/srv/irclogs.ubuntu.com/2016/04/12/#snappy.txt

Swami_sure thanks kyrofa. if you are around00:53
Swami_will get back and look through it00:53
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
stgraberbeuno, jdstrand: Just uploaded a tiny fix for LXD 2.0.0 to the store (use the old lxcbr0 bridge as we don't have lxdbr0 on snappy yet)03:53
stgraberfolks upgrading didn't have any issue with rc9 or 2.0 final, but folks doing a clean install would have got broken networking, this will fix new installs03:53
stgrabercyphermox: ^03:53
cyphermoxstgraber: thanks05:09
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
davidcalleHi snappers, when using the snap find command, I don't see the fully qualified app name, how can I install apps without it?09:31
davidcalleAnyone, zyga maybe? ^09:59
davidcalle"when using the snap find command, I don't see the fully qualified app name, how can I install apps without it?"09:59
zygadavidcalle: store bug10:00
zygadavidcalle: and it's snaps, not apps10:00
davidcallezyga =) ty10:04
mvodavidcalle: beuno mentioned the other day they are working on automating this, for now we may need to set aliases for the store apps (worst case manually until this is automated). but I'm not sure what the status is, maybe its fixed already, beuno will know10:13
ogra_ubuntu@localhost:~$ snap install webdm10:16
ogra_error: access denied10:16
ogra_mvo, can we make that message a bit more user friendly ?10:16
ogra_(it used to tell me i should use sudo etc)10:17
oparozAny idea why this doesn't compile on arm? https://github.com/kyrofa/mdns-publisher/blob/master/main.go10:17
davidcallemvo: thanks10:18
oparozActually, that's wrong, it compiles, but fails in the stripping phase10:18
mvoogra_: yes, I think we should10:27
ogra_Building dependency tree...10:29
ogra_E: Unable to locate package ubuntu-core-libs10:29
ogra_P: Begin unmounting filesystems...10:29
ogra_mvo, s390x and ppc64el fail with that ^^^10:30
ogra_do we not build that package there ?10:30
oparozhttps://bugs.launchpad.net/snappy/+bug/156928010:35
ubottuLaunchpad bug 1569280 in Snappy "Snapcraft fails to strip go binaries on arm" [Undecided,New]10:35
mvoogra_: I'm surprised by that, I think that worked before :/10:36
ogra_mvo, did we ever buikld images for them ?10:36
ogra_(i just added both arches and started a testbuild)10:36
ogra_OH !10:38
* ogra_ wasnt aware it is a meta package :P10:38
mvoogra_: I don't think we did images for them10:40
ogra_right10:40
ogra_where the heck do the seeds live for that package ?10:44
ogra_ah, found them10:46
dakerkyrofa: hi sorry for the noise, store question : if i have rev1 of v1.0 of my snap that's have been automatically rejected, i need to upload v1.0.1(for ex) ?10:51
attenteis there a reason the names of the binaries in /snaps/bin don't match the ones in the snaps themselves?11:25
ogra_attente, got an example ?11:27
attenteogra_: ubuntu-calculator-app for example11:27
* ogra_ never used that 11:27
ogra_i meant an example of the binary names :)11:28
ogra_vs what you expect11:28
attenteogra_: maybe a better question is what is the canonical way of launching a snap11:28
ogra_packagename.binary-name11:28
ogra_i.e. in my nethack snap i defined "run" ... to start my game i now need to exec nethack.run11:28
attenteogra_: so in ubuntu-calculator-app, i think the Exec line of the desktop file is incorrect possibly11:29
ogra_well, i thought dpm has tested that11:29
attenteogra_: it just has the line 'Exec=ubuntu-calculator-app11:30
ogra_and there is no binary of that name in the snap ?11:31
ogra_(not sure if .desktop files are simply handled differently)11:31
ogra_i never used them :)11:31
attenteogra_: no, there's only /snaps/ubuntu-calculator-app.ubuntucoredev/current/usr/bin/ubuntu-calculator-app11:32
ogra_then thats fine i guess11:32
attenteogra_: but /snaps/*/current/usr/bin isn't in the user's path11:32
attenteonly /snaps/bin is11:32
attenteso it finds /snaps/bin/ubuntu-calculator-app.calculator instead11:33
