[00:51] mwhudson: o/ [00:51] Chipaca: hello [00:52] mwhudson: question about snapd on debian [00:52] mwhudson: in 'man snap', if you look for the entry for 'info', is it indented too much compared to the other commands? [00:53] Chipaca: let's see [00:53] if so, that's the main bug we revved go-flags for [00:53] there are others around it using too much space needlessly, but those are less ugly [00:54] in any case it's not something critical :-) [00:54] ha this chroot doesn't have man [00:55] I'd make fun of your chroot, but UC doesn't have man either [00:56] Chipaca: man page looks like this https://paste.ubuntu.com/p/Qq7KPtZrQq/ [00:56] which is pretty bad (what are those Help Options about?) but doesn't sound like what you are saying [00:56] whoa [00:56] mwhudson: well, maybe [00:56] mwhudson: we "hid" the help options [00:57] mwhudson: which starts showing off the other bug, where it doesn't de-dent if a command only has hidden options [00:57] this is 2.30 [00:57] ah ok [00:57] so then we fixed go-flags [00:57] is the fix upstream? [00:57] yes [00:57] i guess i can rev it in debian and see what falls out there [01:00] the fix is https://github.com/jessevdk/go-flags/commit/14b8957836290f46f473f39b104761889b2f0af0 [01:00] but there are a bunch more fixes that are nice to have so if you can rev it, all the better [01:01] bah not in latest release [01:02] psh, "releases", who uses those [01:02] anyway, I should go do that "sleep" thing that seems popular around here [01:02] just out of FOMO [01:02] Chipaca: good night [01:02] \o [01:02] heh https://github.com/jessevdk/go-flags/issues/288 [01:03] yeah what he said [01:03] * Chipaca ⇝ zzzz [02:02] PR snapd#6398 opened: tests: update systems for google sru backend [06:20] morning [06:20] hey mborzecki [06:23] wow amdgpu-pro driver is a mess, no wonder nobody uses it [06:24] mborzecki: meh! on fridays I only want good news ;) [06:25] mvo: good news: today is friday :P [06:29] mborzecki: heh :) [06:55] heh no way that driver is going to work on arch, it's super outdated and i'd have to downgrade xorg and the kernel :/ [07:55] PR snapd#6398 closed: tests: update systems for google sru backend [07:59] hey ho [08:00] mvo: I'm back to work from yesterday [08:01] my two multipass machines lost their disks, I fortunately pushed everything out [08:06] zyga: mvo: left some comments in the debian packaging pr [08:06] mborzecki: thank you [08:07] mborzecki: I think my focus today is to do the downstream package release [08:07] it still fails to build correctly [08:07] unit tests fail in sbuild [08:07] but pass in the same machine running debian 10 normally [08:07] some a bit mysteriously [08:07] oh, sbuild is just chrot? [08:07] chroot [08:08] mborzecki: thanks! [08:10] zyga: I had the "smoke" suite passing and the main ones are running but of course that is not inside sbuild [08:10] zyga: also its still a snapshot, would love to try a 2.37.1 package build from a git tag too [08:10] mvo: yeah, [08:10] ideas on why it would fail in sbuild appreciated [08:10] I'm rebuilding my environment [08:14] zyga: let me try to run the import-debian-salsa-no-history .dsc in sbuild [08:14] (this will take a bit to setup the env) [08:15] import-debian-salsa-no-history? [08:15] that's a branch? [08:15] I haven't look at git much yet [08:15] ah [08:15] mvo: do let me know if you have any unit test failures [08:16] PR snapcraft#2443 closed: schema: allow before and after [08:23] PR snapd#6399 opened: packaging: make sure that /var/lib/snapd/lib/glvnd is accounted for [08:25] zyga: ^^ no rush, you can look at it when you get back [08:25] mborzecki, mvo: shall we merge the opensuse packaging branch [08:25] and then separately have someone look at bcond/with part of it [08:26] because I think we are paying the price for the divergence way too often [08:26] and that is managed there [08:26] zyga: sounds good to me [08:26] mborzecki: +1 on your branch but I'd rather work on reusing that for arch/debian than just keep whack-a-mole [08:26] mvo: if you agree then let's please land the opensuse PR [08:27] and if we do a .1 I would have an easier life for suse release too === pstolowski|afk is now known as pstolowski [08:28] mornings [08:28] pstolowski: hey [08:28] hey pawel [08:28] oh, btw, I will be off on monday [08:30] zyga: i can later look into bcond_with thing, hopefully my opensuse installation