=== jamesh_ is now known as jamesh [05:16] morning [05:53] good morning [06:12] o/ mvo [06:12] how are you feeling? [06:13] zyga: good morning! much better, thank you [06:16] zyga: mvo: hey [06:19] hmm [06:19] on cosmic snaps seem to be broken? [06:19] zyga@fyke:~$ canonical-livepatch [06:19] 2018/09/25 08:19:08 error executing : open /var/snap/canonical-livepatch/42/livepatchd.err: permission denied [06:20] mborzecki: hey [06:20] zyga: all snaps? [06:20] ah, no it's just. a bug in canonical-livepatch [06:20] zyga: this sounds like its time for 18.10 in CI now that we are in beta freeze [06:20] zyga: aha, just c-l ok [06:20] canonical-livepatch uses wrong permission for status file https://www.irccloud.com/pastebin/asnxhatD/ [06:23] * zyga wonders where to report bug on canonical-livepatch [06:26] ok, bug reported [06:27] PR snapd#5858 opened: Skip unsupported architectures for fedora-base-smoke test [06:35] mvo: review on https://github.com/snapcore/snapd/pull/5845 [06:35] PR #5845: interface: add new dotfiles interface [06:37] brb, too cold, need more clothing [06:44] re [06:51] zyga: cool, thank you, I have a look === pstolowski|afk is now known as pstolowski [07:03] mornings [07:07] pstolowski: heya [07:08] PR snapd#5832 closed: [RFC] overlord/{snapstate,assertstate}: parallel instances and refresh validation [07:11] hey pawel [07:20] zyga, those chilly mornings are the best [07:20] kyrofa: best to get ill? :-) [07:20] or are you serious [07:20] my hands are freezing [07:20] Hahaha, I'm serious, I love it when it's cold and I need to bundle up all cozy [07:20] kyrofa: do you type with your gloves on or do you just have warm hands anyway? [07:21] kyrofa: it's not freezing yet [07:21] it just feels like proper autumn already [07:21] zyga, yeah not gonna lie, it's probably 60-something here and I'm working outside [07:21] So not the same situation :P [07:21] kyrofa: hey! how is the place? [07:21] 6C in the morning (42F) [07:22] kyrofa: tonight we will have 3C [07:22] kyrofa: also, unusual to see you online that early ;) [07:22] pstolowski, man, I couldn't ask for better. It's so quiet, and I have this courtyard to myself [07:22] pstolowski, haha, yeah I'm not used to you guys being around either! [07:22] kyrofa: beware of the squirrels ;) [07:22] mborzecki: partial review on 5857 [07:23] I'll make warm tea and be back for more [07:23] zyga, there are some massive bees here, that's about all I've noticed so far [07:23] kyrofa: bees or wasps? [07:23] zyga: thanks [07:23] kyrofa: as in, do you get genuinely larger bees? [07:23] kyrofa: or is "bee" a word for anything bee-like [07:23] * zyga loves bees [07:24] I'd say they are bees, but they may completely different. They're like half the size of my fist, and black [07:24] But they're fat like a bee [07:24] kyrofa: did I tell you I want to have bee hives? [07:24] !!!! [07:24] that's some bee :D [07:24] I'm too busy flailing them away from me to look very closely [07:24] bumblebees? [07:24] if you can snatch a photo I would love to see those [07:24] mborzecki, it's a little bigger than that, and black [07:25] zyga, haha, I'll try, they tend to come out when it gets warmer [07:25] zyga, actually I've always been curious about bee hives as well. Do you have room for them? [07:26] kyrofa: i envy you! [07:26] kyrofa: no, but I would like to buy some land [07:26] kyrofa: and get properly trained to handle bees [07:26] kyrofa: my dad and granddad had honeybees :) [07:27] kyrofa: supposedly the best moment to start learning about them is early summer, the season ends in august and then pretty much nothing happens [07:28] kyrofa: I subscribed to a periodical about bees; so far I know very little though [07:34] PR snapd#5858 closed: Skip unsupported architectures for fedora-base-smoke test [07:36] zyga: thanks for your review on the dotfiles interface - the missing check for "." is intentional - I was thinking we could use this for non dotfiles as well at least potentially. if something neeeds access to a specific non-dotfile in home we could use the interfaces to give it without giving the full home away [07:37] hmmm, I would give the interface a different name then [07:37] like file-access or folder-access? [07:37] or user-data-access [07:39] actually [07:39] eyah [07:40] user data access feels like the right