ogra_well, i dont exactly know how ,.desktop files are operated in snappy ... but i guess it is only run inside the confined env ... where that path will be exported by the launcher11:33
* ogra_ assumes /snaps/bin/ubuntu-calculator-app.calculator is a script11:34
attenteyes11:34
ogra_then it will likely set the path correctly11:35
attenteogra_: so i should probably not be using the desktop file at all and just using the script in /snaps/bin directly?11:37
attentesince it seems like the desktop file refers to the binary in the snap itself and not /snaps/bin11:37
ogra_you should use the ,.desktop files from GUIs obviously ... :)11:38
ogra_no idea how the integration works with that though11:39
ogra_from commandline you shoudl always just run ubuntu-calculator-app.calculator11:39
tsimonq2ogra_: can you make metapackages with Snappy or does it have to all be in the package?11:58
tsimonq2whoops, snap, not package :)11:59
ogra_tsimonq2, whats the difference ?11:59
ogra_:)11:59
tsimonq2heheheh11:59
ogra_a snap *is* essentially a metapackage :)11:59
tsimonq2well11:59
tsimonq2oh I see11:59
ogra_containing all the bits and pieces11:59
tsimonq2so let's say I want to make the bob metapackage, and include like 20 deb dependencies, would that work smoothly?12:00
tsimonq2ooh, could I get it from upstream instead if I wanted to?12:00
ogra_sure, you would just define all the debs in your snapcraft.yaml and snapcraft would include them12:01
ogra_and yes, you could as well use the upstream source12:01
tsimonq2are the debs only from Ubuntu or can they be from Debian as well?12:01
ogra_for one or even for all your deps12:01
ogra_only from ubuntu12:01
tsimonq2oic12:01
tsimonq2is there a tutorial I could follow to do this?12:01
tsimonq2or is it really simple?12:02
ogra_there are examples in the snapcraft-examples package12:02
tsimonq2is that a deb package that I could install now, or a snappy package? :D12:02
ogra_snapcraft is a deb package ... note though that you need xenial if you build for xenial snappy12:02
tsimonq2I'm on Xenial :)12:03
ogra_good12:03
ogra_well, "sudo apt install snapcraft"12:03
ogra_;)12:03
* tsimonq2 does sudo apt update && sudo apt -y install snapcraft snapcraft-examples12:03
tsimonq2:)12:04
ogra_create a workdir ... create a snapcraft.yaml file ... run "snapcraft" in the workdir and if your yasml file is correct you get a working snap ;)12:04
ogra_*yaml12:04
tsimonq2where are the examples then? :)12:04
ogra_i guess in /usr/share7doc12:04
tsimonq2also, can I upload snaps to PPAs?12:05
ogra_dpkg -L snapcraft-examples will tell you as usual ;)12:05
ogra_no12:05
tsimonq2that would be an awesome feature12:05
ogra_you can use launchpad to build snaps from a snapcraft.aml in a bzr branch12:05
ogra_and you can upload them to the store12:05
ogra_snaps *are* PPAs ... like they are metapackages ;)12:06
tsimonq2oh okay :)12:06
tsimonq2so let's say I have my snap built and all ready to go, how do I test it?12:06
ogra_in xenial you shoudl actually be able to install it directly on your system ...12:07
ogra_at least after release, not sure what bugs are still there12:07
tsimonq2it would replace packages and cause a big huge mess, I would prefer a VM :)12:07
ogra_alternatively you can run a complete snappy in kvm or on i.e. a raspberry pi12:07
ogra_no, it wouldnt cause any mess12:07
tsimonq2KVM? I thought there was a guide somewhere for that...12:08
ogra_snappy on classic installs is a feature of xenial :)12:08
ogra_installing a snap will "just work" ... snaps live in their own filesystem space, they wont mess up anything12:08
ogra_(it shoudl already work if your xenial system is up to date, ubuntu-snappy-cli should be installed)12:09
oparozogra_: What's the strategy when we need to ship a firewall, log rotater, etc. It's not really efficient to add it to every single snap part of a solution12:10
ogra_you would creater a library snap that all your other snaps (only yours though) can consume12:11
oparozogra_: Ah, good. Is there doc on that already?12:11
ogra_not sure12:11
tsimonq2ogra_: if I create a snappy package from upstream, will it automatically pull in updates from there?12:13
