=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
=== zyga is now known as zyga_
oparozI have a question regarding: https://developer.ubuntu.com/en/snappy/build-apps/your-first-snap/08:25
oparozIt says that the snap will only work on Core, so how do we pick a target since snaps can also be created for the desktop?08:26
davidcalleoparoz: actually, the snappy is running on core on the desktop, so yes, it should work as long as the snap targets the right arch08:27
davidcalles/the snappy/the snap :)08:28
oparozThanks davidcalle, so it's just a problem with the doc then. I can install snapcraft on a desktop, create a multi-arch snap and deploy on desktop,server and core without any issue as long as they all use the same Snappy engine (16.04)08:29
zyga_good morning08:29
oparozDo we report doc problems on LP, in the Snapcraft project?08:31
davidcalleoparoz: yes, snapcraft doc is being updated to match the latest state of things. Examples in here (https://developer.ubuntu.com/en/desktop/) are the most recent (less detailled, though).08:31
davidcalleoparoz: snapcraft project, yes. https://bugs.launchpad.net/snapcraft08:31
oparozThank you davidcalle08:32
=== zyga_ is now known as zyga
slvnhello ... a few questions : 1) I have a few .click (arm, native binary) package for ubuntu phones, does it make sense to upgrade to snappy package ?09:00
zygaslvn: what kind of app is that?09:02
zygaslvn: I guess right now the answer is "if that snap would run on desktops"09:02
slvnzyga, games based on libSDL209:02
zygaslvn: sure, you can target the desktop with your snap09:03
slvnthe games could also run on desktop, provided I compiled them for desktop09:03
zygaslvn: and inevitably the phone will adopt snaps at some point so you will already know how things work there09:03
slvnalso I tried to install my .click package on my destkop and it refused to be installed (cannot be install on X ...)09:04
slvnso I should look how to provide a snap package that contains both arm and x86 libs for arm and desktop09:05
slvnWill snapy packages run *confined* like .click on phones ?09:06
=== chihchun is now known as chihchun_afk
=== JamesTait is now known as Guest6158
=== Guest6158 is now known as JamesTait
davidcalleslvn: to some extent, yes. Eg. snaps on the desktop can be granted access to $HOME09:14
davidcalleslvn: note that the phone switch to snap will take time.09:15
=== chihchun_afk is now known as chihchun
=== seb128_ is now known as seb128
slvndavidcalle, zyga, thanks for the answers, I will have a look on snapy !09:36
seb128is there an "apt-cache show" equivalent in the snap world?09:37
seb128snap list lists names, but that's not very useful to figure out what the snaps are exactly09:38
zygaseb128: I don't think there is one09:40
zygaseb128: as a hack, you can look at /snap/$name/current/meta/snap.yaml09:41
zygaseb128: but that's just a portion of the information we have about each snap (2nd part comes from the store)09:41
seb128zyga, I'm not hacking09:41
seb128just trying to use it as an user09:41
zygaseb128: I know that Chipaca is working on a richer REST API09:41
seb128k, I'm also trying if we have those info available so we integrate them with the desktop frontend09:42
zygaseb128: you mean in gnome-software?09:46
zygaseb128: AFAIR gnome software should use the rest api09:46
zygaseb128: there we can expose many details easily09:46
seb128I was just trying to see what is currently exposed09:46
seb128and I though I would start as an user with using the command line to see what is displayed there09:46
seb128but I hit that wall ;-)09:47
zygayep, I understand09:48
zygaseb128: perhaps this can be of some use: https://github.com/ubuntu-core/snappy/blob/master/client/packages.go09:48
zygathis is the client side rest API09:48
plarspopey: you still have these problems like we discussed yesterday with building images with udf, right? fgimenez just sent me a rebuilt binary, but it didn't seem to help, and then I found myself wondering if there was some known-fix that I just didn't hear about yet09:48
zygait's in go but you can make the query in any language09:48
popeyplars: i haven't tested since it made my laptop unusable09:49
seb128zyga, is there an easy way to see what is currently available? I'm looking after the license info09:49
zygaseb128: available as in "can install"?09:49
plarspopey: ack, thanks09:49
popeyhappy to test to confirm any thing you might need plars09:50
seb128zyga, no, I was wonder if there is a way to query the license from a snap through the rest api atm09:50
* zyga looks09:50
plarspopey: I just wonder how it works for others09:50
popeyplars: i expect people don't run that commanbd09:50
zygaseb128: no, I don't think so, AFAIR some license code was removed before 16.04 as it wasn't ready09:50
seb128attente, ^09:50
zygaseb128, Chipaca: will know more for sure09:51
seb128zyga, thanks09:51
Chipacano i won't09:51
* Chipaca reads09:51
seb128willcooke, do we maintain a list of things we would need? until ^ is resolved g-s is going to show snaps with a "that software is non free" banner09:51
seb128Chipaca, can we query the license of a snap through the rest api?09:52
Chipacaseb128: nope09:52
Chipacanot sure why not though09:52
Chipacagimme a sec09:52
Chipacaseb128: so, no. But it could be added easily.09:53
seb128Chipaca, we have https://bugs.launchpad.net/ubuntu/+source/gnome-software/+bug/1555567 could you comment/triage the snappy/store side?09:54
ubottuLaunchpad bug 1555567 in gnome-software (Ubuntu) "License information from the store not being used" [Medium,Triaged]09:54
willcookeseb128, we need to get it on to the backlog, so we attend the stakeholders meeting and request it.  (wow! such manager speak)09:54
seb128willcooke, :-)09:54
seb128willcooke, ^ bug reference there09:54
willcookeseb128, so I think we should log bugs and tag them09:54
willcookeah nice09:55
willcookeI'll think up a suitable tag name09:55
Chipacaseb128: done09:57
seb128Chipaca, willcooke, thanks09:57
Chipacawillcooke: our bad, wrt licensing09:57
Chipacaas long as we're not talking of *accepting a license prompt*, shipping the info should be doable rsn09:57
Chipacaif it's urgent i mean :-D09:57
Chipacathe interactive license thing is for in about a month? wild-arsed guess09:58
seb128yeah, no need of interaction09:59
willcookejust spoke to seb128 about this - this issue is that there is a fairly large and unsightly banner in g-s which nags "Freedom hater"09:59
willcookeso it would be /nice/ to have that fixed09:59
willcookefor the first release of the snappy enabled g-s09:59
=== chihchun is now known as chihchun_afk
stevebiscuithttps://code.launchpad.net/~stevenwilkin/webdm/snappy-2-0/+merge/292902 # gets WebDM building against recent Snappy10:37
=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
slvngot some error when doing "snappy -v" : "snappy version 0.3.7" .... "(snappy:16290): GLib-GObject-CRITICAL **: g_object_set: assertion 'G_IS_OBJECT (object)' failed.No media set. '10:57
slvnsorry stupid question, don't know which tutorial makes me install that :)11:10
Chipacaslvn: that's probably not this snappy11:14
