=== ogasawara is now known as Guest14527
=== c74d is now known as Guest58751
=== wgrant is now known as Guest27601
=== kgunn is now known as Guest52384
=== Guest27601 is now known as wgrant
=== bigcat_ is now known as bicgat[afk]
dholbachgood morning06:50
fgimenezgood morning07:09
dholbachogra_, ogra2: do you know who could help figuring out https://bugs.launchpad.net/snappy/+bug/1506480 (link to "Kernel requirements for Snappy" is broken and I'm not sure what else to link to)?07:20
ubottuLaunchpad bug 1506480 in Snappy "Broken link in https://developer.ubuntu.com/en/snappy/guides/porting/" [Undecided,New]07:20
Chipacaasac: beuno: http://jimbenton.tumblr.com/image/13116293422408:26
tasdomasthe wiki specification for snappy confinement mentions an assign.yaml file that can be used instead of running the hw-assign command08:30
tasdomashowever, I couldn't find any more information about it (format, etc.)08:30
tasdomasis it supported?08:30
Chipacatasdomas: where does it mention that?08:33
tasdomasChipaca, https://wiki.ubuntu.com/SecurityTeam/Specifications/SnappyConfinement08:34
Chipacatasdomas: there's no assign.yaml in that doc08:35
Chipacatasdomas: it says "assign yaml"08:35
Chipacatasdomas: no dot08:35
Chipacatasdomas: it's not a file; it's part of the oem config08:35
tasdomasChipaca, ah, sorry - misread that08:35
Chipacatasdomas: so, yes it's supported, for oem snaps08:36
tasdomasChipaca, so for my own snap the only thing to do is use hw-assign, right?08:37
Chipacatasdomas: right08:37
Chipacatasdomas: what're you wanting to do?08:38
tasdomasChipaca, I was writing a simple snappy app to control the piglow device on my orange matchbox08:38
tasdomasChipaca, http://domas.monkus.lt/posts/2015-10-15-writing-a-snappy-application/08:38
tasdomasChipaca, my main problem was that I need to restart the service after assigning the i2c device to it08:39
Chipacathat's surprising to me -- but then i know little about i2c and the rpi208:39
Chipacatasdomas: are you sure it's not that you need to restart the service that uses it?08:40
Chipaca(that'd be bug 1484645 fwiw)08:40
ubottubug 1484645 in Snappy "Snappy hw-assign doesn't restart affected services" [High,Triaged] https://launchpad.net/bugs/148464508:40
tasdomasChipaca, sorry - maybe I wasn't clear - my problem was that I needed to restart the service (exposed by my snappy app) that uses the i2c device08:40
Chipacawow, it's going to be a long day08:41
Chipacabecause you said exactly that08:41
Chipacasorry :-(08:41
Chipacatasdomas: that's a bug, we know about it, should get fixed (on rolling/edge) soonish08:42
tasdomasChipaca, ah, cool!08:42
tasdomasChipaca, can I bother you with one more question?08:42
Chipacatasdomas: i dount you can bother me with a question, but you're welcome to try08:42
tasdomasthere's an mqtt snappy app available in the store - but I can't find the source for it08:42
tasdomasI want to modify it to make it mdns discoverable08:43
Chipacatasdomas: have you tried asking the snap publisher?08:44
Chipacaor gone to the snap support url?08:45
Chipacatasdomas: this one, i presume https://search.apps.ubuntu.com/api/v1/package/mosquitto.kartben08:45
tasdomasChipaca, ah, will do08:46
dpmdavidcalle, dholbach, looking at askubuntu, there seems to be quite a lot of questions around snappy already08:48
dpmmight be worth for us looking at them in the upcoming weeks08:48
dholbachmaybe we can talk about it in our next docs hour call?08:49
dholbachdpm, ^08:50
dpmit could be a good topic, yes08:50
davidcalledholbach, dpm, +108:50
tasdomasby the way, is it possible to use snapcraft to build a snappy package for a different architecture? e.g. build a RPi2 compatible package on my amd64 laptop?08:54
dpmdavidcalle, dholbach, it might be worth creating a snapcraft tag as well. Right now we've got 'snappy' already, which redirects to 'ubuntu-core'08:55
dholbachyep, that makes sense08:56
dholbachciao ppisati, do you know how we could fix https://bugs.launchpad.net/bugs/1506480? (which information should we link to in the article there?)08:58
ubottuLaunchpad bug 1506480 in Snappy "Broken link in https://developer.ubuntu.com/en/snappy/guides/porting/" [Undecided,New]08:58
JamesTaitGood morning all; happy Friday, and happy World Food Day! 😃09:00