name [07:40] but I'll defer to the naming overlords ;-) [07:52] zyga: my thinking exactly :) [08:14] * mborzecki wonders what happens if one has networkd and NM managing different interfaces and competing for the default route [08:17] mborzecki: good question, when NM is in charge it will setup the default routes just with different metrics [08:19] mvo: yeah, would you need to manually tweak those or are the deafults non conflicting? i see the default route is using 600 here, but no clue whether that's dhcpclient's or nm's doing [08:20] mborzecki: I have 100 for eth and 600 for wlan but my two interfaces are both managed via NM :/ [08:20] iirc a default 1000 if you add it by hand right? [08:21] mborzecki: just tried with "route" and got metric:0 [08:21] hmm [08:22] well, i guess if one has a device which uses both netword and NM you craft the setup carefully enough so that it does what your requirements say [08:27] mvo: mborzecki: should I mention classic devices? === cpaelzer_ is now known as cpaelzer [08:56] zyga: quick brainstorm: should the dotfiles interface infere if its a file or dir from the trailing "/" ? i.e. .foo is a file and .foo/ is a dir? or too magic? [08:57] zyga: this way I can get rid of the files/dirs attr and just use paths or something [09:03] mvo: I strongly think this is too magic [09:03] zyga: ok [09:03] So either allow both or make it explicit [09:03] zyga: in this case the validation needs to reject "foo/" for files [09:03] zyga: but allow "foo/" for dirs [09:03] Yes [09:03] Well [09:03] You kind of do clean on it already [09:04] So either allow trailing / on dirs [09:04] zyga: I do but you pointed out it might be confusing to disallow "foo/" on a dir [09:04] Or always reject it and add it internally like you do now [09:04] Right [09:05] Because the reason for that is unclear IMO [09:05] zyga: I will look into it but my gut feeling is to reject it because the alternative is more code [09:05] We routinely refer to directories with / [09:05] Ok [09:05] As long as it is documented [09:05] Files and directories explicitly have [09:05] one more advantage [09:06] If somebody uses a symlink it is not going to work [09:06] But perhaps rightfully so [09:06] zyga: yeah, I think thats ok, symlinks & apparmor are tricky [09:18] PR snapcraft#2291 opened: meta: put environment into runner instead of app wrapper [10:22] PR snapcraft#2292 opened: sources: properly handle pull failures [10:41] PR snapd#5825 closed: tests: add snap install hook with base: core18 [10:42] niemeyer: when you have a moment, a high level review of 5717 would be great. I reworked it based on your input and it will auto-retry to get out of degraded mode now [10:42] niemeyer: this will unblock two more PRs :) [10:43] mvo: Heya, looking [10:43] niemeyer: also 5583 (without snaps stop and wait for socket) would be great but that might be more work :/ [10:43] niemeyer: (but I also addressed your input there) [10:43] niemeyer: thanks! [11:00] brb [11:00] mvo: Sent a couple of notes on the first one [11:14] mvo: Second one reviewed [11:17] mvo, zyga, everyone: I'm taking the afternoon off today to recharge batteries a bit [11:19] niemeyer: I was just thinking of doing the same [11:20] finding it hard(er) to concentrate [11:20] niemeyer: yeah, I needed that yesterday [11:20] niemeyer: enjoy the remains of autumn :) [11:27] PR snapd#5859 opened: many: introduce facts, export experimental features as facts [11:46] ppisati, poke ... [11:46] mborzecki: yesterday I saw code of yours where you took a string like "a:b::c:d" and return []string{"a", "b", "c", "d"}; where was that? [11:47] ogra: poke back [11:47] ppisati, would you be able to include the bvc4 libs in the pi2-kernel initrd ? [11:47] *vc4 [11:47] Chipaca: #5857? [11:47] PR #5857: userd: extend the list of supported XDG Desktop properties when autostarting user applications [11:47] ppisati, https://paste.ubuntu.com/p/v4qCGjTNrz/ would be the list [11:49] ppisati, that makes psplash work even with the vc4-kms-v3d overlay (else it only work with vc4-fkms-v3d, but that is completely unaccelerated) [11:50] when using vc4-kms-v3d /dev/fb0 is only created by the vc4 module ... while vc4-fkms-v3d gives you a kernel /dev/fb0 but without any acceleration [11:51] so to have fb0 in the initrd i'd need the modules available