flexiondotorgIf I create a snap that uses a reserved interface in a snap, such as unity7, is it subject to manual review when submitted to the stire.11:15
slvnChipaca, yes, I figured out :) (though this "snappy" fails anyway).  Just build the hello-world opencv snap :)11:16
slvnIn addition to "snappy", the package "snap" is also confusing .. one should install "snapd"11:22
jamiebennettflexiondotorg, thats the idea, yes11:51
kyrofaGood morning11:56
dpmhm... it seems somehow 'ubuntu-core' got uninstalled from my system?11:58
dpmIs this a known issue, shall I just manually reinstall?11:58
dpmon a 16.04 desktop, that is11:58
dpmflexiondotorg, although I believe right now only 'home' triggers manual review in the store. 'opengl' will trigger manual review too until the store starts recognizing it12:00
dpmso somehow 'snap list' does not list 'ubuntu-core' for me, yet if I try to install it, snap install tells me it's already installed12:02
dpmand 'snap interfaces' does not list any slots available12:03
dpmubuntu-core cannot be removed, either12:03
flexiondotorgdpm, jamiebennett Is there a 'mir' interface as an alternative to x11 and unity?12:05
flexiondotorgIf so, is it confined?12:05
dpmflexiondotorg, these are the interfaces currently available: https://github.com/ubuntu-core/snappy/blob/master/docs/interfaces.md, but there is ongoing discussion about interfaces in general12:07
flexiondotorgdpm, Thanks. Understood. I thought I'd seen mention of a 'mir' interface, but I must have been mistaken.12:09
dpmflexiondotorg, perhaps you've seen it in a discussion here on the mailing list, but I don't think it's available12:09
flexiondotorgMaybe, I thought I'd seen something.12:10
oparozflexiondotorg, it will probably be available once snaps replace clicks12:10
=== mcphail_ is now known as mcphail
josepht'snap refresh <snap>' gives me a wrapper error.  I have to 'snap remove <snap>' twice to remove both revisions and then 'snap install <snap>' to get it working again.12:40
kyrofajosepht, any chance you had an app running when you tried to remove?12:45
josephtkyrofa: no12:45
sergiusensjosepht kyrofa this happens when sideloading, I hope the fix lands soon12:52
sergiusenskyrofa I have gulp working btw! EOD I might get vscode working; then I'll try atom but we need to support grunt12:53
josephtsergiusens: ah, thanks.12:53
kyrofasergiusens, yeah make sure you create bugs so we don't forget to upstream what we're working on!12:54
sergiusensstevebiscuit as a user, would you prefer using a nodejs snapcraft plugin with a possibility to select npm, gulp or grunt as a build driver or individual plugins for each? So we'd have a nodejs plugin (which already exists), then a gulp  plugin and a grunt one?12:55
sergiusenskyrofa upstream? as in snapcraft?12:55
sergiusenskyrofa the bug report depends on stevebiscuit's output ;-)12:55
kyrofasergiusens, haha, yeah12:55
* sergiusens puts some weight on someone else12:55
kyrofasergiusens, same question to you regarding qmake for qt4 versus 512:56
sergiusenskyrofa qmake afaik is tied to Qt versions, is it not?13:07
kyrofasergiusens, sort of. There are different qmake packages for each version, but the qt5 one can call the qt4 one13:08
kyrofasergiusens, but maybe it would be more clear to keep them separate13:08
sergiusenskyrofa yeah, I'd hop on to #ubuntu-touch and ask those guys ;-)13:11
kyrofasergiusens, alright13:11
stevebiscuitjdstrand: http://pastebin.ubuntu.com/16062943/ # AppArmor is giving a denial when starting WebDM, I think similar has already been reported?13:13
stevebiscuitsergiusens: I'd be tempted to see which toolchain the JS community is favouring, build a solution based on that and wait to see if there's demand for the other options13:15
bloodearnestjdstrand, hi there! beuno sent me your way regards finding out more about a) cgroups/namespace used to confine snaps, and b) the kernel boot/rollback process, if you've any pointer to docs?13:15
ogra_i think jdstrand is out this week13:17
stevebiscuitogra_: cheers for the headsup13:17
ogra_stevebiscuit: try tyhicks13:18
stevebiscuittyhicks: http://pastebin.ubuntu.com/16062943/ # AppArmor is giving a denial when starting WebDM, I think similar has already been reported?13:19
stevebiscuitogra_: ta13:20
sergiusensstevebiscuit from what I am seeing, gulp and grunt are a waste of time13:20
sergiusensstevebiscuit used when the project target's windows as people think npm scripts need to be shell scripts13:21
jdstrandstevebiscuit: what version of webdm are you using?13:21
jdstrandogra_: I was only out yesterday13:21
sergiusensstevebiscuit http://blog.keithcirkel.co.uk/why-we-should-stop-using-grunt/13:22
stevebiscuitjdstrand: building from https://code.launchpad.net/webdm13:24
sergiusensstevebiscuit ok, I think I'm sold on separate plugins that inherit from the nodejs plugin13:24
sergiusensso I can deprecate the other ones in an easier fashion13:24
ogra_jdstrand: oops :)13:24
zygajdstrand: hey, how are you13:25
jdstrandbloodearnest: the documentation is all in flux. I believe that the community team (dpm) is handling updating that13:25
jdstrandzyga: hi, good. you?13:25
jdstrandbloodearnest: if you had specific questions, I might be able to help with some13:25
zygajdstrand: good :-)13:25
bloodearnestjdstrand, ah, ok13:25
jdstrandstevebiscuit: I don't see snap.yaml or snapcraft.yaml in that tree. what are you using for that?13:25
beunojdstrand, wasn't there a whitepaper something or another?13:26
jdstrandbeuno: there is a whitepaper for 15.04, yes, but it wasn't ever published (I asked to have it done but don't think that ever happened)13:26
jdstrandbeuno: the whitepaper for 16 is done except it'll need updates for GA13:27
bloodearnestjdstrand, ok, so I've seen various mentions of cgroups and namespace usage for snaps, would like to know what cgroups/namespaces are used, and how they interact with plugs/slots13:27
zygajdstrand: I read that on developer.u.c somewhere (the whitepaper)13:27
beunobloodearnest, there's also apparmor13:27
stevebiscuitjdstrand: http://bazaar.launchpad.net/~snappy-dev/webdm/trunk/view/head:/pkg/meta/snap.yaml13:27
beunoto complete a buzzwork bingo13:27
bloodearnestbeuno, right, that part I understand13:27
zygabeuno, bloodearnest: and dbus and seccomp and udev tagging13:27
bloodearnestor at least, understand enough of13:27
roadmrat least there's no synergy!13:28
stevebiscuitjdstrand: it needs updating to use snapcraft and currenly has it's own build.sh13:28
jdstrandstevebiscuit: ah, I missed that. what are the perms of /snap/webdm/100001/snappyd?13:29
stevebiscuitsergiusens: so you're going to have something that can be installed via npm initially? the JS community seems intent or re-inventing *everything* in JS heh13:29
sergiusensstevebiscuit I am chatting from a snap that uses the nodejs plugin already ;-)13:30
sergiusensbut I ignored gulp and grunt; but turns out vscode uses gulp and atom uses grunt13:30
stevebiscuitjdstrand: 0755. It can be run fine standalone oddly13:30
jdstrandstevebiscuit: sorry, what is the ownership13:31
stevebiscuitjdstrand: it's owned by ubuntu:ubuntu13:32
