3792 closed: cmd/snap-confine: allow running snap-exec without confinement [09:34] why is my daily edge image installing core 2466 on first boot ? [09:34] 2466 feels 1000 builds too old [09:34] yes [09:35] especially for an edge build [09:35] 2739 is the current core in armhf edge [09:36] this smells like firstboot didnt set up the system properly [09:36] :/ [09:36] what's in /var/lib/snapd/snaps ? [09:38] ogra@pi2:~$ snap list [09:38] No snaps are installed yet. Try "snap install hello-world". [09:38] ogra@pi2:~$ [09:38] grmpf [09:38] ogra@pi2:~$ ls /var/lib/snapd/snaps/ [09:38] core_2466.snap core_2739.snap pi2-kernel_39.snap [09:38] ogra@pi2:~$ [09:38] no gadget [09:39] * ogra_ starts over with a fresh flash to capture the log of the first boto [09:39] *boot [09:41] Chipaca: did you see the new "dep" tool? [09:41] ogra_: super odd [09:41] zyga-suse_: in go? [09:41] ogra_: how did youget 2466? from ubuntu-image? [09:41] Chipaca: yes [09:41] zyga-suse_, i didnt .. [09:42] zyga-suse_, core_2739.snap [09:42] sure but 2466 is present there as well [09:42] u-image did the right thing ... but the initial setup failed ... i usually install avahi as first thing after setting the hostname when testing ... that pulled core from stable [09:43] aha [09:46] and this time it takes actually longer to get to "please press enter" ... last round this casme up immediately which indicates the key generation didnt take place [09:47] ogra@localhost:~$ snap list [09:47] Name Version Rev Developer Notes [09:47] core 16-2.27.4+git331.b672daf 2739 canonical core [09:47] pi2 16.04-0.18 39 canonical gadget [09:47] pi2-kernel 4.4.0.1070.70 39 canonical kernel [09:47] ogra@localhost:~$ [09:47] hrm ... so there was a race in the first run ... which is gone now :/ [09:51] i'm always surprised how different the memory usage between the different arm boards is [09:52] on the nanopi-air htop shows 48.7MB used ... the pi2 with the same set of snaps installed and running uses 70.3MB [09:53] PR snapd#3808 opened: apparmor,release: add better apparmor detection/mocking code [09:53] mvo: another small branch pre-approved with jdstrand [09:53] (both armhf, both using the same core ) [09:53] zyga-suse_: ok, looking [09:53] mvo: I have a tiny branch on top that will let debian and suse to enable apparmor [09:53] mvo: and that will let us enable reexec on suse :) [09:54] mvo: the last holdout of reexec will be distro layout (fixed with layouts) and /snap location (not covered by layouts yet) [09:56] fgimenez, ppisati, any idea why all our stable kernels are like 6 revisions behind ? [10:02] * zyga-suse_ -> small break [10:02] everyone: please request my reviews on things you want me to look at [10:03] ogra_: nope, i don't know [10:26] mvo: so the problem test seems a quirk of task without undo handlers, their change is ready "sooner", don't think it should block your branch, but I would like to explore not needed Overlord.Loop there [10:37] pedronis: aha, I see! I can add an undo handler for now as a workaround until we have time to explore further. thanks a lot for looking into this [10:49] mvo: if what's now /etc/profile.d/apps-bin-path.sh suddenly becomes /etc/profile.d/snapd.sh, does apt/dpkg/something freak out? [10:53] PR snapd#3809 opened: tests: copy files with less verbosity [10:53] mvo: trivial for my modem woes ^ [10:53] Chipaca: yes [10:53] you need to rm conffile [10:58] Chipaca: or in other words, you want to ship your computer to the arctic, move north and herd sheep [10:59] zyga-suse_: the name just strikes me as odd, especially now that it'll have XDG stuff in it :-) [10:59] but they are technically both paths [10:59] so ¯\_(ツ)_/¯ [10:59] Chipaca: yeah, the name is bad [10:59] Chipaca: I'd like to have just "snapd.sh", period [10:59] that's what i was proposing above [10:59] Chipaca: new systemd (we cannot assume it though) has nicer way to do this [11:00] zyga-suse_: it does? [11:00] one sec [11:00] anyway, time for me to go run [11:00] * Chipaca waits [11:00] let me pull the discussion with systemd devs [11:00] PR snapd#3810 opened: interfaces/hooks: PlugData and SlotData wrappers [11:00] from my blog ... [11:01] https://new.zygoon.pl/post/case-study-snapd-on-centos/ [11:01] Chipaca: have a look there [11:02] there are links to a PR that has since landed in systemd: https://github.com/systemd/systemd/pull/5131 [11:02] PR systemd/systemd#5131: Environment generators [11:03] and while looking there, enjoy 102 open PRs for systemd [11:03] (we are not doing