[01:36] kyrofa: is there any documentation on the changes to the catkin plugin syntax when you specify a base core? [01:57] <`dw> is there some way to get around the forum's "new users may only include (1 pictures|2 links)" restriction on the forum? [02:45] PR snapd#6377 opened: tests: auto-clean the test directory [05:16] NickZ, sorry, we're not timezone aligned this week at all! When you specify a base, the catkin plugin will use the ROS release that corresponds to it. core/core16 will use kinetic, core18 will use melodic [05:16] We've moved away from being able to specify the rosdistro so that we can ensure we're running the supported ROS distro on the supported ubuntu distro [06:11] morning [06:11] hey mborzecki ! [06:11] mvo: hey hey [06:12] #6370 is finally green [06:12] PR #6370: interfaces/builtin/opengl: allow access to NVIDIA VDPAU library [06:12] mvo: can you take a look ^^ ? [06:12] mborzecki: I just have it open - the author has signed our CLA? [06:13] mvo: the cla check is green, hope that's the same as confirmation he's signed it :) [06:14] mborzecki: yeah, let me do a quick check [06:15] Good morning [06:15] mborzecki: yeah, looks good [06:15] hey zyga ! good morning [06:15] #6326 sound like 2.37 material too, jdstrand requested changes there, but it's fixed already [06:15] PR #6326: interface: raw-usb: Adding ttyACM ttyACA permissions [06:15] zyga: hey [06:16] how was yesterday? [06:16] I saw some failures on master [06:21] zyga: things are mostly ok [06:21] zyga: very few failures [06:21] zyga: afaict at least [06:22] mborzecki: I taged it 2.37 but will consider it a target of opportunity. I would really like to have the +1 from jdstrand on pr#6326 before merging it [06:26] mvo: that's good, must have been a fluke on one branch then === pstolowski|afk is now known as pstolowski [08:02] morning [08:04] hey pstolowski - good morning [08:14] mborzecki: meh, this configure hook with network-control forum topic is a bit of a can of worms, looks like we have a bug in snap-confine for hooks, the cgroup name is incorrect so no devices are part of the cgroup :/ [08:16] oh, what's that? [08:17] pstolowski: there is this forum topic "stracing a configure hook" [08:18] pstolowski: and it points to a real bug it seems in snap-confine [08:20] interesting, thanks [08:24] mvo: which topic is that? [08:24] mvo: ah the strace topic [08:24] mborzecki: yeah, I think I'm close, its quite twisted though [08:25] mborzecki: anyway, just wanted to let you all know that I look into it (to avoid duplication of effort) [08:26] mvo: no worries, i was looking into https://forum.snapcraft.io/t/restart-a-snap-background-service-on-resume-from-suspend/9182/5, power mgmt can of worms [08:28] i don't understand why it's so convoluted [08:29] mborzecki: yeah :/ [08:29] mborzecki: thanks for looking into this one [08:33] mvo: about cgroup, iirc we add a bunch of /dev/* by default [08:33] mvo: including /dev/null [08:34] mvo: https://github.com/snapcore/snapd/blob/master/cmd/snap-confine/udev-support.c#L168 [08:36] mborzecki: yeah, its all good [08:36] mborzecki: I mean, its not :) [08:36] mborzecki: but the code is mostly DTRT, it seems there is a bug in the cgroup name generation [08:37] re [08:37] mborzecki: so we do DTRT just to the wrong cgroup (test-snapd-with-configure-nc_hook.hook_configure vs test-snapd-with-configure_nc_hook_configure [08:37] hey zyga [08:38] mmm [08:38] mvo: interesting [08:38] cgroup name for hooks [08:38] interesting [08:38] I'll let you handle that but one word of advice [08:38] device cgroups sticks around [08:38] zyga: :) ok [08:38] you can run something and then, at your leisure, look at it later [08:38] and you can move a shell there for full hands on [08:39] let me know if you need help with any of that === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [09:31] pstolowski: hi, can you take a look at https://github.com/snapcore/snapd/pull/6373 ? [09:31] PR #6373: overlord/ifacestate: helper API to obtain the state of connections [09:34] guys, if you have any thoughts about presenting disconnected plugs/slots or undesired connections in the output of snap connections, feel free to share them ehre https://forum.snapcraft.io/t/rfc-snap-connections-command/4296/12 [09:39] mborzecki: o/ [09:39] Chipaca: hey [09:39] mborzecki: I'll reply there [09:39] Chipaca: sure [09:48] pstolowski: yeah, maybe the endpoint should only return connections, connected or undesired, but no plugs/slots list like we get from /v2/interfaces [09:50] mborzecki: ah, going even further. might be a good idea [09:52] PR snapd#6378 opened: cmd: fix snap-device-helper to deal correctly with hooks [09:53] mborzecki: there ya go [09:53] hah [09:53] Chipaca: thx, reading [09:53] pstolowski: agreed on -d :-) [09:54] woops typo [09:54] * Chipaca fixes [09:54] mborzecki: wrt my last comment, I can do that in a followup if you'd rather not bash your head against tabwriter and escape sequences :-) [09:55] Chipaca: tis fine, go ahead [10:23] PR snapd#6379 opened: ifacestate/tests: extra test for hotplug-connect handler [10:26] degville: o/ [10:26] Chipaca: hello! [10:27] degville: PM [10:33] degville: should I create the topic for #6369, or would you rather do it? (if I do it it'll be a placeholder for now) [10:33] PR #6369: Add check for snap binaries dir not being in path