jdstrandstevebiscuit: ok, that is your problem. how are you building your snap?13:32
jdstranddholbach: note, https://developer.ubuntu.com/en/snappy/guides/security/ is terribly out of date13:32
jdstranddpm: ^13:33
dholbachjdstrand, yes, we're on it13:33
dholbachAFAICT it's not more than 2 weeks out of date13:33
stevebiscuitjdstrand: it's still being built with sergiusens origional build.sh, I've not yet looked into what it needs to be moved to using snapcraft13:33
jdstranddholbach: zyga mentioned the 15.04 security whitepaper was on developer.u.c. I can't seem to find it. do you know where it is?13:33
dholbachjdstrand, no, I don't think it ever was on there13:33
jdstrandstevebiscuit: ok, let me look at that real quick13:33
ogra_you need to at least use snapcraft snap instead of snappy build13:34
zygathat was a while ago13:34
jdstrandstevebiscuit: right, build.sh is using 'snappy build $builddir'13:34
jdstrandstevebiscuit: I think if you change that to 'snapcraft snap $builddir' it should work13:35
stevebiscuitjdstrand: cool, I've give that a whirl13:35
jdstrandbloodearnest: I shared the link with you13:37
jdstrandbloodearnest: for 1613:38
jdstrandbloodearnest: note in general most things should be all settled, but there might be a few things that will change (nothing architecturally)13:38
jdstrandbloodearnest: but for the benefit of people in this channel, Ubuntu Core uses a combination of technologies to implement the sandbox. the heart is apparmor, but we also use seccomp, device cgroups, private /tmp and devpts newinstance13:39
bloodearnestjdstrand, great, thanks!13:40
jdstrandbloodearnest: the only namespace is the mount namespace for /tmp. we chose to implement the snadbox in this manner because it allows snaps to integrate more fully with the system and to better interact with each other. for people wanting full containers, people can use lxd or docker (though atm neither is available on 16, but will be before GA)13:41
thomas25I have a question about the configuration file13:41
thomas25How are they handle since the snaps are read only13:42
jdstrandtyhicks: fyi, in case you didn't notice, I answered stevebiscuit13:42
thomas25I tried the mosquitto snap but I can't edit the mosquitto.conf file.13:42
bloodearnestjdstrand, right, I assumed a fairly limited namespace usage, was just looking for details13:42
jdstrandthomas25: I'll answer since I happened to have seen the question. In general, the snap would be set up to copy the configuration file from $SNAP to $SNAP_DATA and have the daemon use the file in $SNAP_DATA so that the file is then readable. I don't know how mosquitto is packaged though13:43
ogra_thomas25: there used to be a "snap config" command that would use the conf as input and put it into the writable space ... that is temporarily gone and being re-worked13:44
thomas25ok thanks13:44
sergiusensthomas25 mosquitto from snapcraft sources?13:44
sergiusensthat is just an example, not meant to be configurable afaik13:45
jdstrandbloodearnest: if you want the implementation details of what we do with the devices cgroup, see http://bazaar.launchpad.net/~snappy-dev/ubuntu-core-launcher/trunk/view/head:/README13:45
sergiusensyou can add the smarts for it thought (like have it read from SNAP_DATA or SNAP_USER_DATA)13:45
sergiusensogra_ the examples are also a stress tester for snapcraft itself; not for snaps13:45
sergiusensogra_ I wish the core guys wrote more snaps as that would really make sure the system is good13:46
jdstrandI should have said "is then *writable*"13:46
ogra_they will come over time :)13:46
thomas25SNAP_DATA nad SNAP_USER_DATA are used to override the path of configuration file in snaps ?13:48
bloodearnestjdstrand, thanks! already skimming the whitepaper has answers most of my questions! :D13:48
kyrofathomas25, this question might help: https://askubuntu.com/questions/762354/where-can-ubuntu-snaps-write-data13:48
thomas25Or the snap just copy the configuration file in SNAP_DATA path ?13:49
ogra_thats up to you :)13:49
ogra_the snap only provides the path ... weather you copy it and make your service use it is up to you13:50
ogra_(you could ship an immutable config just in the readonly dir and not allow the user to alter it at all)13:50
thomas25In the snapcraft.yaml how to you say that this goes to SNAP_DATA and that in SNAP_USER_DATA ?13:53
thomas25Just using "copy" plugin ?13:54
ogra_no, you would write a wrapper that does the copying for you on service startup13:54
sergiusenskyrofa with great pleasure I deliver this bug to you https://bugs.launchpad.net/bugs/1575188 ;-)13:55
ubottuLaunchpad bug 1575188 in Snapcraft "Fix for bug #1572664 had broken my snap package build" [Undecided,New]13:55
kyrofasergiusens, yes I just saw that come in13:55
kyrofasergiusens, thanks :P13:56
thomas25ogra_: Thanks13:56
sergiusenskyrofa while the snapping might work, not sure it would ever be accepted by the store with so many dead symlinks13:56
kyrofasergiusens, yeah might be a question for beuno13:57
sergiusensjdstrand rather13:57
sergiusensI'm sure it won't13:57
kyrofasergiusens, so is this a wontfix, or should the review tools change?13:58
sergiusenskyrofa I'd ask him if the review tools passed without warnings in his previous working snaps13:58
kyrofasergiusens, alright :)13:59
=== zyga_ is now known as zyga
asacthe pi2 really has no reset button right? (or do i need glasses?)14:14
zygaasac: it has no buttons at all AFAIK14:15
zygalooking at mine now14:15
asacgood :)14:15
ogra_you can use a strobe light to reset it though14:15
zygawe need strobe interface14:15
asacmaybe i should use my ttl power and then i can use some tricks with /dev/ttyUSB to repower?14:16
zygaasac: whad are you trying to do?14:16
asacjust rebooting14:17
asacwithout having to fiddle myself through 5 layers of cables :)14:17
asachoping to not unplug some other board or ttl14:18
asacok installing hello-0world to look at hello-world.sh14:18
asacto see what is going on14:18
asacdoesnt work :)14:18
asacbah the pi2 kernel really always get a reset by peer14:19
asacso i cannot even try the latest14:19
asacerror: cannot perform the following tasks:14:19
asac- Download snap "canonical-pi2-linux" from channel "stable" (read tcp> read: connection reset by peer)14:19
ogra_you really dont make friends with that nordic guy today eh ?14:20
asache seems to not like the pi214:20
ogra_i just built aa set of images here14:20
asacdb worked fine14:20
ogra_no issues14:20
asacto update14:20
asacthink its only when you use it from the device14:20
asacnot during udf14:21
zygayep, same here14:21
zygaasac: are you on 3g?14:21
asacno stble landline14:21
asacsuper stable14:21
zyga4g :-) ?14:21
ogra_who would do a kernel update of his pi2 over 3G ?14:21
zygasuper proxy in betweeen to  read your mail?14:21
asacnot that i know14:21
asaci am direct14:21
* zyga looks from device14:21
asacits really something else14:22
asacbut the device goes through my desktop14:22
asace.g. connection share14:22
asacso could be due to that14:22
zygahmm, all up-todate14:22
asaclet me think14:22
zygamaybe, try direct14:22