tsimonq2for example GitHub12:13
ogra_if you re-build it12:13
tsimonq2that seems a bit tedious :|12:15
tsimonq2is it easy to rebuild and reupload?12:15
ogra_if your snapcraft.yaml is correct it should be one click in launchpad12:16
ogra_(modulo build failures that the upstream changes introduce indeed ...)12:16
tsimonq2can you set up daily builds?12:16
ogra_i think launchpad receipes are supported ... not sure though12:17
tsimonq2I'll try later today :)12:18
tsimonq2then to update everything is sudo snappy update?12:18
ogra_sudo snap update12:19
ogra_the "snappy" command is dead12:19
tsimonq2why?12:19
tsimonq22 extra letters? :P12:19
ogra_dunno12:19
* ogra_ wasnt involved in that discussion :)12:20
tsimonq2alright, thank you ogra_, I'm off for a few hours, have a nice day. :)12:21
ogra_you too12:21
dholbachkyrofa, sergiusens: can you maybe advise on https://bugs.launchpad.net/click-reviewers-tools/+bug/1569226?12:36
ubottuLaunchpad bug 1569226 in click-reviewers-tools (Ubuntu) "click-review crashed with KeyError in check_apps_plugs_mapped_oldsecurity(): 'plugs'" [Medium,New]12:36
dholbachI don't know what the right fields and stuff are12:36
kyrofaHey dholbach!12:36
dholbachhey hey - how are things?12:36
sergiusensmorning12:37
ogra_eeek !12:53
ogra_sergiusens, !12:53
ogra_sergiusens, all our initrds use lzma now ... seems the kernel plugin has no check about the compression method and just assumes gzip12:54
sergiusensogra_, well, not really; it said we only supported that for now12:58
sergiusenslzma or lzma2 or xz?12:58
ogra_bug 156933712:58
ubottubug 1569337 in Snapcraft "snapcraft kernel plugin needs to check compression method of the initrd before trying to unpack it" [Undecided,New] https://launchpad.net/bugs/156933712:59
sergiusensogra_, thanks12:59
sergiusensogra_, probably an EOW thing12:59
ogra_yeah13:00
kyrofadholbach, things are frantic, haha13:07
kyrofadholbach, any chance that bug relates to the removal of old-security from snappy?13:07
dholbachkyrofa, I don't know which fields are current and which old fiels should still be allowed13:07
dholbachbut this breaks store reviews right now13:07
kyrofadholbach, join the club, things are changing fast :(13:08
kyrofadholbach, I believe old-security is gone and one needs to use the actual interface names now. But then I heard maybe that would come back13:08
kyrofadholbach, zyga would know more on that one13:08
zygaold-security is gone now13:09
zygabut we'll probably bring it back later, for now use real interfaces13:09
zygathey are listed in docs/interfaces.md13:09
zyganote that only auto-connecting interfaces really work as connection/disconnection commands are not wired to the backend yet13:09
zygathat will land today though so all the interfaces will be fully functional13:09
dholbachok... I guess I need to talk to jdstrand to find out what policy should be in this case... reject snaps which don't have proper interface names?13:09
kyrofadholbach, I guess so. I thought we had old-security as a migration plan, but I guess not?13:10
zygadholbach: snappy validates that internally13:11
zygakyrofa: it's all crazy :)13:11
kyrofazyga, I've noticed :)13:12
zygakyrofa: I want to add old-security back but that's not top-priority13:12
zygatop priority is fully functional REST api for managing interfaces13:12
kyrofazyga, I'm not sure it's necessary since everyone is broken now-- everyone will have to move to interfaces if they want stuff to work13:12
zygakyrofa: true13:12
zygakyrofa: but some edge cases need old security13:13
dholbachzyga, right... but the store needs to check it too13:13
zygasergiusens: can you release snapcraft that allows something other than old security today?13:13
zygadholbach: no13:13
zygadholbach: the store doesn't have to check it in any way13:13
oparozzyga there is no docs/interfaces.md in snapcraft13:13
zygadholbach: we'll add some checks via assertions but that's not today13:13
kyrofaoparoz, that's snappy13:13
sergiusensppisati, fwiw ogra_ logged a bug about the gzip issue13:13
zygaoparoz: sorry, look at snappy13:13