is still working [08:31] mborzecki: thank you [08:31] actually [08:33] zyga: I need to look at this again, I'm slightly wary that it might lead us to a tangent that makes the debian release slower, but I really need to look first [08:34] mvo: it's not affecting debian [08:34] mvo: it's an opt-in [08:34] only suse uses it now [08:34] any work to enable it for more distros is explicit [08:34] but [08:34] it's a reference list to look at [08:35] zyga: ok, I check it out in a wee bit, just need to do one more remodel thing [08:35] thank you! [08:35] it's not urgent but I'd love to improve upon that [08:36] and not stay in the current split model where we pay the price for each distro/release [08:37] PR snapcraft#2444 opened: snap: move to core18 [08:52] mvo: any luck with sbuild and tests? [09:01] hey Chipaca :) [09:02] moin moin [09:02] Chipaca: https://github.com/snapcore/snapd/pull/6395#pullrequestreview-193983092 :) [09:02] PR #6395: cmd/snap: use a fake user for 'run' tests [09:03] :-) [09:03] PR snapd#6395 closed: cmd/snap: use a fake user for 'run' tests [09:07] thank you sir :) [09:09] btw, I know this is a bit big but I'd love have some quick pass on the design in https://github.com/snapcore/snapd/pull/6360 [09:09] PR #6360: cmd/snap-update-ns: refactor of profile application [09:09] I think it's sufficient if one reads one function there [09:09] the rest is plumbing into that idea [09:10] specifically this one https://github.com/snapcore/snapd/pull/6360/files#diff-d5f861571a1e368f7c7f933b29316750R37 [09:10] the whole PR is about allowing that function to exist [09:10] and the fuction is a generalization of the existing algorithm (that part is unchanged) [09:11] but the details of system vs user profile are abstracted behind the interface [09:11] so reuse, easier testing and less code (the tests cover more code now so there's more +es than -es) [09:26] one interesting spread failure: https://paste.ubuntu.com/p/Tz4JCptgPb/ pstolowski something you may want to look at [09:27] pstolowski: on a second look, doesn't seem to be related to hotplug [09:28] mborzecki: hmm something went terribly wrong with reboot? [09:29] localhost.localdomain systemd[1]: etc-shadow.mount: Failed to check directory /etc/shadow: Not a directory [09:29] etc [09:30] pstolowski: ah, watchdog timeout [09:31] zyga: re sbuild> running but slow, hopefully finishing soon [09:32] pstolowski: actually itnersting, snapd was waiting for system reboot, and got killed by systemd because of watchdog timeout [09:37] zyga: hrm, I'm confused, it fails in sbuild early but works in pbuilder, need to look closer [09:51] zyga: I get errors in sbuild using an ubuntu kernel I think you had this too, trying on pure debian noe [09:51] now [09:52] zyga: TestSetupSnapConfineGeneratedPolicyError bits [09:54] mvo: iterating on sbuild [09:54] mvo: oh! [09:54] that's interesting [09:54] mvo: I'll pastebin my errors [09:54] we can compare [09:54] mvo: note, I got errors on both kernels [09:55] mvo: both ubuntu and debian-10 [09:56] zyga: interessting, my debian-sid sbuild is still building [09:58] I had to spend a moment to add caching to my setup as otherwise iteration was mostly network bound [09:59] now it should be quite a lot faster [10:02] * Chipaca ⇝ break [10:07] mvo: FYI: https://salsa.debian.org/debian/snapd/merge_requests/1 [10:07] mvo: that's just last evening state [10:08] zyga: nice [10:18] zyga: yeah, same errors in debian too [10:20] zyga: http://paste.ubuntu.com/p/5D3dggjbvM/ full log [10:20] that's the same that I see [10:20] ok [10:20] I think you can put this on hold now [10:20] let's attack it tomorrow in master [10:20] I'll work on packaging only [10:21] unless you see a point I think we should not both spend time on it today [10:22] zyga: its fine, I try to minimize the time on this [10:22] zyga: working mostly on remodel now [10:35] pedronis: if-and-when you have a chance, a short glance at #6400 would be good [10:35] PR #6400: overlord/snapstate: use an ad-hoc error when no results [10:35] PR snapd#6400 opened: overlord/snapstate: use an ad-hoc error when no results [10:35] pedronis: it's +30-7, so should be quick [10:36] this is to block the "yeah sure switch to non-existent channel" [10:36] but, it has more consequences, because of course it does :-) [10:42] mborzecki: nice catch (6317) [10:49] Chipaca: do you have a moment for me? I was