asachave no lan14:22
asacbut i never had problems14:22
zygauntil today ;) AFAIR n-m doesn't use a proxy but I might be wrong14:23
asacno it doesnt14:23
asacits pure routing14:23
asactoo bad nothing works on image so i cannot download a big iso to see14:23
asaci am sure that would work :)14:23
asacwell  not so14:23
asaclet me get a wget14:23
asacoh i have chroots14:24
asaclet me try there14:24
ogra_snap install doesnt work either ?14:24
asacfor instance i had zero issues setting up those chroots14:24
zygaworks for me14:24
zygajust tried on pi214:24
zygaimage built yesterday14:24
asacok wget http://cdimage.ubuntu.com/ubuntu-core/releases/xenial/release/ubuntu-core-16.04-core-amd64.tar.gz is running14:25
asacwget finished14:25
* zyga goes to pick up kids from school14:25
asacno issues14:25
asacogra_: tells me its already installed14:26
ogra_asac, i mean a random snap14:26
* asac tries14:26
asacyeah nmap gives me same error14:27
asacxkcd-webserver works14:27
asacinstall mvos image, try refresh14:28
asacnot working14:28
oparozIs it asking for trouble if bundling python 2.7 in a snap?14:52
ogra_just use snapcrafts python plugin14:52
oparozogra_, I don't need to build anything in python, it's just that dtrx needs python 2.714:53
oparozogra_, So I'm just wondering what should be done for such scripts14:54
ogra_using the python plugin ;)14:54
ogra_it makes sure the right bits get bundled14:54
ogra_there was some trick to only get the interpreter ...14:54
ogra_source: .14:55
ogra_or some such14:55
oparozogra_, So you mean, not use the deb, but the source archive?14:55
ogra_no, define a snapcraft part that uses the python plugin14:55
ogra_in your snapcraft.yaml14:55
oparozand use the deb as a staging deb14:55
ogra_which deb ?14:55
ogra_yeah, with teh copy plugin or so14:56
oparozThe problem is how to run it on the other side14:56
oparozBut I'm guessing Snapcraft will include python2.714:56
oparozThus my question whether it's a good idea since python 3 is already present14:57
oparozBecause the script calls /usr/bin/python14:58
ogra_well, what does the deb depend on ?14:58
oparozDepends: python (>= 2.7), python (<< 2.8), bzip2, unzip, cpio, rpm, binutils, p7zip-full, cabextract, unshield, lzma, xz-utils14:58
ogra_smaller than 2.8 ...14:58
ogra_there you got your answer :)14:58
oparozOK, so Snapcraft will ship Python2.7 and create a wrapper which points /usr/bin/python to /usr/bin/python2.7 ?14:59
ogra_i dont know, thats a sergiusens or kyrofa question :)15:00
kyrofaogra_, oparoz if python is pulled in via stage-packages snapcraft won't do anything special15:00
ogra_but i think it replaces the shebang line in all py scripts it finds15:00
ogra_kyrofa: no, via the python plugin15:00
oparozkyrofa, Ah, then the script will fail in the snap15:01
kyrofaAh, but ogra_ is right-- the shebangs should be fixed via the stage-packages15:02
oparozAh, no, the default in core is 2.7, so I'm guessing v3 scripts in debs use /python315:03
oparozkyrofa, Ah, that's good if snapcraft fixes shebangs15:03
kyrofaoparoz, should be replaced with "#!/usr/bin/env python"15:04
oparozkyrofa, OK and the wrapper will provide the correct env15:04
kyrofaoparoz, assuming you're using the python plugin, yeah15:05
oparozOK, I'm going to try that and see if it works, thanks15:05
sergiusenspython is very unflexible about installation paths which is sad for such a flexible language in itself15:05
ogra_wobbly-python :)15:06
_morphiszyga: whats with the changes you did for the bluez itnerface? did jdstrand review them?15:19
jdstrandI wanted to sync on that today. I'm not sure what is expected at this point15:19
ogra_will be interesting to see what happens if you install it on a desktop :)15:20
jdstrandthat's another converstation that needs to be had. sdoc interfaces vs core interfaces15:22
seb128jdstrand, hey, I think you had a go at making at snap from gnome-calculator, is your work available somewhere? we are at a sprint and would like to have a look to start tomorrow, would be nice to restart from scratch if you resolved some of the issues already though15:25
jdstrandseb128: I tried, it failed. I talked to dsert and he gave some tips but I was unable to continue. someone I think last week was trying something similar and so I gave my notes. let me find that15:26
seb128jdstrand, thanks15:27
jdstrandseb128: this is the branch (note, it is horrible) -- lp:~jdstrand/+junk/gnome-calculator . Still digging up notes15:31
seb128jdstrand, thanks15:32
jdstrandright, it was sergiusens I talked to about it15:32
jdstrandseb128: http://paste.ubuntu.com/15923754/15:32
seb128sergiusens, ^ did you look more into that one (before we dup work)?15:33
jdstrandseb128: sergiusens was looking at snapping firefox (which is gtk) and I gave him those notes. not sure how much farther he got if at all15:33
seb128jdstrand, thanks, hopefully we can move things forward at the sprint this week15:33
sergiusensseb128 no I haven't15:34
jdstrandthough I imagine you're going to need some interfaces work. I was trying to get it to a point where I could look at that, but couldn't at first and then was busy helping with interfaces/etc for release15:34
seb128sergiusens, no worry15:35
jdstrandso a working snap in --devmode would be great15:35
_morphisjdstrand: not sure, zyga just said he wanted your eyes on what we he implemented15:36
jdstrandI thought I did that...15:36
_morphisjdstrand: https://github.com/ubuntu-core/snappy/pull/1037#issuecomment-21438681615:37
* jdstrand looks at the PR again15:37
_morphisjdstrand: hah, sounds like we're deadlocking :-)15:37
jdstrand_morphis: I need zyga to comment. I gave zyga a patch for him to build off of. I haven't seen anything else15:38
_morphisjdstrand: he implemented https://github.com/zyga/snappy/commit/16228ad739da17d0ff975a1d781db4a53dfdbc7915:38
jdstrand_morphis: ok, commented in both15:44
_morphisjdstrand: thanks15:44
_morphisjdstrand: didn't looked through the snappy code yet for this, but isn't there an easy way to figure the plug name?15:45
jdstrand_morphis: not at that point in the code15:45
_morphisjdstrand: ok15:46
_morphisssweeny: can you pull those changes in from zyga?15:46
_morphisjdstrand: will implement the same for networkmanager tomorrow15:46
jdstrand_morphis: that is the bit that zyga needed to look at (I gave him a similar patch as the commit which he was going to work off of to get the app name, which isn't available in any of the structures in that function atm)15:46
ssweeny_morphis, already pulled them locally and tested them15:46
ssweeny_morphis, pushing now15:46
_morphisssweeny: awesome!15:47
jdstrandplease add a FIXME comment though15:47
_morphisjdstrand: could you have a look on the minimize dbus policy at https://github.com/ubuntu-core/snappy/pull/1036 too?15:48
_morphisjdstrand: will add the same changes we're doing for security label tomorrow15:48
ssweenyjdstrand, it should end up as "snap.bluez." ?15:48
ssweenytrying to parse the punctuation in your comment15:48
jdstrandssweeny: what 'it' are you referring to?15:50
jdstrandthe future fix or the current implementation?15:50