bad) [11:03] well [11:03] though 806 contributors is x10 more than we have [11:03] from our downstram POV thats not acgtually much different [11:04] (environment.d vs profile.d ... ) [11:04] you still need to ship the snippet that assembles the variable [11:04] just in another place [11:04] environment.d happens earlier than profile.d [11:04] sure sure [11:04] though I hate putting things in either, really [11:05] just saying that iit doesnt matter much regarding the naming of the fil ;) [11:05] *fiile [11:05] sigh ... [11:05] that's okay, one day you'll get it :) [11:05] my laptop keyboard slowly gives up [11:06] zyga-suse_, we'll probably not have anywhere close to that number of contributors [11:06] Son_Goku: as I said, x10 less [11:06] Son_Goku: good to see you [11:06] Son_Goku: did you see my reply on the forum there? [11:06] that would require rejiggering the project in a way that currently isn't possible :( [11:07] the forum makes me angry, so I haven't looked since I posted yesterday [11:07] * zyga-suse_ hugs Son_Goku again [11:08] * Son_Goku sighs [11:20] kyrofa: thanks for the 11.0.4 nextcloud update. Is 12.x on the roadmap? (I'm not complaing, btw. I don't even know what version 12 brings. Just curious!) [11:33] ogra_: what does it take to add /var/lib/snapd/apparmor/snap-confine.d to the core snap? [11:33] two lines ? [11:33] (assuming you want it writable, else one line) [11:33] yes, two lines [11:33] er [11:33] yes writable [11:33] where? [11:33] heh [11:34] note that it's under /var/lib/snapd/apparmor which was writable before [11:34] does it need any work? [11:34] or do I just need to fix CI prep to mkdir that too [11:34] zyga-suse_, https://github.com/snapcore/core/blob/master/live-build/hooks/20-extra-files.chroot for the mkdir -p ... [11:35] why there's no /var/lib/snapd/apparmor there/ [11:35] zyga-suse_, https://github.com/snapcore/core-build/blob/master/config/etc/system-image/writable-paths to make it writable [11:35] -> /var/lib/snapd auto persistent transition none [11:35] that's already present, so I think I'm good [11:36] well, you want a mkdir somewhere ... if snap-confine creates it on startup or package install that should be sufficient [11:36] else you need it in some of the build scripts like above [11:36] ogra_: the snapd package has that directory [11:37] ok [11:37] ogra_: it's made and packaged in the deb [11:39] PR snapd#3809 closed: tests: copy files with less verbosity [11:40] mvo: thank you [11:40] mvo: please help me out with one thing [11:40] mvo: when I want to add a new directory under /var/lib/snapd/ (specifically /var/lib/snapd/apparmor/snap-confine.d) should I do anything more than just update packaging? [11:41] mvo: tests are failing and I assume it is because of this [11:41] mvo: I have a patch that looks like : [11:41] https://paste.gnome.org/pkhbjpoju [11:42] mvo: but I'm unsure if that will really hide the problem until release [11:43] hmm [11:43] zyga-suse_: lets get to the bottom of this, the dir in the package should be enough, let me try to reproduce via spread [11:44] you might have a problem if you need it on non-new systems ... [11:44] since /var/lib/snapd is "transition" not synced [11:44] can you explain? [11:45] transition only copies the dirs on the first run ... [11:45] from core to writable [11:45] "transition"? [11:45] aha [11:45] I see [11:45] so an upgraded ubuntu-image install wont simply add it [11:45] well [11:45] suggestions? [11:46] systemd job ... or switch to "synced" but i'm not 100% sure about the latter [11:46] (have to check the code) [11:46] when making a change to an interface, is building snapd cmd sufficient to pick up the changes? `go build -o /tmp/snapd github.com/snapcore/snapd/cmd/snapd` [11:46] ugh [11:46] willdeberry: yes, you need to re-start it though [11:46] willcooke: which backends are you changing [11:47] willdeberry: some things are not refreshed on startup [11:47] willcooke: (sorry, bad tab completion) [11:47] zyga-suse_, for refrence http://manpages.ubuntu.com/manpages/xenial/man5/writable-paths.5.html [11:48] ogra_: I just need an empty directory [11:48] but it _must_ be there [11:48] ogra_: and it may need to be there very early [11:48] ogra_: before apparmor profiles are loaded [11:49] and I think those are done before regular systemd units start [11:49] well, given apparmor itself relies on writable-paths, that should be a given [11:49] the issue is transition vs synced [11:49] WARNING: This is a one-off operation which requires that the [11:49]