=== shadeslayer_ is now known as shadeslayer === beowulf is now known as Guest75286 [12:55] Good morning everyone [12:57] shuduo_, I'd like to discuss your ROS problems today if you're around [13:05] utlemming: do you have all-snap builds already for gce? [14:08] hmm... latest ubuntu-core all-snap breaks my etherpad :( [14:08] good news is rollback ubuntu-core resurrected it [14:09] hmm. who broke stuff? [14:10] * asac updates again [14:12] jdstrand: tyhicks: ok latest core bails my ethpad because it restricts syscall=49 which seem to not have caused issues before [14:14] * asac reboots and tries to find where the security pol info now lives [14:16] hmm. what was the script againthat spits out syscall names? [14:26] kyrofa: snapcraft generated something odd for caps for me: http://paste.ubuntu.com/14164111/ [14:26] any idea? [14:26] my input was not having caps as a list item: [14:26] asac, eww [14:26] cat snapcraft.yaml | pastebinit [14:26] http://paste.ubuntu.com/14164118/ [14:27] anything that strikes you? [14:27] Hmmmmm [14:27] wow :) [14:27] interestingly this worked though [14:27] it gave my pad the network-servbices permission [14:27] Wait, what? [14:27] Hahaha [14:27] and now the node process does not go defunct with bind syscall bailing anymore [14:27] weird [14:28] let me double check [14:28] that i am on the ubuntu-core that has that problem with that that caps entry [14:28] kyrofa: do you see in code whats going on with the caps transform? [14:28] There's got to be a bug here. Two I'd say, if that actually runs :P [14:29] yeah [14:29] :) [14:29] asac, I'm looking now [14:29] let me commit this so you can run it yoursel;f [14:29] give me one sec [14:30] so confirmed that without that odd caps it bails on bind [14:30] now adding it again and committing [14:31] asac, which snapcraft are you using? [14:34] kyrofa: master HEAD ... https://github.com/asac/etherpad-lite/tree/snap-support [14:34] get that branch, cd bin/snappy [14:34] run snapcraft snap [14:34] from master HEAD [14:35] Alright, let me play with it for a minute [14:35] https://github.com/asac/etherpad-lite/tree/snap-support/bin/snappy [14:35] also works with 1.x [14:35] which is coolie :) [14:36] but i suspect that we never really prevented bind etc. because it stopped working on all snaps from -2 to -4 when i didnt have that caps [14:36] (in case you care about ubuntu-core runtime as well) [14:36] ok stepping out for 5 min [14:36] bbiab [14:37] kyrofa: oh one more thing... i rfind it odd that we dont run "snap" as default on master... it just bails if i dont pass any target [14:37] anyway... bbi5 [14:37] asac, yeah it surprised me as well [14:38] asac, we'll pass that by sergio when he gets back, see how he feels about a default target again [14:49] kyrofa: i get the same caps syntax on 1.x [14:50] asac, alright. We were just about to release, so I'll hold off [14:50] lets see if 15.04 also eats this fine at least [14:51] asac, and yeah, I can reproduce so I'm working on this now [14:51] kyrofa: release without sergio around or might he pop in today? [14:52] asking bc i had two pulls i think woudl be nice to have included [14:52] https://github.com/ubuntu-core/snapcraft/pulls [14:52] topmost [14:52] not needed for the variant above, but for my other build from outside recipe its needed [14:52] asac, I was hoping he might pop in long enough to make sure I didn't blow it to kingdom come, though that may not have happened anyway (in which case I wouldn't release, of course) [14:53] e.g. for https://github.com/asac/etherpad-lite-snap [14:53] ok... well, since i made this inline approach now the pull above could wait if it feels risky [14:53] but have it for 1.x and master in case [14:54] anyway, the caps is probably more interesting :) [14:54] asac, ah okay. I'll make sure I take a look at those :) [14:54] ok so seems 15.04 also eats the - caps [14:54] ODDDD [14:54] :) [14:54] asac, yeah that's quite an impressive screwup === shuduo_ is now known as shuduo [14:54] asac, you should be proud [14:54] kyrofa, hi [14:54] shuduo, hey! [14:55] kyrofa, let me post ROS build log [14:56] shuduo, please do. If you can post the ROS code, that would be great as well [14:57] kyrofa, http://pastebin.com/RHbPTeWv actually i am trying to build ros example code of snapcraft I git clone from github.com [14:58] shuduo, ah that makes it easy [15:02] ok filed bug for odd runtime beahviour here: https://bugs.launchpad.net/snappy/+bug/1528871 [15:02] Launchpad bug 1528871 in Snappy "snappy accepts "wrong?" caps syntax in package.yaml (both on 16.04 and 15.04)" [Undecided,New] [15:03] kyrofa: https://bugs.launchpad.net/snapcraft/+bug/1528873 is the other side [15:03] Launchpad bug 1528873 in Snapcraft "snapcraft produces supposely bad caps formatting in package.yaml" [Undecided,New] [15:05] asac, excellent, thank you :) [15:05] shuduo, wow, probably the most helpful error ever, huh? [15:06] shuduo, I'm particularly sorry that happened while you were trying to show it to someone [15:06] shuduo, that error, while being insanely unhelpful, I believe indicates that you're trying to launch a binary that can't be found, so it can't be wrapped [15:07] shuduo, I only know this because it happened to me just a bit ago (an error message I plan on fixing) [15:07] shuduo, can you give me a link to what you're trying to build? [15:08] asac, give me a sanity check? [15:09] kyrofa, that's fine if you are already aware. I'm trying to build https://github.com/ubuntu-core/snapcraft/tree/master/examples/ros [15:10] asac, http://pastebin.ubuntu.com/14164908/ == good yaml? [15:10] (in python form) [15:10] kyrofa: which one? [15:10] ahh [15:10] thougth pull etc. [15:11] kyrofa: give me a patch and i try the real snap [15:11] but the first looks ok now [15:11] the binaries list [15:11] asac, I'm not to that point just yet. To