ogra_dholbach, ppisati should be able to help with the link09:22
=== soee_ is now known as soee
dpmdholbach, davidcalle, do we have any document that specifies the format of the snapcraft.yaml file?09:30
dholbachdpm, no spec-like doc in our docs yet, but there's 1) https://developer.ubuntu.com/en/snappy/snapcraft/your-first-snap/ and 2) ./examples in lp:snapcraft and 3) https://bazaar.launchpad.net/~snappy-dev/snapcraft/core/view/head:/docs/snapcraft-advanced-features.md which will land as soon as the md importer is working09:32
davidcalledpm, closest thing we have to a spec for it in the the App dev manual09:32
dholbachah yes, and that09:32
davidcalleIt's actually quite detailed09:32
dpmdavidcalle, yeah, I've gone through 1) and 2), but I've been missing a doc that is specific to snapcraft.yaml09:33
davidcalledpm, https://docs.google.com/document/d/1Rj9nVBttx62BvGlbnkmKOzAlAIEWuLNr1QaSGe3gQDA/edit#heading=h.8o7w1g30gucy09:33
dpmI had actually seen this, but couldn't figure out where, glad I wasn't going mad09:35
davidcalledpm, not sure the link is world-readable though, hopefully next week in some form on duc...09:35
dpmdavidcalle, dholbach, http://askubuntu.com/questions/686167/what-is-snapcraft, this might help with AskUbuntu/Stackoverflow searches to give a 5 min intro and direct folks to the site09:52
Chipacamvo: ping09:53
davidcalledpm, this is excellent09:54
davidcalledpm, love it :)09:54
mvoChipaca: pong09:54
dpmglad it if it's useful :)09:54
Chipacamvo: this race condition with the automount unit09:54
Chipacamvo: tell me about it09:55
mvoChipaca: its not that much that I know - so the code will activate and start the automount unit, right after that it runs the click stuff09:56
dholbachdpm, nice one!09:56
dholbachdpm, now we can't change the URLs anymore! :-)09:56
mvoChipaca: when I do not unpack the meta,.click dirs it will fail to generate the apparmor profiles09:56
mvoChipaca: however ls a bit later is fine09:57
dpmdholbach, oh, I can update the link after we've done the IA rearrangement. Or just make it point to https://developer.ubuntu.com/snappy09:58
mvoChipaca: hm, maybe its something else, need to look deeper09:58
dholbachdpm, or we add redirects - in any case, no problem :)09:58
Chipacamvo: aiui, systemctl should not return until automount has actually started09:58
Chipacamvo: maybe we're creating the units but not starting the .automount until later?09:59
* Chipaca looks09:59
mvoChipaca: adding a sleep did not solve it, so its something else09:59
mvoChipaca: here we go, its a missing .click/manifest in the snap :/10:00
mvoChipaca: sorry10:00
mvoChipaca: let me fix that10:00
Chipacahue hue hue, to quote a brazilian10:01
mvoChipaca: we also may want to keep the .meta for now because of "snappy list -v"10:01
mvounless we go husks10:01
Chipacamvo: that one i don't follow10:01
Chipacamvo: once automount is started, any access under it will trigger the mount10:01
Chipacamvo: even if you're accessing things that are there10:01
mvoChipaca: so only the active snap has a automount unit10:02
mvoChipaca: so snappy list -v (old snap) will not have a meta/packages.yaml10:02
Chipacayep yep10:02
Chipacaso, yeah, leave it for now, flag it for removal when we switch to lightweights10:02
mvoyeah, will do10:03
Chipacamvo: have you checked it for speed?10:03
mvoChipaca: not yet, no10:04
Chipacamvo: the lightweight thing of looking at remote manifests does not work for sideloaded stuff though10:04
mvoChipaca: one more complication - the hooks expect .click/$name.$origin.manifest - and the $origin is not known at build time. so we can not ship that in the snap and need to generate it10:04
mvoChipaca: once hooks are gone this is no longer a issue of course10:05
Chipacahooks were going to be gone "soon" since i joined10:05
Chipacanot lookin' good10:05
mvothere is only one left10:05
mvobut that one is pretty sticky :P10:05
mvoChipaca: do you want me to split the branch for the snapfs-mount up more?10:57
Chipacamvo: how would you split it?10:58
Chipacaunless you moved all the systemd stuff to .. ahem .. systemd :-p10:58
Chipacamvo: anyway, you don't have to, but my review is slower :)10:59
mvoChipaca: I can do that I think if it helps (I will do that in any case in this MP, your suggestion makes sense)10:59
mvobut lunch first10:59