jdstrandcurrent proposed* implementation15:51
jdstrand_morphis: commented15:51
_morphisjdstrand: thanks15:52
ssweenyjdstrand, I mean, should the comment be "FIXME: this glob turns into snap.bluez.* where it *SHOULD* be snap.bluez."15:52
jdstrandssweeny: 'where it *SHOULD* be snap.bluez.<app>'15:54
ssweenyjdstrand, ok that's what I wanted to understand15:55
jdstrandssweeny: also, really, it should say 'where it *SHOULD* be snap.<name>.<app>'15:55
ssweenyjdstrand, I think your <app> got lost in the comment15:55
* jdstrand shakes fist at github15:55
ogra_what do you expect from a site that requires flash to provide you tarballs of the trees :P15:56
jdstrandssweeny: fixed github comment15:56
mvojdstrand: snappy-debug is your baby, right? would be nice to have an update for 16 - looks like it is using old-security, anything we can do in the short term with it? do we have an interface that works ?15:58
jdstrandmvo: it is completed broken15:59
mvojdstrand: and we can not bring it back because it has to be unconfined?15:59
jdstrandmvo: it has to be rewritten. best to remove it at the moment15:59
jdstrandthe tool itself is broken15:59
jdstrandinterfaces moved all policy into go and so there is nothing to grep16:00
mvojdstrand: oh, I see16:00
mvojdstrand: ok, I updated the bug about it (doing triage now and stumbled over this one)16:00
jdstrandis it in 16?16:00
jdstrandI didn't ask for it to be there16:01
jdstrandonce it is rewritten it should work with log-observe16:01
jdstrandbut that isn't autoconnected16:01
jdstrandmvo: ^16:02
mvojdstrand: its not in 1616:02
mvojdstrand: there is a bug that its broken in rolling and I was checking the state16:02
jdstrandare you taking my irc comments and putting them in the bug?16:03
jdstrandor shall I? I can, but what is the bug number?16:03
mvojdstrand: I will add them16:06
slvnhello. I'm trying to build/install my first snap package. I use SDL2 which needs to access "libmirclient". After building, "libmirclient" is not in my "/snap/mytest/current/..." whereas libSDL2 is. I think I should set a magic line in cmake to install it ? any idea ?16:06
mvojdstrand: https://bugs.launchpad.net/snappy/+bug/154311816:06
ubottuLaunchpad bug 1543118 in Snappy "Snappy-debug installation on 16" [Wishlist,Triaged]16:06
jdstrandmvo: thanks! fwiw, this is the next thing I will move to after various interfaces work16:07
mvojdstrand: ta16:07
sergiusensjdstrand can I invoke your wisdom http://pastebin.ubuntu.com/16065450/ ?16:14
sborovkovogra_: Hello, are there still 2 gadget snaps for rpi?16:15
sergiusensjdstrand oh, seems to not be related to security; crashes with --devmode too16:16
sergiusensI guess I need to do some sort of gtk dance16:16
ogra_sborovkov: yes, canonical-pi2 and canonical-pi316:16
ogra_they share the canonical-pi2-linux kernel package16:16
jdstrandI'll still add those denials as something to consider16:17
* jdstrand is gathering a list16:17
ogra_sergiusens: videos though, or it didnt happen16:17
ogra_i guess seb128 knows the exact steps for all gkt dances16:18
sergiusensyeah, it might not be gtk16:18
sergiusensogra_ jdstrand I'm running into this code path https://chromium.googlesource.com/chromium/src/+/lkgr/crypto/nss_util.cc#20616:18
ogra_uh, nss16:19
jdstrandsergiusens: you might ask ChrisCoulson if he knows anything about that chunk of code. I have a feeling he had to work with that before with webapp-container16:20
jdstrand(as webapp-container relates to oxide)16:20
sergiusensjdstrand yeah, it will be a hard fix as this is electron16:29
sergiusensnecessary though as all apps are written with this now :-)16:30
ogra_"all apps"16:30
sborovkovogra_: so for RPI2 and RPI3 different images will need to be built always?16:45
ogra_sborovkov: unless ppisati made the uboot binary work on both, yes ... the pi3 uboot breaks the pi2 serial16:46
ogra_i'll look into an arm64 build too in the near future16:47
ogra_so there might even be a third gadget16:47
=== chihchun_afk is now known as chihchun
sborovkovIs it possible to mount directory with binaries instead of squashfs temporarily? So that I have write access but still run inside of the snap16:52
* ogra_ doesnt understand that question16:56
ogra_you can only deliver snaps as squashfs ... so there is no way to make the snap itself writable16:57
ogra_you can mount overlays on top of files manually i guess ... for temporary changes and tests16:57
sborovkovogra_: yeah, I don't want to deliver, just to be able to make quick changes locally16:58
zyga_morphis: not that I know of yet, could you push them to the branch so that they appear in the pull request?17:31
* zyga wonders if this works17:43
zygajdstrand: hey, around?17:46
jdstrandzyga: yes, hi17:58
zygajdstrand: hey17:58
zygajdstrand: I sent a pull request the other day, that changed the x11 interface17:58
sborovkovzyga: Hello, do you have any time estimations when RPI interfaces going to be in? If not - is that a difficult change to do to allow access to /dev/vchiq and may be I could it myself then?17:58
zygajdstrand: I also made a change to bluez but I'm not sure if you've seen that17:59
zygasborovkov: hey, no estimate yet, our focus is on devices but I'm sure I'll be split between desktop and iot world for now17:59
zygasborovkov: I can guide you with interface work, I will gladly review and merge new interfaces17:59
zygajdstrand: https://github.com/zyga/snappy/commits/bluez-fix-rules18:00
sborovkovzyga: ah, alright cool, could you point me to the source code of some similar interface may be?18:00
zygajdstrand: https://github.com/zyga/snappy/commit/e0f7f3bec17b0f33f3aaf4fa0c0f4273c98a788a18:00
zygasborovkov: there aren't any yet, please familarize yourself with the two articles I've published on interfaces already, I have the third one half-written (specifically about how interfaces work internally), reading interfaces/core.go would be useful18:01
sborovkovzyga: Ok18:02
jdstrandzyga: I commented on https://github.com/zyga/snappy/commit/16228ad739da17d0ff975a1d781db4a53dfdbc79 a few minutes ago18:02
* zyga looks18:02
zygajdstrand: replied18:03
jdstrandzyga: I didn't see the x11 PR. where is it? (I didn't see the bluez one either until _morphis pointed it out to me)18:03
zygajdstrand: trying to find the pull request18:03
zygajdstrand: mvo merged it18:04
jdstrandzyga: replied to your question18:06
jdstrandzyga: getsockname is fine18:07
zygajdstrand: I think we need a protocol to ensure that in the future all such requests are acked by you before they land18:07
jdstrandI agree. we said before that any changes to policy should go through the security team18:07
jdstrandits fine if that is me for now, but it could be anyone on the team once things settle a bit18:08
zygajdstrand: thanks for clarifying the bluez patch, I will fix it shortly!18:08
jdstrandnp, thank you! :)18:09
zygajdstrand: offtopic, can we do a pulseaudo interface so that games can have sound :)18:09
jdstrandwell, that is a very interesting conversation18:09
jdstrandcause it brings out a problem we have18:09
zyga(perhaps just for playback as 1st attempt)18:09
jdstrandUbuntu Core systems vs sdoc18:09