there (and i doubt we should compile vc4 into the kernel since many headless use casess wont want it) [11:51] ogra: xenial or bionic? i guess both [11:52] ppisati, yeah., likely both, i have only worked with xeinal yet though [11:52] ogra: if you open me an LP with the list of modules to include, i'll start working on it [11:52] \o/ [11:52] * ogra hugs ppisati [11:53] ppisati, hmm, LP just against linux-raspi2 ? (since i want it in the snap)# [11:53] ogra: yep [11:53] on it [11:58] Chipaca: thanks for the tip [11:59] ppisati, Bug #1794279 [11:59] Bug #1794279: please include vc4 modules in pi2-kernel snap initrd [12:01] mborzecki: note you can get megabytes into the environment before it starts to fall apart [12:01] I seem to recall that 8MB in an env block was enough for execve to fail, but that might be old (32 bits?), i haven't tested the limits in this way in ages [12:02] you can easily creae an env var that'll make bash crawl: FOO=$(yes) [12:02] some of those used to be limited to ~~~ 32KB but that was long ago [12:02] (killall yes when you get bored; then 'echo "$FOO"' to be more bored [12:02] AFAIK there's no limit now [12:03] zyga: xargs --show-limits thinks still thinks there is a limit [12:03] ~2MB? [12:03] but that's mostly for execve (which applies in this case, but the other is fun) [12:07] ogra: ack [12:08] zyga: i wonder, when doing mount mappings for parallel installs under /var/snap/, would people expect this to have shared propagation [12:09] mborzecki: note also there's bytes.FieldsFunc which would save you the conversion (but then you want a []string, so ¯\_(ツ)_/¯ [12:09] ) [12:09] * Chipaca ~> lunch, and then probably nothing but probably guitar playing and maybe minecraft with the boys and dog walking and some more guitar and early bed [12:09] mborzecki: shared propagation is actually established as a default by snapd, we only get slave propagation though, right (because snap-confine) [12:09] but, hey, maybe some irc too. ttfn. [12:09] mborzecki: btw, I found a bug in propagation setting, something I need to investigate and fix ASAP [12:09] mborzecki: I have a test that shows htis [12:09] *this [12:10] mborzecki: we can sync after standup [12:11] zyga: i've extended fuse-support spread test a bit and obviously the mount point is only under the mapped hierarchy, now we've said (have we?) that _foo will be mapped under snap name, but i guess we did not promise anything about propagating mounts [12:11] yeah [12:11] I think that is the same bug [12:12] ok, let's talk after standup [12:19] mborzecki: https://github.com/zyga/broken-layout-migration [12:20] mborzecki: that shows this and one more issue found last week [12:37] PR snapd#5860 opened: interfaces/houtplug: helpers and struct updates [12:55] mvo: hey! Today I quickly sat down to finally see what's up with my raspi3 when running core18 on it, since I wasn't able to reproduce your original problem (the one with the invalid netplan config) [12:55] mvo: I wasn't able to reproduce it as console-conf was segfaulting [12:55] mvo: I did some debugging and ugh, it's segfaulting on probert when trying to probe for wifi [12:55] With the wifi module unloaded console-conf and probert all work fine [12:55] mvo: so apparently my raspi3 is suffering from probably the same (or very similar) issue as the dragonboard [12:57] mvo: so generally I'll poke mwhudson that probert seems to segfault on two different boards, I guess on two different architectures as well [12:57] sil2100: yeah, that sounds a lot like the segfault we had on the dragon [12:58] sil2100: did you push a debug build of probert into the image ppa? [13:01] zyga: pstolowski: standup? [13:11] hi all [13:11] https://snapcraft.io/nongnu-gsequencer/listing [13:11] ^^ please, do the review. [13:14] mvo, sil2100, note that we had the same issue with core16, caused by netplan (not probert) forcefully un/reloading the wlan drivers when probing [13:15] mvo, sil2100 there was a blacklist in netplan to avoid unloading most sdio based wlan drivers due to exactly that reason ... perhaps thats gone in 18.04 ? [13:21] zyga: snappy-devel? [13:22] mborzecki: actually, can we postpone for a sec, I need to see the stuff that my daughter asked about [13:22] zyga: ok [13:29] ogra: oh! Good to know! I'll investigate that after lunch [13:30] sil2100, here