Chipacamvo: i'm not sure the total time would be less, so i'd say nah10:59
beunoChipaca, lol12:18
longsleepyaml question, how can i have a binary with a _ in the name?12:44
tedgsergiusens: So I think I may need to copy the options locally to stay under the 79 character limit :-)13:07
sergiusenstedg, harr harr :-)13:30
* sergiusens notices that today is not talk like a pirate day13:30
tedgsergiusens: It can be here!13:30
sergiusenstedg, I couldn't sleep last night thingking about the override of setup_stage_packages; maybe it is best to use repo.Ubuntu directly instead13:31
tedgSince we can't have casual Friday's working from home, we should have talk like a pirate Fridays.13:31
sergiusenstedg, right on13:31
tedgsergiusens: Heh, I never meant for you to lose sleep over it :-)13:31
sergiusensalthough my vocabulary is rather scarce13:31
sergiusenstedg, it is one of those things13:31
sergiusensI also stayed up late as I went to a docker meetup to see a bunch of old friends13:32
sergiusenslearned a lot about docker and why developer like it so much13:32
tedgI'm feeling like we might need to detach the plugin lifecycle from the overall one. Like have more phases for plugins.13:32
tedgI think it is nice. I just haven't see the overlays actually work well. Seems like they always get confused for me.13:33
ogra_because you talk like a pirate to them ?13:33
tedgStupid overlays are no fun, they never talk like pirates.13:38
dholbachsergiusens, tedg: snappy clinic planning?14:01
jdstrandmvo: hey, curious if you had a chance to review my email re snappy-debug. I know we were both pretty busy yesterday14:07
mvojdstrand: I saw it but have not acted on it yet, sorry14:18
jdstrandmvo: ack14:20
dholbachsergiusens, I found an interesting one: http://pastebin.ubuntu.com/12799579/15:03
dholbachnot sure if we support this15:03
dholbachtedg, ^15:03
dholbachI just used: apt-get source bamtools15:04
sergiusensdholbach, thanks, perfect example for custom plugin15:04
dholbachI'll leave that one for you then :)15:04
dholbachtedg, sergiusens: I added two more examples to the pad15:22
Chipacasergiusens: i should probably have pointed you at that talk earlier15:23
Chipacasergiusens: 's good stuff15:23
elopiomvo: Chipaca: yes! with this last test I saw the moment when the last cloud init message appeared and the moment when the mode switched to regular. It's just twice as slow as the last time I tried.15:23
elopiofgimenez: ^15:23
elopioChipaca: tell me more about waiting for a systemd service.15:23
Chipacaelopio: it's just a for loop with a sleep and a check15:24
Chipacaelopio: give me a sec15:24
elopioChipaca: ah, we have that.15:25
Chipacaelopio: ah! show me the code :)15:25
elopioI thought you were talking about subscribing a listener to the service or something like that.15:25
Chipacaelopio: well, you could, but why bother15:25
elopioChipaca: https://bazaar.launchpad.net/~snappy-dev/snappy/snappy/view/head:/_integration-tests/testutils/wait/wait.go15:26
elopioto work around the workaround I will just have to increase the maxWaitRetries.15:26
Chipacaelopio: and you're waiting for bootok?15:27
elopioChipaca: I'm waiting for snappy_mode=regular15:27
Chipacaelopio: and is that better, or worse than waiting for boot ok?15:28
* sergiusens wonders if it is beer o clock already15:28
Chipacasergiusens: it is!15:28
elopioChipaca: better, I think. Instead of checking the service, we check it's effect.15:29
Chipacasergiusens: you should come visit. can't have beer on my own (and ellie doesn't do beers)15:29
elopioit's beer-on-hangout o'clock.15:29
Chipacaelopio: so, yeah, looks like you're already doing the right thing and all :)15:30
Chipacaelopio: wrt increasing maxWaitRetries, maybe instead have an arch-dependent base?15:30
Chipacaelopio: ie scale checkInterval by some bogomips factor15:30
=== kickinz1 is now known as kickinz1|away
elopioChipaca: that would be nice.15:31
Chipacavar bogomips = map[string]int{"arm": 100, "amd64": 1, etc}15:31
Chipacaand index it on GOARCH15:31
elopioI'll make a card to be able to configure the wait, with a delay factor.15:31
elopiobut anyway, this is a bug :) Instead of working it around, somebody should fix it.15:32
* elopio looks at somebody.15:32
Chipacaelopio: which is the bug, exactly?15:32