wondering where the api of remodel should life: /v2/remodel? or /v2/model/ and a GET to get the current and a POST to post a new one? [10:49] mvo: where is the model kept? [10:53] Chipaca: in the state [10:54] mvo: what would you GET? [10:56] Chipaca: the currently get model [10:56] Chipaca: I implemented this for testing already as part of debug but was wondering if it should be more official [10:56] Chipaca: hence my wondering [10:57] Chipaca: the important use case for this endpoint is to post a new model :) [11:13] mvo: I think /v2/model would be fine [11:13] mvo: how is it used? [11:14] Chipaca: thanks, I think that makes sense [11:14] Chipaca: we will provide it via an api endpoint mostly [11:14] Chipaca: but I think there will also be a "snap remodel" command [11:14] mvo: ah, no 'snap model --re [11:14] ' [11:14] oh [11:14] Chipaca: interessting, maybe `snap model {get,set}`? [11:15] Chipaca: we have not talked much about the cli yet [11:29] mvo: I like snap model get [11:29] though I very much like "snap remodel" [11:29] it spells out explicitly this is an operation that can change things more than "assign a value" [11:33] zyga: yeah [11:33] zyga: I did "snap debug get-model" for now, not sure if we need something on the cli, will talk with samuele [11:42] mvo: I like that [11:56] amdgpu-pro is so much fun, i'm typing something in the terminal, suddengly gnome-session fails and the whole Xorg goes down :/ [12:04] and it's mind boggling qt wants to compile shaders to display a piece of text from qml [12:20] PR snapd#6401 opened: many: add /v2/model API, `snap remodel` CLI and spread test [12:44] mborzecki: i don't think it's unique to amdgpu ;) [12:45] pstolowski: frankly the windows drivers aren't much better, i've deliverately skipped the last 3 updates :P [12:50] :) [13:41] mborzecki: qml's text renderer uses distance field rendering technique, good for UIs where text can change size a lot (especially animated) to avoid lots of cpu-bound rasterisation. Yep it needs shaders, but why not? :) [13:41] * greyback puts https://github.com/snapcore/snapd/pull/6361 out there [13:41] PR #6361: kvm: load required kernel modules if necessary [13:47] greyback: interesting, sounds like i should search qt blog archives for the details :P [14:05] mvo, my internet connection is breaking up [14:05] can't connect ot the hangouts [14:58] PR snapd#6361 closed: kvm: load required kernel modules if necessary [14:58] mvo: https://github.com/snapcore/snapd/pull/6399 super simple review [14:58] PR #6399: packaging: make sure that /var/lib/snapd/lib/glvnd is accounted for [14:59] PR snapd#6317 closed: overlord/snapstate/backend: call fontconfig helpers from the new 'current' [15:00] mvo: thanks! [15:00] PR snapd#6399 closed: packaging: make sure that /var/lib/snapd/lib/glvnd is accounted for [15:01] mborzecki: thanks, merged and I cherry picked it into 2.37 in case we do anohter release of that [15:16] PR snapd#6397 closed: tests: Use backend: [autopkgtest] for smoke test suite [15:26] booo, multiple progress bars doesn't work :-( [15:26] * Chipaca takes a breaks === phoenix_firebrd is now known as murthy === murthy is now known as phoenix_firebrd_ === phoenix_firebrd_ is now known as murthy [16:37] * cachio lunch [16:43] mvo: hey, do you have a moment for https://github.com/snapcore/snapd/pull/6379 ? [16:43] PR #6379: ifacestate/tests: extra test for hotplug-connect handler [16:44] pstolowski: do you need a review? [16:44] mvo: yes, 2nd review [16:44] it's just a test [16:44] pstolowski: sure, looking [16:44] thanks! [17:00] I'm off to the gym. Will bbl, but if I don't see y'all, have a good weekend! [17:01] PR snapd#6379 closed: ifacestate/tests: extra test for hotplug-connect handler [17:02] mvo: ty! === pstolowski is now known as pstolowski|afk === murthy is now known as phoenix_firebrd_ === phoenix_firebrd_ is now known as murthy === murthy is now known as phoenix_firebrd_ === phoenix_firebrd_ is now known as murthy === murthy is now known as phoenix_firebrd_ === phoenix_firebrd_ is now known as murthy === murthy is now known as phoenix_firebrd_ === phoenix_firebrd_ is now known as murthy === Luke is now known as bitnomial === bitnomial is now known as Luke [23:00] so, I have added wget in a stage-packages list, but that command isnt found when I try to run it from a shell script inside the snap... am I missing something?