=== JoshStrobl is now known as JoshStrobl|zzz === chihchun_afk is now known as chihchun === JanC is now known as Guest58454 === JanC_ is now known as JanC === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [07:14] mornings [07:42] Good morning [07:48] good morning [08:34] o/ === JoshStrobl|zzz is now known as JoshStrobl [09:44] zyga-ubuntu, can we merge this now? https://github.com/snapcore/snapd/pull/3984 [09:44] PR #3984: release,cmd,dirs: Redo the distro checks to take into account distribution families [09:45] Son_Goku: hello [09:45] Son_Goku: looking [09:45] morning [09:47] Son_Goku: +1'd, I'll ack with mvo and merge [09:47] mvo is probably not going to be alive for some time... [09:47] Son_Goku: I'm a bit absent minded today with the news about catalan news [09:48] Son_Goku: how was the event? [09:48] Son_Goku: at least you didn't have to fly far :) [09:48] I didn't fly at all [09:48] it's too close to fly ;) [09:48] it was pretty good [09:48] though now I'm sick :( [09:49] Son_Goku: me too :) [09:49] Son_Goku: it was 3C in the morning today [09:49] Son_Goku: and the temperature jumps around to 15 and back all the time [09:50] it's 8C right now [10:05] PR snapd#3966 closed: cmd/snap-seccomp,osutil: make user/group lookup functions public [10:05] PR snapd#3988 closed: Added note in HACKING file [10:10] zyga-ubuntu, hey, pondering about your comment https://github.com/snapcore/snapd/pull/3972/files/46c79aed60e0b8a3cfe7d827a32af297f2d4950e#r141618790 [10:10] PR #3972: repo: sanitize plugs and slots early in ReadInfo [10:10] zyga-ubuntu, ie. the check if snap.SanitizePlugsSlots is not set. it's tricky considering the tests [10:11] zyga-ubuntu, would you expect it to panic if already set? or just print a warning? [10:13] pstolowski: hey [10:13] pstolowski: I was thinking about it myself [10:13] pstolowski: it'd be easier if it would allow a nil value and just handle that elsewhere [10:13] pstolowski: then any non-nil value could panic [10:14] pstolowski: as for tests they should be fine as they can skip that check and assign the variable directly [10:14] zyga-ubuntu, I did with a global bool for a quick test, but the problem is tests seem to be creating more than one instance when they e.g instantiate overlord [10:15] aha [10:16] pstolowski: well, think about it [10:16] pstolowski: maybe you can come up with something [10:17] zyga-ubuntu, sure, will keep investigating === chihchun is now known as chihchun_afk [10:42] * zyga-ubuntu finishes NFS spread test and picks up another topic [10:54] hmm [10:54] so I have 8GB of ram now and I still see swap [10:54] :/ [11:00] zyga-ubuntu: Same here, I can't find a way to avoid it even if to my best knowledge memory never exceeds what I have [11:01] kalikiana_: I know linux swaps out unused memory just to use the RAM better but I really want to avoid that [11:01] meanwhile, I'm in shell quoting hello [11:01] *hell [11:03] I thought that's what "shell" stood for if you spelt it out ;-) [11:03] shell hell :) [11:10] I'm already feeling the reverse jetlag... this week is not gonna be any easier to get through [11:12] so I have quoting fixed, now for the other half of the test === ShalokShalom is now known as ShalokShalom2 [11:51] kalikiana_: well, at least you're not sick [11:51] ok, more progress, expanded the NFS test, now just to ensure it restores and I can push that [12:03] woot, passed [12:03] ok [12:11] ok, now 14.04 to check it works too [12:13] Son_Goku: now you're making me wonder... but I sure hope it's just the jetlag, at least now I can take power naps during the day [12:14] on the sprint that was difficult, with it being a constant hallway conversation and such [12:16] I hope I can get over this cold quickly... [12:17] pstolowski: hey, can you please have a look at https://github.com/snapcore/snapd/pull/3958 [12:17] PR #3958: many: add support for /home