Chipacaelopio: that you can ssh in before you can log in?15:32
elopioChipaca: you can sign in before boot-ok. both ssh and normal log in through serial. And if you are fast with a rollback, that makes the boot crazy.15:33
ubottuLaunchpad bug 1498293 in Snappy "after a successful update and reboot, the bootloader snappy_mode is set to 'try'" [Critical,Triaged]15:33
elopioon bbb you don't have to be super fast. I actually can reproduce this manually.15:33
Chipacaelopio: so that's the bug15:35
Chipacaelopio: not that "try" thing :)15:35
Chipacaso, we want the boot-ok to be done *before* ssh and getty lets people in15:36
Chipacapitti: do you remember offhand whether there's a target that would let us do that easily?15:37
elopioChipaca: I think so. You could fix the rollback so it waits for boot-ok, but then it might affect something else.15:37
elopioChipaca: and if you just move the log in after boot-ok, the reboot will take twice the time to let you access the board.15:38
Chipacaelopio: well, but it's letting you in before things are ready15:39
sergiusensChipaca, I don't mind visiting; when are we having a sprint in London?15:39
Chipacasergiusens: i have space for two for half a week every week15:40
Chipacaas long as they're not super tall :)15:40
elopioChipaca: well, before cloud-init has finished. Most of the things are ready.15:40
dholbachtedg, sergiusens: is the announce text roughly what you expected?15:41
sergiusensdholbach, where is it?15:42
dholbachsergiusens, ah sorry - on the pad :)15:42
sergiusensdholbach, looks good, just added on clarification just in case15:43
dholbachthanks sergiusens15:43
sergiusensdholbach, btw, how hard would it be to patch clog?15:43
dholbachrock and roll15:43
dholbachlet me check15:43
sergiusensdholbach, so ideally, patch CMakeLists.txt to take an arg to provide an alternate ~/.clogrc path15:44
tedgIf sergiusens is happy I'm good.15:44
* tedg didn't keep a link to the pad :-)15:45
sergiusens-DLOGRC_PATH=$SNAP_APP_DATA_PATH or something like that, and then the code would need to expand the var15:45
tedgdholbach: +115:46
dholbachsergiusens, I guess it's not too hard to do - maybe something to look into some other time?15:53
dholbachtedg, sergiusens: thibautr suggested next time we ask for proposals on what people would like to see snappified :)16:05
fgimenezhave a nice weekend o/16:05
sergiusensdholbach, sounds like a good idea16:10
sergiusensdholbach, maybe at the end of the clinic we ask that question too16:10
dholbachfor next time we can maybe leave it a bit more time in advance, so we can collect suggestions beforehand16:10
sergiusenselopio, very simple MP https://code.launchpad.net/~sergiusens/snapcraft/tested_or_not/+merge/27474117:04
elopio sergiusens: why is that needed? I find it good to see if all your test code is being run.17:06
elopioif it isn't, you can probably remove it. Or you have a condition that has to be faked.17:06
=== willcooke_ is now known as willcooke
sergiusensChipaca, can we get the release from the rest api already or not?17:21
Chipacasergiusens: i don't think that's in stable yet17:22
Chipacasergiusens: why?17:24
sergiusensChipaca, to be able to ask it to be able to ask the store if you know what I mean.17:24
Chipacasergiusens: updating to latest edge to see if it's there yet17:29
sergiusenselopio, ok, it just bothers me to be honest :-)17:31
Chipacasergiusens: wait, the release?17:36
Chipacasergiusens: you could always get the release!17:36
sergiusensChipaca, as in 15.04-core .. btw, I'm not saying you can't, I'm asking if you can! :-D17:37
Chipacasergiusens: the store id is what was missing, and is there now17:37
Chipacasergiusens: http://pastebin.ubuntu.com/12801235/17:37
Chipacasergiusens: ^ that's the output in rolling/edge :)17:38
sergiusensChipaca, yup, nice17:38
Chipacasergiusens: http://pastebin.ubuntu.com/12801274/17:39
Chipacasergiusens: that's 15.04, now17:39
Chipacasergiusens: note the socket name has changed17:40
sergiusensChipaca, life is good!17:42
sergiusensChipaca, mind looking at this https://code.launchpad.net/~sergiusens/snapcraft/build-deps-request-something-missing/+merge/274749 among other complicated things in your life?17:43
Chipacasergiusens: i don't *mind*, but it's friday afternoon and the malbec has already confessed its crimes17:44