zygaoparoz: github.com/ubuntu-core/snappy/ docs/interfaces.md13:13
dholbachzyga, what if somebody uploads an old style app to the store13:13
sergiusenszyga, today? wow13:14
oparozThanks guys13:14
dholbachzyga, with no plugs defined13:14
zygadholbach: nothing13:14
zygadholbach: it will just run without any extra permissions, fully confined13:14
zygadholbach: snappy ignores unknown or invalid interfaces13:14
sergiusenszyga, we can, but we need old-security replacements for our examples13:14
zygasergiusens: I'd be fine with a half-broken release that lets people snapcraft non-broken snaps without old-security13:15
zygasergiusens: I suspect that most of those can just use normal interfaces but I didn't check13:15
sergiusenszyga, right, but we won't be able to know if snapcraft itself is working13:15
sergiusensor doing the right thing13:15
zygasergiusens: ideas welcome, right now you cannot build a snap that uses unity7 + x11 plugs13:16
zygasergiusens: I don't know snapcraft as well as you do, what would be the best way forward?13:16
sergiusenszyga, 1st; is there an os image in edge that works already?13:16
zygasergiusens: mvo built a package, the image is soon after that I bet13:17
sergiusenszyga, 2nd; do we have replacements for everything in `examples` wrt interfaces?13:17
zygasergiusens: I don't know what's in all the examples13:17
sergiusenszyga, 3rd; is old-security coming back or gone forever?13:17
sergiusensif it is gone, I'd rather have it gone forever13:18
zygasergiusens: 3 not sure, I can implement it but if powers that be demand it gone, what can we do?13:18
zygasergiusens: https://github.com/ubuntu-core/snappy/blob/master/docs/interfaces.md13:18
zygasergiusens: that is what we have today13:18
dholbachthanks... I'll just catch the error if no plug is defined13:18
dholbachthink that should be fine13:18
sergiusenszyga, create a snapcraft bug and I'll work on a branch; kyrofa heads up, we might release today13:19
kyrofasergiusens, thanks for the heads up13:20
kyrofasergiusens, though that probably means we need to do the examples as well13:20
zygasergiusens: sure; I'll open the bug in a moment13:27
attentehas something changed with the /run/snapd.socket api in the past day? i seem to only be getting 404 errors now13:28
zygaattente: many things were removed13:28
zygaattente: look at docs/rest.md (it has been updated AFAIK)13:28
=== chihchun is now known as chihchun_afk
zygaattente: which endpoints were you calling13:29
attentezyga: for example GET /2.0/snaps?sources=local13:30
attentezyga: docs/rest.md was removed too?13:31
attentezyga: ok, i'll take a look, thanks for the pointer13:32
zygaattente: no, it is still there?13:33
zygahttps://github.com/ubuntu-core/snappy/blob/master/docs/rest.md#v2snaps13:33
zygaattente: the prefix changed so you just have to update your client code13:33
attentezyga: ok, thanks. is the api going to be frozen at some point?13:35
zygaattente: when we release 1.0 I suspect13:35
zygaattente: though there will be less changes made now13:35
zygaattente: you can expect additions rather than changes13:35
jdstrandzyga: fyi, there are checks that the review tools need to do in the absence of assertions13:39
zygajdstrand: perhaps but I don't know what the plan is for that13:40
jdstrandzyga: I will handle the review tools13:40
jdstranddholbach: ^13:40
zygajdstrand: thank you!13:40
jdstrandbut there is no point in changing them atm until we have a working system13:41
jdstrandso, perhaps later this week13:41
zygajdstrand: do you know where we are today?13:41
jdstrandnot yet13:41
zygajdstrand: we got auto-connection feature landed, also for x11 and unity713:42
zygajdstrand: I'm working on some tweaks to connection control so that it persistst properly and that users are in control13:42
jdstrandzyga: x and unity7 are autoconnecting? can you explain what you mean by autocontrol?13:43
zygajdstrand: x11 (it was renamed)13:43
zygajdstrand: it gets connected to os snap on install13:43
jdstrandI understand what that means13:43
jdstrandthat also is a security hole13:44
jdstrandso we had the idea that there would be some sort of prompting13:44