[10:34] Chipaca: feel free to create the topic. I guess you'd want it to be as short as possible. I can always add the text. [10:35] degville: I'll leave it unlisted for now :-) === chihchun is now known as chihchun_afk === cpaelzer_ is now known as cpaelzer [11:26] PR snapcraft#2439 closed: snap: add xslt dependencies for lxml === chihchun_afk is now known as chihchun [11:46] PR snapd#6380 opened: Nm iface add perms === chihchun is now known as chihchun_afk [12:31] Chipaca: https://paste.ubuntu.com/p/VVNpnmJ6QY/ [12:31] still have to add the count of disconnected/unconnected ifaces === chihchun_afk is now known as chihchun [12:47] hey marcustomlinson :) [12:48] hey pstolowski ! [13:02] off to pick up the kids [13:08] mborzecki: can you take a look at https://github.com/snapcore/snapd/pull/6379 ? [13:08] PR #6379: ifacestate/tests: extra test for hotplug-connect handler [13:40] Hey all. [13:41] Is there any fix for broken electron snaps on 18.10 yet. I now have no text in atom's menu bar! :/ [13:50] Gargoyle: I wasn't aware they were broken :-) [13:51] let me start an 18.10 vm to have a look [13:51] Gargoyle: is there a bug or forum post talking about the issue? [13:53] Gargoyle: I have text in atom's menu bar here [13:53] Gargoyle: https://i.imgur.com/56r236E.png [14:16] mborzecki: Conflicts=sleep.target WFM AFAICT FWIW [14:17] but maybe I should test it a bit more :-) [14:18] Chipaca: interesting, i'll play a bit more with these on the laptop [14:23] mvo: hi! miss you in sunny cape town :) [14:23] mvo: it seems like you are looking into this, so fyi, https://forum.snapcraft.io/t/stracing-a-configure-hook/9452/7 [14:23] mvo: also, I'll look at PR 6326 again [14:23] PR #6326: interface: raw-usb: Adding ttyACM ttyACA permissions [14:26] mborzecki: to be clear, Conflicts=sleep.target works to stop the service before suspending [14:27] mborzecki: you need something else to wake it after resuming [14:27] mvo: re 6326, just need a one character fix then feel free to commit [14:28] mborzecki: you'd think there'd be a resume.targt, but lennart says "Why would a service need something like this? This sounds systematically flawed" [14:29] Chipaca: hahah [14:30] Chipaca: hard to disagree with him [14:30] mborzecki: that's https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=744753#30 fwiw [14:32] Chipaca: on a side note, the use case from the forum should probably hook to nelink and act on interface state changes [14:33] * Chipaca covers his ears and goes LALALALA [14:39] https://github.com/snapcore/snapd/pull/6380 [14:39] PR #6380: Nm iface add perms [14:39] this is one terse commit message [14:40] I re-wrote the PR title [14:42] jdstrand: cool, will look at it [14:43] jdstrand: heh - I miss the sun as well! next time I can make it again. I am looking into the strace right now, I think I have a fix [14:43] jdstrand: not strace, the device cgroup with configure hook support [14:58] Chipaca: left a note in the forum, imo the guy should really be using avahi [15:04] PR core18#114 opened: Set the C.UTF-8 locale for the 001-extra-packages.test [15:13] PR snapd#6381 opened: devicestate: add initial Remodel support [15:14] Chipaca: hm that interface count is quite interesting, the API actually does some filtering and returns only connected interfaces by default, unless the client asked for all [15:23] mborzecki: isn't this the api that you're adding in the same PR? [15:24] #6373 [15:24] PR #6373: overlord/ifacestate: helper API to obtain the state of connections [15:24] ah no [15:24] * Chipaca looks more [15:24] #6333 [15:24] PR #6333: daemon: introduce /v2/connections snapd API endpoint [15:25] Chipaca: not the same PR, but yes, that's the API i'm adding, that's why it's helpful to get some feedback on the functionality people expect [15:25] oh and it's 'changes requested' by me and I never looked again [15:25] I'm a horrible person [15:25] * Chipaca wanders off [15:25] haha :P [15:25] :-) [15:26] Chipaca: so the API actually takes ?select=all to include disconnect plugs/slots and undesired conections, otherwise it's just connected [15:26] i could push everything and filter in the client [15:27] mborzecki: or, add counts to the interfaceJSON (if I'm reading the code right) [15:27] Chipaca: hm some itneresting bits are still missing in that pr [15:28] mborzecki: how is gnome software going to use v2/connections? [15:29] Chipaca: that's what I discussed with pedronis but haven't pushed out yet: https://paste.ubuntu.com/p/G4h3q4n2Zz/ [15:30] aha [15:31] Chipaca: Established is what's currently connected, so one does not have to dig inside Plugs/Slots, Undesired is what's