sergiusensChipaca, look at it and then decide ;-)17:44
* Chipaca obeys17:44
Chipacasergiusens: no, that's too many lines of code for this hour17:45
Chipacasergiusens: also, the complexity is astounding17:45
Chipacasergiusens: flabbergasting17:45
Chipacasergiusens: maybe split it in two or three and have a team work on it on monday17:45
sergiusensChipaca, btw https://github.com/docopt/docopt.go17:46
Chipacasergiusens: also, in that branch, i think you're using the Schwarzschild metric wrong; you need to use Kruskal–Szekeres coordinates to express it like that17:49
pittiChipaca: Before=systemd-user-sessions.service? That's "before people can log in" (I didn't read the full backscroll)17:52
Chipacapitti: that includes ssh and getty (both tty and serial)?17:52
Chipacaelopio: could you try that?17:53
Chipacapitti: thanks!17:53
Chipacapitti: can something be after: multi-user.service *and* before: systemd-user-sessions.service?17:55
Chipacathat seems to be a no-no17:55
* Chipaca could test, but me has already checked out17:55
pittias multi-user is after user-sessions17:55
Chipacasuspected as much17:55
Chipacawe need to check on monday, then, whether sessions is enough17:56
pittiit has multi-user in it, after all :)17:56
pittiright; I'm just on a drive-by, need to leave again17:56
sergiusensChipaca, let me fix!18:34
sergiusensChipaca, I like the docopts thing, but I can't easily add help to commands18:56
sergiusenstedg, final comment wrt setup_stage_packages20:00
sergiusensand there's a merge conflict (very minor)20:01
* tedg isn't sure if he wants to look20:01
tedgsergiusens: I don't think we can use pull with the current configuration of the command. Because the superclass is just empty.20:02
sergiusenstedg, right, I mean, pull() -> handle_source_options;20:04
sergiusenstedg, then have your private implementation of setup_stage_packages keeping self.stage_packages empty (or reuse if you want)20:05
tedgOh, I see.20:05
tedgThat seems a little scarier to me...20:05
sergiusenstedg, in other words, setup_stage_packages should have always been _setup_stage_packages20:05
tedgBecause then we have two overrides20:05
sergiusenstedg, 2?20:06
tedgsergiusens: We'd have to override setup_stage_packages to be null, then have pull get sources, setup deps, and then call it.20:07
sergiusenstedg, let me try and propose something20:08
sergiusenstedg, meh, I'll do this in another proposal (I'll fix the priv/pub thing first)20:13
tedgsergiusens: I think restructuring pull() also would help.20:13
sergiusenstedg, indeed, that's why :-)20:14
sergiusenswell another reason why20:14
tedgI think there may need to be a "get_deps_from_source()"20:14
tedgOr something for plugins.20:14
tedgsergiusens: I'm getting an odd error with trying to copy system libs. Basically the -dev packages are bringing in symlinks and that seems to be confusing snapcraft.20:16
tedgsergiusens: Does that sound like something you've seen?20:16
tedgThe first results in the second :-)20:20
sergiusenstedg, that is weird, the path exists but can't be copied to20:31
sergiusenstedg, does /home/ubuntu/readlinetest/parts/ginac/install/lib/x86_64-linux-gnu/ exist?20:31
tedgsergiusens: No20:32
sergiusenstedg, that is the darn problem then :-/20:32
tedgsergiusens: I'll get an MR here in a sec.20:32
sergiusenstedg, its just a mnissing os.makedirs(os.path.basename(real_path, exist_ok=True)20:33
sergiusensbefore the copy20:33
sergiusenstedg, what I don't get is that that lib should of been relinked to the internal one20:35
tedgsergiusens: There is no internal one because it's on the list of blocked packages.20:40
tedgsergiusens: I'm not sure that readline should be on that list though.20:40
sergiusenstedg, ah, then the makedirs missing is the correct thing to do or if it is one of those that is always on the system, add it to the skip list20:41
sergiusenstedg, right, maybe not; the only real problematic one is libc20:41
sergiusenstedg, all others come inherited from deb2snap20:42
tedgsergiusens: So, I think it found a bug, but perhaps it isn't a bad bug once we get everything more refined.20:43
sergiusenstedg, what bug?21:04
tedgsergiusens: Not making the directory21:07
sergiusenstedg, right, that is indeed a bug21:11

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!