on NFS [12:17] yep [12:18] thanks [12:20] I need some tea and medicine, back soon [12:20] (the joys of working from home is that you can work while ill) [13:00] hmm, are all snaopcraft.io URLs (including the store) dead or is it me ? [13:00] ogra_: well, https://snapcraft.io/ works for me [13:00] pstolowski: and https://github.com/snapcore/snapd/pull/3965 is also easy to review [13:00] PR #3965: interfaces/mount: add support for parsing x-snapd-mkdir-{mode,uid,gid}= [13:00] Running late, but will be in the standup in a couple of minutes [13:00] * zyga-ubuntu joins [13:06] weird [13:08] Son_Goku, thanks ... seems it is related to the canonical VPN (my desktop got stuck) === JoshStrobl is now known as JoshStrobl|AFK [14:46] * zyga-ubuntu is back but needs to do another errand, I'll be online but off irc as connection will be hit/miss [15:40] flexiondotorg, did that alsa fix actually end up working for you on that snap you were working on? [15:44] kyrofa: It did. [15:45] I want to make it a remote part because anything needing just ALSA will require that bespoke set of config files. [15:45] Yes. Definitely [15:46] flexiondotorg, I _think_ this is the case but want to verify: those */card/* files are generic, right? They're the entire set of config files for every card supported by alsa, not somehow specific to the hardware on which alsa is installed. Agreed? [15:46] flexiondotorg wait, you got things working without the need to snapcraft-preload? [15:46] i.e. it actually does makes sense to make it a remote part? [15:47] kyrofa: I'm guessing you said something like 'have a nice day' but it was slow motion robot voice :-P [15:47] sergiusens, yeah baby [15:47] kalikiana_, hahaha [15:47] kyrofa: The card configuration is generic but requires the paths are modififed in each one. [15:47] sergiusens, this is what happens when you go get coffee [15:47] flexiondotorg, of course. Excellent [15:47] I was thinking of just biting the bullet and creating snapcraft quirks logic for stage-packages [15:47] sergiusens: Not quite. It works in classic but still needed to ALSA config work around. [15:48] flexiondotorg, can I help you with the remote part? I want to make sure we don't drop that [15:48] we already do it implicitly for other things, we should probably just make it an official thing [15:48] To work in strict I have to manually add `/dev/shm/* rwl,` to the AppArmor profile. [15:48] a remote part for now should be ok [15:48] flexiondotorg: that's what preload is fixing ;-) [15:48] flexiondotorg that is a bit broad though [15:49] Sadly, we couldn't get snapcraft-preload to work. [15:49] I dicussed with sergiusens and he has some ideas about what can be added to snapcraft-preload. [15:50] flexiondotorg I want to make a minimal "configurable" version of it so you do not unnecessarily have to preload everything you may not want to [16:15] sergiusens: so are my fixes still blocking on the rewrite? [16:16] or should the user of preload define stuff like /dev/shm or /tmp manually instead? [16:17] Might be worth opening a forum topic [16:21] * kalikiana_ wrapping for today [16:24] kalikiana_: o/ [17:31] jdstrand: thank you for the review, have you had a good trip home? === JoshStrobl|AFK is now known as JoshStrobl [17:33] PR snapcraft#1581 opened: catkin plugin: support rosdep pip dependencies [17:35] sergiusens, that one is the last bit for one roadmap item ^^ [17:36] kyrofa anything pending before that needs to go in? [17:36] PR snapcraft#1420 closed: add new "no-wrapper" property to apps [17:37] kalikiana_ if your preload fixes are already converted to c++ you should be fine and just requires a review [17:37] sergiusens, that has no dependencies [17:37] zyga-ubuntu: yes, it was uneventful. direct flight, home friday night. couldn't have gone better. thanks for asking :) [18:00] jdstrand: and you didn't even wind up sick :) [18:00] well, knock on wood :) [18:04] Son_Goku: my weirdest patch to snapd *ever*: https://github.com/snapcore/snapd/pull/3958/commits/56aaca91a5d7e7aed076703186c729a3283d4c2c [18:04] PR #3958: many: add support for /home on NFS [18:05] wtf [18:05] that... literally makes no sense as a patch [18:05] it makes the PR green :) [18:05] also, you're really not going to like this... [18:05] the previous patch failed because "exportfs" is obviously a mis-spelling of "exports" [18:05] rot13 isn't in /usr/games in Fedora [18:05] * Son_Goku grumbles [18:05] Son_Goku: I know, I know, I just made this test to run on 16.04 [18:06] Son_Goku: I will expand it to do better and use python to do the rot13 [18:06] Son_Goku: I just wanted to push something quickly not and not re-design it today [18:06] this is such a stupid PR [18:06] for such a dumb reason [18:06] Son_Goku: no, I mean, the PR is nice [18:06] Son_Goku: but the patch that made it green is silly [18:06] (read the rest) [18:06] err, the patch I mean [18:07] right, no disagreement tere [18:07] *there [18:07] fixing /home on NFS is absolutely good [18:07] and necessary [18:08] zyga-ubuntu: https://github.com/SELinuxProject/cil [18:08] this looks like it'd be useful for snapd [18:08] as it has its own HLL [18:08] Son_Goku: if I have a moment next week I'll update this to work with cifs [18:08] thanks, I'll check it out [18:09] err: https://github.com/SELinuxProject/cil/wiki [18:10] wow, a documented project [18:10] I wonder how that happens [18:10] are all developers a great tech writers? [18:10] are the docs written ahead of code? [18:11] CIL docs were written before code [18:11] the whole thing had a huge design process ahead of time [18:11] it was explicitly designed to make writing SELinux policies and policy modules easier [18:11] and allow developing custom languages on top [18:12] and since snapd already has its own custom format, it would make sense to use CIL rather than the normal language (HLL) [18:12] that does look sensible, yes [18:12] CIL stuff can be directly loaded by libselinux [18:12] and semodule [18:12] * zyga-ubuntu sees lisp-like language [18:13] haha [18:13] intermediate languages tend to look very LISPy [18:13] * zyga-ubuntu needs to run (not because of lisp, lisp/scheme are great) [18:13] yep :) [18:13] I'll be back later [18:25] kyrofa heh, you changed the PR while I was looking it seemed ;-) [18:25] sergiusens, just the commit, no content. Forgot to ref the bug [18:40] PR snapcraft#1582 opened: plugins: add ros2 boostrapper [18:45] sergiusens, that one is ready as well ^^ [19:00] kyrofa do you have a 14.04 environment handy? [19:00] kyrofa almost done that review, got distracted by someone wrong on the internet [19:00] sergiusens, I'm afraid not, but I can make one. Who's wrong, let me at 'em [19:01] kyrofa lol, I am sort of joking fwiw ;-) [19:01] :P [19:04] (cough relatively trivial to setup a 14.04 env with either `lxc launch ubuntu:t` or `uvt-kvm sync release=trusty arch=amd64; uvt-kvm create trusty-vm release=trusty arch=amd64`, but I suspect you knew those :) [19:14] nacc it is a matter of having the correct bandwidth :-) [19:14] sergiusens: heh [19:16] diddledan does your corebird currently support wayland? [19:41] kyrofa added comments to the review, looks good, tell me what you think and will do another quick round [19:43] jdstrand, I'm getting "snap-confine has elevated permissions and is not confined but should be" on 14.04, using the hwe kernel. Am I missing something? [19:44] kyrofa: that's probably because it's setuid [19:45] Son_Goku, well, that doesn't explain why it's not confined [19:45] probably the AppArmor profile didn't apply [19:46] kyrofa: did you reboot into that kernel? [19:46] it is like Son_Goku said. [19:47] jdstrand, Yeah [19:47] just because I don't use AppArmor doesn't mean I don't know how it works ;) [19:47] what does 'sudo aa-status|grep snap-confine' have to say? [19:48] kyrofa: ^ [19:51] jdstrand, https://pastebin.ubuntu.com/25662590/ [19:52] kyrofa: hmm, what about just 'sudo aa-status' [19:53] jdstrand, https://pastebin.ubuntu.com/25662599/ [19:54] kyrofa: and: snap version ; snap list ; cat /proc/version_signature [19:55] jdstrand, https://pastebin.ubuntu.com/25662608/ [19:56] kyrofa: did you set SNAP_REEXEC=0 somewhere? [19:57] jdstrand, no, this is a fresh install, nothing weird here [19:58] let me try in a vm [20:04] kyrofa: it worked here. just a simple 'sudo apt-get install snapd ; sudo reboot ; sudo snap install hello-world ; hello-world' [20:05] kyrofa: what is the output of 'dpkg -l|grep apparmor' [20:05] jdstrand, wait... I've rebooted several times now in order to play with VM networking settings, and all of a sudden things are working [20:07] As rebooted after I installed the core snap, so I have no explanation for this [20:07] kyrofa: so, that is only supposed to happen if the profile fails to load on a distro that supports apparmor [20:07] Hmm [20:07] so, the deb is 2.27.5~14.04 [20:07] but the core is 2.27.6 [20:08] idk [20:08] hard to say what it was since it is now working. it worked right away here [20:11] kyrofa: did you try to install the snap before rebooting? [20:11] jdstrand, yeah I did. But I installed from the netboot image with the hwe kernel [20:18] kyrofa: I don't know. I used a desktop install. maybe snapd is missing a dependency? iirc, apparmor_parser isn't in netinst (which uses ubuntu-minimal) [20:18] jdstrand, that's possible [20:19] I should've phrased that differently. apparmor isn't in netinst, iirc netinst uses that instead of ubuntu-standard [20:59] Uh. sergiusens have you ever seen pip-installed packages get chmod'd 700? [21:00] kyrofa nope [21:00] It's suddenly happening to me and I have no explanation as to why [21:06] I just tested this a few days ago... it's like pip updated from underneath me, but it hasn't been updated for a while [21:09] sergiusens, it looks like it's expected behavior from pip install --user [21:10] kyrofa new behavior? [21:10] * sergiusens needs to run and pick up his kid from day care [21:10] No, I'm seeing stuff from 2013 about this. I'm so confused [21:11] kyrofa wait, check if we have any postprocessing in the python plugin itself [21:12] Haha, _fix_permissions, dangit [21:13] Well, easy fix at least [21:17] I still can't explain why my testing ever passed, but I'll let it slide [21:28] Maybe I tested as root for some odd reason [21:28] In a container, perhaps [21:35] sergiusens, alright, snapcraft#1581 is ready for another look [21:35] PR snapcraft#1581: catkin plugin: support rosdep pip dependencies [21:36] That was annoying [21:52] kyrofa: yeah, containers as effective root have made of my testing in the past invalid too [21:52] kyrofa: might be worth adding a local cloud-config snippet to ensure the ubuntu user has your ssh key and go that route [21:53] nacc, yeah I actually do that. The only way I would have tested like that would have been unusual and a mistake. But it's the only way I can explain it :P [21:53] kyrofa: ah ok [22:02] sergiusens, +1 from OSRF on snapcraft#1582 [22:02] PR snapcraft#1582: plugins: add ros2 boostrapper [23:15] niemeyer, https://forum.snapcraft.io/t/error-when-updating-snap-and-cleaning-old-revisions makes for a very sad experience in lxd :( [23:16] But only after one has time to get invested in it [23:18] kyrofa: Thanks for the ping.. let's find someone to have a look at this so zyga can focus on layouts [23:19] niemeyer, no problem, I wasn't sure if anyone else was familiar with snap-confine other than zyga or jdstrand [23:20] kyrofa: zyga worked a lot on it, but we can find someone else to have a look [23:20] niemeyer, excellent, thank you [23:37] kyrofa: np! [23:37] * niemeyer looks for dinner