jdstrandwhich we are seeing already18:09
jdstrandsnappy dimension on classic18:10
jdstrandsnappy on desktop18:10
zygayes, I see18:10
zygasame with network-manager and pulseaudio really18:10
zygaand bluez18:10
jdstrandfor example, we have a bluez interface, about to have nm and pulseaudio18:10
jdstrandyou see what I'm getting out18:10
zygaI was thinking that we should auto-create the slots on desktop systems and not create them on IOT systems18:10
jdstrandthe interfaces are with slots/plugs snaps in mind18:10
jdstrandbut classic has these as debs18:10
ogra_and if my IOT wants to use sound notifications ?18:10
jdstrandogra_: then you install the pulseaudio snap18:11
zygaI think the value should be that the interface is the same (if you have pulse snap)18:11
ogra_jdstrand: and that provides the same interface ?18:11
jdstrandzyga: I'm not so sure18:11
zygaand the same client snap would work on desktop (with deb based pulseaudio) and on iot with a pulse snap18:11
jdstrandogra_: that is the question18:11
zygaeeek, no power18:11
ogra_get an M10 !18:12
jdstrandtraditional desktop apps might use things in pulseaudio that we wouldn't want to allow to apps18:12
jdstrandfor example18:12
jdstrandpulseaudio has a socket interface and a dbus interface18:12
jdstrandon touch, we only allow access to one because the other is dangerous18:12
jdstrandso, core systems might want a different policy than classic systems18:12
zygaogra_: I'm not sure it could replace my activities today, I'm tempted to see how it plays out at the upcoming sprint18:12
zygajdstrand: ok, simplifying this for now, I'd like a sound interface that works for 80% of the desktop games out there18:13
ogra_zyga: well, it is good if you do a lot on remote machines ... i doubt i'd want to use it for loacl image builds or so :)18:13
zygajdstrand: because that's a vaiable target for snappy today and because it's a good thing to have :)18:13
jdstrandzyga: obviously I'm not saying no to accessing pulseaudio. I'm just saying all this brings up questions18:13
zygaogra_: I hack on snappy18:13
ogra_mail, irc, browsing and terminals definitely work fine ... and it lasts a good 16-18h18:13
zygajdstrand: yes, but I think we have to be pragmatic, we should populate 16.04 with useful desktop interfaces rapidly18:14
ogra_zyga: and ?18:14
zygaogra_: and nothing else, I use udf to build snap images18:14
zygaogra_: not heavyweight18:14
ogra_zyga: i hack on all sorts of stuff ... rarely on teh machine i'm typing on :)18:14
jdstrandzyga: no one is arguing that. I think a conversation with niemeyer would be worthwhile though :)18:14
zygaogra_: I don't want two machines, tried that, went back to one18:14
zygajdstrand: no disagreements there :)18:14
jdstrandmy inclination is actually maybe put these in unity7, or maybe unity7-audio18:15
jdstrandand leave pulseaudio interface as a proper interface for a pulseaudio snap18:15
zygaogra_: I'd be sold iff I could get usable editor (vim) with my settings and ability to run it without squinting my eyes in the terminal18:15
zygaogra_: assuming I can do a xenial chroot18:15
zygajdstrand: hmm, interesting18:15
ogra_and even natively ...18:16
zygajdstrand: I'm sure we'll bring this up at the sprint18:16
jdstrandzyga: because these are all of the 'classic desktop' persuasion and shoehorning stuff that wasn't designed for snappy18:16
zygaasac: offtopic, I just built the bbb image18:16
jdstrandzyga: on an image that isn't ubuntu core plus a bunch of magical dbus things that appear to be from no where (from the app's perspective)18:17
jdstrandanyway, niemeyer said he is back tomorrow I think. we can pick it up then18:18
sergiusensjdstrand ok, so my vscode snap turns out to fail even when running locally but preferring the libraries from the snap; coincidentally most of them are glib/gtk ones ;-)18:22
zygasergiusens: once we make it work, do we sent that to microsoft to release?18:23
sergiusenszyga yeah, I already learned gulp and created a `snap` target and have snapcraft.yaml fully building18:24
sergiusensit just doesn't work ;-)18:24
zygasergiusens: does a pre-made binary work?18:24
sergiusenszyga I don't think so; the electron binary is prebuilt (fetched from the build system)18:25
sergiusenszyga and the `snap` works as long as ld_library_path (toup())  is not set to point to the snap18:25
zygasergiusens: any idea what the problem is/18:25
sergiusenszyga I suspect it is gtk initialization18:26
zygasergiusens: did you try to snap the gtk demo app with all the widgets?18:26
sergiusenszyga no; gtk is black magic to me18:27
sergiusenszyga you try ;-)18:27
zygasergiusens: why? :)18:27
zygasergiusens: I can try on Friday18:27
sergiusenszyga ask jdstrand :-)18:27
zygajdstrand: ^^ :-)18:27
jdstrandask me?18:28
zyga(I want to know what I'm getting into)18:28
jdstrandI'm no gtk guru18:28
jdstrandseb128 is sprinting this week and are going to look at this stuff18:28
sergiusensjdstrand I didn't tell him to ask you to do it, but rather how your endeavour came through ;-)18:28
jdstrandI suggest talking to him18:28
jdstrandmy endeavor failed18:28
zygain what way?18:29
sergiusensfrom the looks of it; there are more vars than for Qt and there are dependencies on fixed paths18:29
jdstrandI was able to start the calc without crashing, but menus, themes, gsettings, etc all failed18:29
zygagtk loves to have various processes18:29
zygaI wonder what the approach should be18:29
zygaisolate each snap18:29
zygaor have them see a part of gtk (daemons)18:30
zygafrom the system18:30
zygae.g. themes will be a problem otherwise, I don't know what qt does here18:30
zyga(apart from fully-skinned apps)18:30
jdstrandzyga: note, I talked to dsert about this a bit a while ago, but couldn't go farther due to other snappy work. I forwarded that conversation to seb128: http://paste.ubuntu.com/15923754/18:30
zygaI suspect gtk will need a few patches to love $SNAP first18:31
jdstrandthat would not surprise me18:31
* ogra_ whispers "static builds"18:31
sergiusensreason I love go ogra_18:32
sergiusensjust drag and drop!18:32
ogra_so just do it in go-gtk :)18:33
ogra_trivial ... just port the app :P18:33
sergiusensogra_ go-gtk dyn links against gtk18:33
sergiusensit would have to be an api compatible rerwite using Qt underneath ;-)18:33
zygaogra_: what's the battery life like on m10?18:41
ogra_zyga: i'm at 60% after doing 6h of constant work on it18:42
zygaogra_: wow, what perhiperhials are you using?18:43
zygaogra_: do you have a screen attached?18:43
ogra_and i have a bunch of apps excluded from lifecycle mgmt ... meaning it actually consumes more than a default system18:43
ogra_only a k480 Bt keyboard18:44
zygaogra_: no mouse?18:44
zygaogra_: do you work with the only screen being the 10" tablet?18:44
zygaogra_: man, you must have good eyes18:45
zygaogra_: how do you physically keep it in place on your desk? any stands/arms holding it?18:45
zygaogra_: I was wondering if I'd switch to that device entirely or just use it in addition to my laptop18:49
sergiusenszyga jdstrand GI is s glib/gtk thing, right? http://pastebin.ubuntu.com/16069260/18:50