me, that paste looks good. It's handed off to yaml.write, then if I read the write that's written, it looks terrible (as you've seen) [15:12] I think yaml.write is a third-party lib [15:12] oddd [15:12] asac, excuse me-- yaml.dump [15:12] Still investigating [15:13] asac, have you tried making your caps lists regular lists (i.e. not inline [item1, item2] but: [15:13] -item1 [15:13] - item2) [15:13] so yeah i think it might be correct actually :( [15:13] * asac looks again [15:13] yeah its correct [15:14] just super unfortunate ordering [15:14] * asac marks invalid [15:15] asac, ohhhh it IS isn't it! the hyphen caps was really throwing me [15:15] right [15:15] But yeah, still a valid map [15:15] a way to guuide the yaml formatter? [15:15] /list of maps [15:15] asac, I doubt it :( [15:15] think we would love to see the serializer to produce name first [15:15] and exec then [15:15] etc. [15:15] and caps last :) [15:15] lol [15:15] ok neat [15:16] asac, I'll look into it though [15:20] asac, why did you look at the generated package.yaml? How often do you think devs do that? [15:21] many times [15:22] caps will be very frequent [15:22] it hink we should investigate if we can subclass Dumper [15:22] or something and give some sanity hints somehow [15:22] so it looks neat [15:22] like you would write it [15:22] start with name [15:22] exec/start: [15:22] description: [15:22] end with caps list [15:23] maybe we could use an event parser to remember the order for those elements that are just pass through? [15:23] let me file a bug that the formatting looks ugly [15:23] :) [15:23] but nothing important afaics [15:23] asac, yeah if people look at that very often you're probably right. Do you mind logging a separate bug? [15:23] Haha, yeah thanks :) [15:24] Yeah I agree-- I'll get back to squashing release-blocking bugs. I'll take a look at your PRs as well :) [15:26] ok filed https://bugs.launchpad.net/snapcraft/+bug/1528878 [15:26] Launchpad bug 1528878 in Snapcraft "snapcraft package.yaml generation could be more beautiful and more human friendly" [Undecided,New] [15:28] oh i think the real thing is that my stuff gets transformed from map style to list style... where the name is in the name: field... maybe we could just keep same style to make things better for now [15:29] ok added comment to bug for this [15:30] oh seems my pull request adds too long line :) [15:31] asac, thanks for working out the kinks! [15:31] * asac fixes [15:50] kyrofa: i have no idea how the test stuff works. where does it find the test package.json for the node test? [15:50] asac, the nodejs plugin test? [15:51] kyrofa: yeah [15:51] asac, master branch? [15:51] sure [15:53] guess i dont get the mock [15:53] asac, after a quick look, there seems to be only one test that would require one, and it makes it itself (test_build_local_sources) [15:53] asac, the others just mock out the npm install [15:54] right [15:54] hmm [15:54] let me think [15:54] so i added a new option [15:54] that allows to point to a dir other topdir for part to have the package.json [15:54] asac, well, since npm install is essentially shelled out (using snapcraft.common.run), if one mocks snapcraft.common.run, you can verify that npm install was called correctly without actually calling it [15:55] fguess we are not testing npm /path/to/where/package.json is right now [15:55] asac, no, since (I'm assuming) /path/to/where was always the root? [15:56] in the past it was [15:56] now its allows to use a subdir of the part [15:56] ok let me try something [15:56] asac, okay, so make a new test that sets up that environment, with the subdir etc [15:56] how do i run those tests though locally? [15:57] Then assert that the run_mock was called with the right path [15:57] asac, that's outlined in the readme, but you can run the unit tests with ./runtests.sh unit [15:59] git diff | pastebinit [15:59] http://paste.ubuntu.com/14166326/ [15:59] how about that? [15:59] err bad syntax [15:59] moment [16:00] git diff | pastebinit [16:00] http://paste.ubuntu.com/14166382/ [16:00] that oen [16:00] ok let me in the readme [16:08] Sorry asac uhh... my mouse pointer disappeared :P [16:09] Having trouble getting those links. I guess X needs a restart [16:09] asac, be back in a minute and I'll be happy to help :) [16:10] ok i think i fixed it :) [16:11] lets see what travis thinks [16:13] Back [16:13] kyrofa: ok so i think i fixed those i have a clue about [16:13] those that are left i dont understand [16:14] https://travis-ci.org/ubuntu-core/snapcraft/builds/98540359 [16:14] oh [16:14] one sec [16:15] asac, alright so you may have a bug here [16:16] The npm install call happens with a working directory of the builddir [16:16] But you seem to have modified it to pass an extra parameter which in the case of test_build_local_sources is the same as the working directory [16:16] That may work fine, but the test will need to be updated to assert a different call on the mock [16:16] (i.e. including that extra parameter) [16:18] asac, the second failure is similar, though it seems like you're actually using subdir there. You need to change the mock assertion to make sure that extra parameter is there [16:18] right will update test [16:19] ok let me do this after dinner etc. [16:19] thanks for your help [16:19] no need to hold back release [16:21] asac, no problem! And yeah, I don't think the release is happening today. Sergio hasn't gotten back to me (I don't blame him) and I found another bug :( [16:22] asac, so a) don't work too hard, and b) releases are easy. We can always make another [18:01] Ugh... who ever thought that "Santa Baby" was a decent song === devil is now known as Guest96027 === Guest96027 is now known as devil_ === Guest75286 is now known as beowulf === beowulf is now known as Guest52016 === Guest52016 is now known as beowulf