jdstrandwhat happened with that?13:44
jdstrandbeuno: ^ ?13:44
zygajdstrand: there's no propmpting for that as there's no UI for install13:46
jdstrandI didn't mean an install prompt13:46
zygajdstrand: oh?13:46
jdstrandwhat I suggested was no autoconnect on install. then on first use, prompt the user somehow, letting them know the app wants privileged access to the user's session. the user says yes or no. if yes, there is an interface connect13:47
jdstrandthis prompt could be done via zenity, or anything really. mvo was in agreement on the approach13:47
jdstrandthere is a card on the idea, but I think beuno was shopping it around13:47
zygajdstrand: that's not realistic today13:48
zygajdstrand: we'll improve it with SRUs13:48
jdstrandmaybe not 'today' but interface connect is surely something for 16.04 release?13:48
zygajdstrand: yes13:48
mvojdstrand: quick question, we don't need ubuntu-core-security-{apparmor,seccomp,utils} anymore, correct?13:49
jdstrandso it should be realistic for 16.04 release13:49
jdstrandmvo: correct13:49
mvota13:49
jdstrandmvo: they should be removed from the archive too13:49
zygajdstrand: there will be no prompting for 16.0413:49
zygajdstrand: we can explore that after the release13:49
jdstrandzyga: why? who decided that? we are making announcements for this. where was this discussion?13:50
zygajdstrand, mvo: FYI we only run apparmor_parser but templates seem to refer to included files13:50
jdstrandzyga: the prompt can be as simple as updated the shell script13:50
jdstrandzyga: the includes don't come from ubuntu-core-security. they come from apparmor13:50
zygajdstrand: there are no APIs to implement the prompt13:50
jdstrandwe don't need it13:50
zygajdstrand: it's not just the UI13:51
jdstrandall we need is interface connect foo unity713:51
jdstrandor whatever the api is13:51
zygajdstrand: yes and that's not ready yet13:52
zygajdstrand: I'm trying to get it ready13:52
jdstrandzyga: but you said a moment ago it would be for 16.04 release13:52
zygajdstrand: auto connect is internal13:52
zygajdstrand: connect/disconnect requires an API13:52
jdstrandI don't care about *today* so much as when someone tries to use this thing on release day, people see the choice13:52
zygajdstrand: I'm trying to implement the API so that we can use it but it's not working as of this second yet13:53
zygajdstrand: the goal is to get it working today (API) but I don't think we can manage a GUI at this stage13:53
jdstranda gui is a zenity prompt13:59
jdstrandsnappy sees if x11 or unity7 interface (or home) is requested. oh, it is, insert this zenity prompt in the shell script13:59
zygajdstrand: can you implement one quickly? perhaps we can manage that?13:59
zygajdstrand: note that there's a REST API there14:00
jdstrandI need beuno to confirm it is ok14:00
zygajdstrand: so it's not a simple sequential client14:00
jdstrandzyga: where is the rest api?14:00
zygajdstrand: you'd have to install the snap, inspect available plugs14:00
zygajdstrand: deamon/api.go, the docs are in docs/rest.md14:00
zygajdstrand: then see if your new snap has x11 or unity7 plug14:01
zygajdstrand: and then offer a quick prompt (even without zenity, just a simple stdin prompt)14:01
zygajdstrand: and then call .Connect()14:01
zygajdstrand: note that there are GO bindings for the api (look at client/interfaces.go)14:01
zygajdstrand: I think someone could look at doing that but I need to finish connect/disconnect first14:01
zygajdstrand: also note that you need to work with cmd/snap/cmd_ops.go (AFAIR) because that's where install is14:04
zygajdstrand, mvo: can tell you more14:04
jdstrandwhen I spoke to mvo before he said that he thought this was all managable. that was a few weeks ago14:04
zygajdstrand: I thnk that explains it14:05
zygajdstrand: there were many things that got us busy14:05
jdstrandsure, I understand14:06
jdstrandI'm not upset about that or anything-- I just want to make sure this is handled in some way14:06
zygajdstrand: I understand, I'm sorry if I seem defensive :-) I'm just tired and we're all racing to get bits in place14:08