ogra_zyga: http://i.imgur.com/apYL565.png18:51
zygalooks like gobject introspection18:51
zygaogra_: it looks big on my 15" screen ;)18:51
jdstrandglib, yes18:52
zygaI wonder if someone could come up with hdmi-over-ethernet snap for x86 boxes18:52
zygapick old laptop18:52
zygainstall that snap18:52
zygaand use it as a networked monitor18:52
zygause modern hw with all the spare "monitors" :)18:52
sergiusensogra_ what chat app is that18:53
ogra_excluded from lifecycle18:53
ogra_the app from the store18:54
zygahow do you do that?18:54
zygaI was actually thinking that lifecycle elision would be a nice interface18:54
zygaI'm sure we'll have to solve a lot of very interesting cases before phone and snappy merge18:54
ogra_its a gsettings key that takes a list of app ids18:54
ogra_i got dekko, kiwi and the terminal added there18:55
sergiusensogra_ why dekko?18:56
sergiusensI use dekko all the time, never had the need to exclude it from lc18:56
ogra_because my mailserver is a single core machine running off a 4200rpm disk18:56
zygaogra_: do you have a chroot or did you made / writable?18:56
ogra_takes a century to deliver a mail ... so i want dekko to sync in bg18:56
zygaogra_: (as in, will you get OTAs?)18:56
ogra_nothing writable, no18:57
ogra_i have a local libvertine container for X apps18:57
ogra_the device is proper on teh stable channel, system partition untouched18:57
zygaogra_: you should blog about how you set that up18:58
zygaogra_: I'm not sure I even heard about libertine :)18:58
ogra_i was planning to do that on teh weekend ;)18:58
ogra_it gives you containers hooked into xmir18:58
ogra_so you can run firefox and stuff18:58
zygaogra_: is firefox readable on the fhd screen?18:59
ogra_well, it doesnt use the proper DPI settings yet ...18:59
ogra_it is very small ... but you noticed already that i have good eyes ;)19:00
ogra_but i obnly used it twice for a few mins ... the ubuntu browser is really superior to it imho19:00
sergiusensI am using the ubuntu browser on desktop as well fwiw ;-)19:00
sergiusensso light weight!19:01
=== devil is now known as Guest52703
=== Guest52703 is now known as devil_
sborovkovIs /writable still the place to write a lot of stuff? I don't want to write to SNAP_USER_DATA because it would be replicated (from what I understood) and since I am on RPI and there is a lot of stuff I need to store that would be pretty bad for SD card19:16
ogra_your snap cant see /writable ...19:16
ogra_but yes, it is still the writable area (as the name kind of implies ;) )19:17
ogra_and actually our only system partition nowadays19:17
sborovkovWait, so what do I do if it can't see it19:17
ogra_you write to SNAP_DATA19:18
ogra_afaik there was work going on to allow your snap to override the duplication/copying on upgrade ... not sure where that stands though19:18
sborovkovogra_: SNAP_DATA won't be replicated and will be persistent, right?19:19
ogra_it will be replicated by default, but yes, it is persistent19:19
sborovkovogra_: Ok. May be it would be better to use HOME interface? it allows access to /home/ubuntu as far as I understand. We keep playlist which can be pretty large19:20
sborovkovany additional replications won't be good for SD card19:21
ogra_not sure if teh home interface actually works on IOT ... thats a desktop thing19:21
ogra_also note that you can not attch to the home interface automatically19:21
ogra_requires user interaction19:21
ogra_as i said, there was work to allow you to prevent the duplication ... but i dont know where that stands19:22
oparozogra_, regarding "your snap cant see /writable" is this a new thing which is going to stay that way?19:29
sborovkovogra_: Yeah, I understand, but while it's not in I'd prefer to use something else without replication. We implemented slot in gadget snap (which we use custom version of) before to allow access to /dev/vchiq. Is it possible to do the same with new security to grant access to /writable or /home/ubuntu or any other directory?19:29
ogra_oparoz: nope, has always been that way19:30
sergiusenselopio kyrofa standup today?19:30
kyrofasergiusens, ah yeah on my way19:30
ogra_oparoz: a snap can only see its own subdirs19:30
oparozogra_, I'm able to access data from another snap by using a /writable path19:30
ogra_(one of them is a writable one)19:30
sergiusenskyrofa let me get some coffee first19:31
ogra_oparoz: that would be a bug if yuor snap is actually confined19:31
kyrofasergiusens, ugh, you remind me I need to roast some19:31
oparozI'm on the older mvo image though...19:31
oparozAh, good point, ogra... I was testing with unconfined19:31
ogra_yeah, then you have access everywhere :)19:32
ogra_sborovkov: install the hello-world snap ... then use hello-world.sh .... that gives you a shell inside a confined snap env19:32
ogra_so you can inspect what you can access and what you cant19:33
oparozogra_, so the way forward is to wait for you guys to implement an interface giving writable storage access to other snaps19:33
oparozthat would help make snaps more atomic19:34
ogra_well, not sure that is planned at all ... you can have a "library" snap that all your other snaps can access ... i guess that way you could have a shared dir19:34
ogra_but i think these library snaps are also still a bit away19:34
oparozah, that famous library snap :)19:34
ogra_zyga could probably tell you more ... i'm mainly guessing here from what i pick up in drive-by conversations19:35
jdstrandtyhicks: can you advise how I should fix https://bugs.launchpad.net/ubuntu/+source/ubuntu-core-launcher/+bug/1574556 ? it seems like I should add to ubuntu-core-launcher's policy the ecryptfs workaround rules from the base abstraction19:53
ubottuLaunchpad bug 1574556 in ubuntu-core-launcher (Ubuntu) "apparmor denials reported for encryped HOME" [Undecided,New]19:53
tyhicksjdstrand: that's the best option for now - there'll soon be a kernel fix upstream for this and then we can SRU that and drop the workaround rules19:54
jdstrandthat would be fantastic! :)19:54
* jdstrand hugs tyhicks :)19:54
tyhicksjdstrand: actually, it was a contribution from chromeos (I still need to review and test v2 of their patch)19:55
wililupyOk, let me ask if this makes sense for adding kernel modules to my kernel snap...19:55
wililupyI get my source (apt-get source linux-image-`uname -r`19:56
jdstrandI don't know how to hug chromeos, so I'll leave that with you19:56
wililupyI build it with snapcraft (snapcraft build)19:56
tyhicksjdstrand: :)19:57
wililupyI then build my kernel modules against the build (make all KSRC=/tmp/ksnap/parts/kernel/build)19:57
wililupyI then stage (snapcraft stage)19:57
wililupyI copy the modules into ./stage/lib/modules/`uname -r`/extra/*.ko)19:58
wililupydepmop them (depmod -b ./stage/kernel `uname -r`)19:59
wililupythen build the snap (snapcraft snap)19:59
wililupyDoes that make sense? I want to script this for the customer and want to know if this is a good solution.20:01
=== devil is now known as Guest16642
=== Guest16642 is now known as devil_
* zyga looks at backlog for that one bit 21:25