is an example bug (there were three or four like this) https://bugs.launchpad.net/netplan/+bug/1741910 [13:30] Bug #1741910: ath6kl_sdio does not support unbinding [13:31] that black/whitelist was extremely ugly, i can imagine that cyphermox tried to replace it with some more intelligent code (which has perhaps bugs) [14:05] mvo: no c18 meeting? [14:07] zyga: we can have one if you want [14:07] zyga: but I think its fine to skip given that gustavo and samuele are not around [14:07] yeah [14:07] let's skip [14:13] 2018-09-25 13:32:01 Cannot allocate google:arch-linux-64: cannot allocate new Google server google:arch-linux-64 (sep251328-278432): cannot find ready marker in console output for google:arch-linux-64 (sep251328-278432) [14:13] cachio: have you seen something like this ^^ [14:15] mborzecki, when did you get that? [14:15] 45 minutes ago? [14:15] cachio: yes [14:15] could you please try again? [14:15] ok [14:15] otherwise I'll revert last image which I manually updated today [14:32] mborzecki, seems to be failling in many branches [14:32] I'll revert it [14:36] ok [14:37] cachio: can you debug what's wrong with the image? [14:37] mborzecki, yes [14:37] I'm on that [14:38] this is already deprecated [14:38] so arch should be up again [14:44] mborzecki, https://paste.ubuntu.com/p/XD87JBKzxs/ [14:44] have you seen that? [14:45] zyga, ~ [14:45] nope [14:56] ogra: I doubt that's it as it's segfaulting earlier, but who knows - thanks for the context o/ [15:01] sil2100, good luck anyway :) [15:11] random thought for the day: make snaps able to tell APT/DPKG that they "supply" the same as a debian package - the specific example I came across that would benefit me personally, a deb package depends on docker-ce deb package but I had the docker snap installed so it refused to install this other package because it thought I didn't already have docker-ce [15:12] obviously that only benefits debian-based distributions so we'd need an abstraction to allow the same function on rpm or other distros [15:13] PR snapd#5811 closed: tests: add test that runs snapctl with a core18 snap [16:18] * zyga returns for evening work [16:32] hi, so it seems like all of the snappy desktop applications on my system tend to hang/crash [16:32] like I get a window for libreoffice, but nothing is drawn in it, I can't close it, and I can't start a new one [16:32] same thing with the spotify snap [16:33] halfbit: hello, that's unfortunate, sorry about that - can you tell me more about your system (output of "snap version" helps) [16:33] this is on a fresh install of ubuntu 18.04 [16:33] tburdick@jupiter  ~  snap version  ✔  8627  11:31:33 [16:33] snap 2.35 [16:33] snapd 2.35 [16:33] series 16 [16:33] ubuntu 18.04 [16:33] kernel 4.15.0-34-generic [16:34] what does snap changes say? [16:35] "snap changes" [16:37] error: no changes found [16:37] hmm, ok [16:37] what happens when you run "snap run spotify" from command line? [16:38] I re-installed spotify with the deb which works perfectly, let me try with libreoffice [16:38] ok [16:40] ok now I'm confused [16:41] yeah its just working now [16:41] of course [16:41] infuriating [16:41] perhaps it was the one-time setup that some apps are doing? [16:41] I have no idea [16:42] but its beyond frustrating to go from a working system to something where I'm mucking around with this :( [16:42] if you want you can "snap install spotify" and see if that starts [16:43] yeah it started spotify and its working [16:43] PR snapcraft#2292 closed: sources: properly handle pull failures [16:44] well I guess its good for now [16:44] I'll be back if it stops working again [16:44] good luck [17:46] PR snapd#5717 closed: snapd: go into degraded mode when the selftest fails [19:16] https://github.com/auchenberg/volkswagen === epod is now known as luk3yx === epod is now known as luk3yx [19:32] * ogra hopes thats not a diesel [19:33] :-) that repo is full of win [20:23] PR snapcraft#2293 opened: build providers: use the new snapcraft: remote for multipass [20:38] PR snapcraft#2294 opened: snap: workaround the dirty tree [20:40] PR snapd#5861 opened: cmd/libsnap: add functions for handling facts [20:40] PR snapd#5862 opened: cmd: fix C formatting [21:50] PR snapcraft#2295 opened: tests: use SNAPCRAFT_PACKAGE_TYPE everywhere === matteo| is now known as matteo