jdstrandno worries14:08
jdstrandmvo: can you ping me when you adjust the seeds for ubuntu-core-security?14:28
dholbachjdstrand, thanks14:34
beunojdstrand, I'll follow up on the notify-when-x14:34
jdstrandthanks!14:34
elopioogra_: any news about the new archs? I'm digging all the emails from yesterday but haven't seen anything.15:06
ogra_elopio, well i got no mails (probably olli forgot to forward something) ... but i have the rootfs snaps building (sadly they arent published yet because i still need to teach livecd-rootfs to not try to build kernel snaps for them)15:08
ogra_https://launchpad.net/~ubuntu-cdimage/+livefs/ubuntu/xenial/ubuntu-core-system-image/ both fail right after creating the ubuntu-core os snap15:09
elopiothe image PPA is still not building for s390x. mvo: can you enable that? or do you know who can?15:17
elopionoise][: can you help me with spi? I'm getting 401 checking for the results of the run I launched. https://paste.ubuntu.com/15793410/15:21
mvocjwatson: who should I talk to about s390x builds for https://launchpad.net/~snappy-dev/+archive/ubuntu/image/+packages ? is that something you could enable or do I need to ask IS? or can I do it myself if I find the right knob :) ?15:28
mvoelopio: I ask someone who knows everything ;)15:28
mvoelopio: I think I can not do it myself15:29
noise][elopio: hey.. that's odd15:29
cjwatsonyou can't do it yourself because s390x has no virtualised builders yet, so it lacks sandboxing15:29
cjwatsonbut I see that PPA is already devirtualised15:30
cjwatsonmvo: done15:30
* ogra_ cureses about livecd-rootfs code ....15:30
cjwatsonmvo: new builds aren't auto-created, but if you want to create new builds of an existing source in that PPA without reuploading, you can copy the source over itself including existing binaries and that has the side-effect of creating missing builds15:31
ogra_feels like i painted myself into a corner, skipping kernel creation for one arch is really hard now :/15:31
cjwatsonmvo: (also obviously only xenial, since s390x didn't exist before that)15:31
mvocjwatson: \o/15:32
mvocjwatson: thanks a bunch15:32
cjwatsonnp15:34
elopiocjwatson: mvo: thanks!15:34
ppisatisergiusens: saw it15:35
elopionoise][: here in the dragonboard section is all I did, in case it helps: http://pad.ubuntu.com/spi15:35
ppisatisergiusens: i'm a bit nervous about the 'snappy login' step, i bet there's no way to skip it15:35
zygappisati: snap login, snappy is gone15:36
ppisatizyga: snapcraft login i mean15:36
zygaah15:36
ppisatizyga: what does it do btw?15:36
zygappisati: I'm not quite sure15:36
ppisatisergiusens: ^15:36
* ogra_ guesses it logs you in15:41
ogra_:P15:42
ppisatiogra_: ok, but why it needs to log me into UU build a kernel/snap?15:43
ogra_heh, no idea ... i was just stating the obvious :P15:43
* ogra_ doesnt even know where to it logs you in15:44
ppisatiah ok15:44
ogra_i assume "snapcraft snap" should still work without login though15:45
ogra_(it does on the livefs builders ... and we wouldnt be able to use any kind of login there(15:45
=== shuduo is now known as shuduo-afk
noise][elopio: can you try to query the results w/o the test opp query param?15:46
noise][just https://spi.canonical.com/products/$PRODUCT_ID/tests/events15:47
ppisatiogra_: nope, it requires the login15:47
ppisatihttp://pastebin.ubuntu.com/15794106/15:47
ppisatibtw i thought 'snap' was the default target15:47
ogra_ppisati, well, it has to be skippable, else we cant build images15:47
ogra_zyga, sergiusens ^^^15:48
ogra_there is no way we can add credentials on the image builders, nor can they see the store15:48
sergiusensogra_, ppisati the login is to download the os.snap. I wasn't sure it was possible to download without being logged in15:49
ogra_sergiusens, just grab it from cdimage :P15:49
* zyga doesn't know what login is for in the long term, cannot comment15:49
ppisatisergiusens: does it store my credentials somewhere?15:50
sergiusensppisati, yes, in the form of oauth; not the password15:51