manually disconnected [15:33] mborzecki: what about things that aren't undesired and aren't connected? [15:33] i mean established [15:33] i mean: established means connected right now, undesired means manually disconnected? [15:34] Chipaca: yes, established === connected, undesired === manually disconnected [15:34] mborzecki: so what about the other ones? [15:35] where do they appear? [15:37] Chipaca: that's where you dig into Plugs/Slots [15:37] Chipaca: iow, pairs that could be connected, but aren't right? [15:37] yeah [15:38] Chipaca: yup, so that's in Plugs/Slots, unconnected ones have len(Plug.Connections) == 0 [15:38] Chipaca: but there is no indication what given plug/slot could be connected to [15:39] fun! core18 fails to boot in spread :( spread qemu:ubuntu-core-18-64 [15:40] Chipaca: to be fair, you could connnect any pair that uses matching interface [15:41] mvo: excessive collisions & not enough packet ambulances ? [15:41] mborzecki: I'm never fair [15:41] hah :) [15:44] mounting /dev/sda3 on /tmpmnt_writable failed: device or resource busy [15:47] cachio: ^- could you please have a look? maybe a new kernel or a new core18 that breaks this [15:47] cachio: I need to leave for ~2h for a kids appointment unfortunately so can't look right now === chihchun is now known as chihchun_afk [16:20] mvo, in 10 minutes after lunch :) [16:21] mvo, Ill leave the tests running [17:02] PR snapd#6377 closed: tests: auto-clean the test directory === pstolowski is now known as pstolowski|afk [17:58] PR snapd#6382 opened: tests: use pc-kernel from stable instead of edge to build core-16 and core-18 [18:21] ouch, EOD [18:21] * Chipaca runs [18:49] mvo, hey [18:50] mvo, same issue I see when I start a core-18 vm locally [18:50] related to lack of entropy [18:51] we are currently faking the entropy source on google instnaces, perhaps we could do the same on this core instance before booting it [19:01] cachio: I was getting a busybox shell when running spread on ubuntu-core-18 here locally [19:02] cachio: you can run SPREAD_QEMU_GUI=1 spread ... [19:02] cachio: and then you get the qemu window and can get around the entropy problem [19:03] cachio: and thanks for looking into it! if reverting to the stable kernel fixes all issues we should probably let the kernel guys know :) [19:09] mvo, it fixes the problem [19:10] but now I see other issues becuase it refreshes the kernel snap [19:10] during the execution === grumble is now known as AMENDMENT_EFFFFF === AMENDMENT_EFFFFF is now known as grumble [19:12] mvo, I'll need to omve the IMAGE_CHANNEL to stable too [19:12] mvo, not sure if it if gonna work [19:21] PR core18#114 closed: Set the C.UTF-8 locale for the 001-extra-packages.test [19:34] cachio: ok [19:34] cachio: not sure what is going on, please keep me updated! I can start digging into that in my morning [19:35] using the image and gadget snap = stable it wokrs [19:38] it is a workaround until the issue is fixed [19:38] cachio: yeah, interessting! [19:38] there is a PR with thnat changge [19:38] cachio: I wonder if we can figure out what happend, i.e. what the root cause is [19:39] well, the pc-kernel snap has changed [19:39] from edge to candidate [19:40] no idea yet which change on the kernel is producing the error [19:40] still researching [19:42] cachio: thank you [19:44] cachio: interessting, there was a new pc-kernel upload 8h ago and a new one 2h ago [19:45] cachio: maybe that helped with things? [19:45] klebers: hey, I saw two kernel uploads recently. one 8h ago and one 2h ago. are there any known issues in the prev kernel or just an update? [20:18] cachio: do we have all must-have 2.37 PR (that fix the tests etc) tagged in github? [20:23] I think there are nice to have [20:23] mvo, [20:24] It is not needed to add them to 2.37 [20:24] cachio: great, thanks [20:25] mvo, yaw [20:26] cachio: I go over the PRs in the morning, I hope we can finalize 2.37 then [20:29] mvo, great [20:29] I'll check the status tomorrow morning [20:29] thanks!! === chihchun_afk is now known as chihchun [21:46] PR snapd#6383 opened: tests: provide a fake random device to the core images === chihchun is now known as chihchun_afk [22:12] hi everyone, I build a snap with PhantomJS that works well in classic mode, but when in full confinement it cannot load libfreetype and libfontconfig libraries, even though they are packaged (in $SNAP/usr/lib...) [22:14] I'm a bit confused - from where the confined binary should load the libraries? from system-wide /lib & /usr/lib or the $SNAP/lib $SNAP/usr/lib ? in such case why cannot it see them by default? [22:15] LD_LIBRARY_PATH="/snap/webcookies/x23/usr/lib/x86_64-linux-gnu" seems to help but why isn't it set by default if libraries are snapped? [22:47] In answer to my own question from earlier - In synaptic you can select a package and choose "lock version". So I was able to pick my way through and change from PPA version to official cosmic version. [22:59] Chipaca: https://i.imgur.com/YVeUqls.png