sergiusensppisati, ~/.config/snapcraft/snapcraft.cfg15:52
ogra_sergiusens, how would that work for i.e. a launchpad based build ?15:52
sergiusensit is only the kernel plugin15:52
ogra_where you simply dont have these credentials15:52
sergiusensogra_, ^15:52
sergiusensonly the kernel plugin15:53
sergiusenskernel plugin is experimental15:53
ogra_sure, but what stops me from using the kernel plugin ibn an LP build15:53
sergiusensnothing is set on stone15:53
sergiusensand as I said multiple multiple multiple multiple times15:53
sergiusensthis download thing goes away as soon as snappy does the initrd dance the correct way15:53
sergiusensogra_, ^ what stops you is the fact that you need to login15:54
ogra_(i doubt that will happen before next LTS :P )15:54
* ogra_ is obviously very pessimistic wrt the SRU process of massive architectural changes :)15:54
elopionoise][: same result.15:58
noise][elopio: hmm, and is that repeatable? starting w/a new product?16:00
elopionoise][: let me try that.16:26
elopionoise][: I can't create it anymore. Now I get 401 on the first step.16:33
elopiooh, wait, no there was a typo.16:33
elopiogoing on...16:33
=== qengho_ is now known as qengho
sergiusenszyga, I logged https://bugs.launchpad.net/snapcraft/+bug/1569452 since I am impatient16:35
ubottuLaunchpad bug 1569452 in Snapcraft "Update snapcraft to use new interfaces" [Undecided,New]16:35
elopionoise][: reproduced again, I created everything from scratch, made a POST on the images endpoint and got 201. But when I query the tests/events, I get 401.16:36
noise][elopio: ok, I don't have time to dig at the moment, but can try to repro later this afternoon16:46
sergiusensjdstrand, zyga can you check if I made as "an accurate as possible" translation of things here https://github.com/ubuntu-core/snapcraft/pull/441/files17:16
sergiusensI am aware some things will just not work17:16
sergiusenslike busybox which would need a read_path of / at least17:17
zygalooking17:17
sergiusensnot sure what `network-management` was for, I hope it was covered by `network-bind` and also not sure I need `network` if I have `network-bind`17:18
kyrofasergiusens, I was just about to ask that question :P17:18
sergiusensnor if `network-service` is covered in any way17:18
kyrofanetwork-service is now network-bind, I believe17:19
sergiusensmy gut says it would17:19
sergiusensor anything needing network-bind would need network17:19
* sergiusens is sad for removing all those tests though17:20
zygasergiusens: network-management is more like "I want to be network-manager"17:20
zyganetwork == client; network-bind == server)17:20
sergiusenszyga, ok, so for something that listens on a port and also talks to the network I need both?17:21
zygait won't hurt17:21
zygabut I suspect network-bind gives you network17:21
zygajdstrand: can clarify17:21
sergiusenszyga, I'll wait for your first pass17:21
zygawe should add those kind of FAQs to the docs :/17:21
zygasergiusens: I reviewed it already17:21
sergiusensoh; yeah, the docs aren't really all that complete ;-)17:22
sergiusenszyga, the mosquitto one, elopio can you look ^17:23
sergiusenszyga, if you look at how the listener plug is defined it seems to indicate it needs `network` and not `network-bind`17:23
zygasergiusens: sorry, I'm so drained today I feel like I'm about to fall asleep right here;17:23
zygamy comments were based purely on the names17:24
jdstrandsergiusens: network-management was network-admin on 15.0417:29
jdstrandsergiusens: it is network-control now17:29
jdstrandsergiusens: network-bind and network client are not the same. network-bind currently gives you all of network, but there is no guarantee that will always be the case17:30
kyrofajdstrand, good to know, thank you17:35
sergiusensjdstrand, ok, so my shout client I should add network and network-bind (basically a web irc bouncer)17:36
sergiusensshout app (not client)17:36
jdstrandsergiusens: gopaste technically would need network-control for a 1-1 mapping, but I think gopaste is suffereing from bug #146572417:37
ubottubug 1465724 in Snappy "net_admin apparmor denial when using Go" [High,Confirmed] https://launchpad.net/bugs/146572417:37
jdstrandsergiusens: I would, yes17:38
sergiusensjdstrand, yeah, most likely the case for gopaste17:38
jdstrandsergiusens: it is safe to leave network-control out for gopaste and just live with the denial until we can fix the kernel17:38
sergiusensjdstrand, what is your suggestion for it?17:38
sergiusensok17:38
jdstrandtyhicks: iirc, that is the current recommendation, right? ^17:38
tyhicksjdstrand: yes17:38
tyhicksI think I know the proper fix for the denial now17:39
tyhicksI'll have another look at it soon17:40
jdstrandtyhicks: oh nice :)17:40
dholbachelopio, thanks for the update on bug 1569478 and everything18:02
ubottubug 1569478 in Snappy "integration tests fail on armhf and s390x" [Undecided,New] https://launchpad.net/bugs/156947818:02
attenteare there any plans to add some api for extracting the metadata out of a snap that isn't yet installed?18:11
kyrofaattente, a store API already exists. Are you talking about a local snap, though?18:12
attentekyrofa: yeah, for sideloading18:12
kyrofaattente, not that I know of, but feel free to log a wishlist bug for it18:13
attentekyrofa: sure, thanks18:13
ogra_You can mount the snap or use unsquashfs to get the snap.yaml18:13
ogra_(it is just a squashfs file after all)18:15
kyrofaogra_, hardly an API, but yeah18:15
kyrofaattente, if you're just scripting something, that may be a good direction18:16
ogra_Well, yaml iis kind of an api :P18:16
kyrofaogra_, hahaha18:16
kyrofaattente, may I specifically call your attention to `unsquashfs -e`18:17
kyrofaattente, even more specifically: `unsquashfs -e meta/snap.yaml`18:18
attenteyeah, i was just wondering if maybe there should be some friendlier tooling for it18:18
attentekyrofa: cool18:18
kyrofaattente, can you explain your use-case?18:18
kyrofaattente, though I guess you should do that in the bug anyway-- no need to make you repeat yourself :)18:18
attenteheh18:19
oparozIs it possible to merge unsquashed snaps and resquash the result?18:36
beunooparoz, you'd have to unpack them, and re-pack them18:37
beunowhat are you trying to do?18:37
oparozbeuno build parts on different machines18:37
sergiusenszyga, seems even after changing to interfaces I still have failing tests18:47
jdstrandmeh, the security policy is messed up for the path renames21:07
jdstrand$ focuswriter.run21:07
jdstrandmkdir: cannot create directory ‘/home/jamie/snap/focuswriter’: Permission denied21:07
* jdstrand takes a look21:07
tsimonq2can I make a snap part that just consists of a package from the Ubuntu archives?21:07
tsimonq2or does it *have* to be from source?21:07
jdstrandoh no, that wasn't hit. I hit the old ~/snap owned by root bug21:11
jdstrandsigh21:11
jdstrandbut the path rename broke the apparmor system unit21:11
jdstrandtyhicks: hey, have you uploaded already?21:12
jdstrandwe need better integration tests21:12
tyhicksjdstrand: I haven't yet21:13
jdstrandtyhicks: ok, can you add a super-quick change? just need to update a path. let me get you a bug and details21:14
tyhicksjdstrand: yes, this is good timing21:14
jdstrandtyhicks: https://bugs.launchpad.net/snappy/+bug/156957321:19
ubottuLaunchpad bug 1569573 in apparmor (Ubuntu) "recent snapd path renames causes apparmor to not load profiles on boot" [Critical,Triaged]21:19
oparozIs SNAP_FULLNAME going away?22:16
tyhicksjdstrand: uploaded23:01
sergiusensjdstrand, tyhicks is that also the reaosn for what I pastebined not working?23:48
sergiusens<sergiusens> zyga, jdstrand it seems that with these new interfaces I can't create files in SNAP_DATA http://paste.ubuntu.com/15800694/23:48
sergiusensthat just in case ^23:48
zygasergiusens: thanks for noticing this. I don't recall us removing this explicitly. I will look at the apparmor template we use23:49
zygaI see23:50
zygalooks like another victinm of moving files around23:50
zygasergiusens: https://github.com/ubuntu-core/snappy/pull/91023:54
zygaafter the dust settles we should use ... variables for that :P23:55

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