kgunnso, if i'm adding an interface...and i go to install my built snap for the very first time...bout how long should the  'Setup snap "mir-server" security profiles' take?21:31
kgunnseems to be spining quite some time...21:31
kgunnany hints what i might have done wrong21:31
zygajdstrand: around21:31
zygakgunn: can you share your branch please21:32
kgunnwill do21:32
jdstrandkgunn: it would be nearly instant. no more than a second or two21:32
jdstrandzyga: yes21:32
sergiusensjdstrand kyrofa zyga nailed the vscode problem down to libnss3 being included in the snap21:32
zygajdstrand: I want to show you the bluez code with 2nd fix, let me quickly share it21:32
zygasergiusens: \o/21:32
jdstrandsergiusens: nice! :)21:33
=== chihchun is now known as chihchun_afk
sergiusensI don't know how to solve it yet though :-P21:33
zygasergiusens: solve it?21:34
sergiusenszyga I know what causes the SIGABRT21:34
sergiusensbut not why21:34
sergiusenszyga - Remove snap "vscode" from the system (remove /snap/vscode/100001/CHANGELOG.md: read-only file system)21:35
sergiusenswhat is going on?21:35
zygasergiusens: oh, nice bug21:36
zygasergiusens: please report this, note that it wants to remove the *changelog*21:36
zygasergiusens: I saw it before but I wasn't sure how to reproduce it21:36
zygajdstrand: https://github.com/zyga/snappy/commit/e9b2da3d1459d1f0dfa92036f0de477757024a4b21:36
zyga(no more globs :-)21:37
sergiusenszyga I've just installed and removed repeatedly21:37
zygasergiusens: please report it with snap changes and syslog parts (journalctl -u snapd)21:37
zygasergiusens: keep your state around if you can, I'm sure pedronis will want to see it21:37
zygasergiusens: (or attach it to the bug as well please)21:37
sergiusenszyga seems to be a systemd mount issue21:40
sergiusenszyga https://bugs.launchpad.net/snappy/+bug/157538521:41
ubottuLaunchpad bug 1575385 in Snappy "After installing and removing for a while I can't remove anymore" [Undecided,New]21:41
zygaApr 26 18:34:53 lindon /usr/lib/snapd/snapd[16408]: overlord.go:142: Failed to stop "/etc/systemd/system/snap-vscode-100001.mount": [stop snap-vscode-100001.mount] failed with exit status 1: Job for snap-vscode-100001.mount failed. See "systemctl status snap-vscode-100001.mount" and "journalctl -xe" for details.21:42
zyga                                                    , but continuing anyway.21:42
zygaApr 26 18:34:53 lindon snapd[16408]: 2016/04/26 18:34:53.066860 overlord.go:142: Failed to stop "/etc/systemd/system/snap-vscode-100001.mount": [stop snap-vscode-100001.mount] failed with exit status 1: Job for snap-vscode-100001.mount failed. See "systemctl status snap-vscode-100001.mount" and "journalctl -xe" for details.21:42
zygaApr 26 18:34:53 lindon snapd[16408]: , but continuing anyway.21:42
zygaI suspect this is key21:42
zygathanks sergiusens!21:42
sergiusenszyga np21:42
zygajdstrand: is the patch sensible?21:43
zygajdstrand: if so I can propose it and we can close the earlier branch21:43
* jdstrand looks21:44
zygajdstrand: as a full pull request: https://github.com/ubuntu-core/snappy/pull/1078/files21:44
zygaI also closed the earlier one21:45
jdstrandzyga: so, I think this looks generally fine. I've been working under the assumption that you can do a connection between snaps, between apps or between apps and snaps. if this is the case, will slot.Apps() contain all the apps if connection to a snap?21:49
zygaslot.Apps is the map of all apps bound to that slot21:50
zygait's not about connections21:50
jdstrandzyga: ie, what this does is get rid of the glob entirely, but the glob is useful for connecting to a snap21:50
zygaconnections are handled by calling the interface many times21:50
zygaso if both, say, bluetoothctl from the same snap is connected21:50
zygaand a 3rd party snap with some apps are connected21:51
zygathose will get separate snippets21:51
zygathat both specify the peer label precisely21:51
sergiusenszyga ok, another weirdness; now on every `snap install` I get this old snap installed that can't be removed21:51
zyganot sure if that answers your question21:51
sergiusenszyga already ran your cleanup script twice!21:51
zygasergiusens: oh?21:51
zygasergiusens: did you restart snapd too?21:51
zygasergiusens: I think the script stops it21:51
sergiusenszyga your script does that, does it not?21:51
zygaso run the script, don't start snapd21:52
zygasee if anything else remains, systemd mount units maybe?21:52
zygabut actually21:52
zygasnapd restarts changes + tasks21:52
* zyga has no idea21:53
sergiusenszyga not unmounting now21:53
jdstrandzyga: I understood that. what I'm saying is this: snap connect foo.bar:bluez bluez5:bluez vs snap connect foo.bar bluez5.bluezd:bluez21:54
jdstrandwhere foo has:21:54
jdstrandname: foo21:54
sergiusenszyga /var/lib/snapd/snaps/vscode_100001.snap (deleted) on /snap/vscode/100001 type squashfs (ro,relatime)21:54
jdstrand  bar: ...21:54
jdstrandand bluez5 has:21:54
jdstrandname: bluez521:54
jdstrand  bluezd: ...21:54
jdstrand  bluetoothctl: ...21:54
zygajdstrand: note that you cannot connect an application, if you connect foo's plug, you connect the bar app from foo snap (reading rest)21:55
zygajdstrand: I'm not sure if I understand what you are saying, in your example you would do the following connect:21:55
jdstrandzyga: ok, fine, so snappy does magic21:55
zygajdstrand: snap connect foo:something bluez5:bluez521:56
jdstrandlet me come up with a paste21:56
zygaor s/bluez5:bluez5/bluez5:bluez/21:56
zygasergiusens: we merged better remove today21:56
zygasergiusens: remove removes all revisions now21:56
zygasergiusens: maybe it hit the ppa by now21:56
zyga(though I don't know if you want to follow the ppa)21:57
wililupycrap. Ok. So I built my kernel snap using snapcraft, I add it as the --kernel= with u-d-f but when I try to boot up, I get the following error: http://pastebin.ubuntu.com/16071576/ and then I go back to GRUB menu...21:57
wililupyAny ideas?21:59
zygawililupy: nope, sorry21:59
wililupyEverything looks good, just no ideas. At first I thought maybe it was becuase I wasn't root when building the kernel snap, but even that doesn't work...22:00
zygaI don't think you should ever be root with snapcraft22:01
zygaogra_: any idea if I can try convergence/firefox on nexus 7?22:02
wililupyzyga it builds fine with my normal user account as well, everything builds fine, I just haven't been able to build a working kernel snap since 2.5...22:02
zygaogra_: I have a channel (pd-proposed or something) that has the firefox icon but it does nothing22:02
zygawililupy: I think you want to talk to sergiusens22:02
sergiusensno, I'm not a kernel person22:02
zygahe is just teasing ;)22:03
sergiusensjdstrand zyga ok, apparently I can't "relocate" nss, but it is not in ubuntu-core22:03
zygasergiusens: isn't nss using lots of plugins and a socket to talk to various parts and a daemon22:04
zygasergiusens: doesn't sound like something we should put into snaps22:04
zygasergiusens: it does name resolution among other things IIRC22:04
sergiusenszyga right, but if core doesn't provide it, what to do?22:04
zygasergiusens: provide it in core22:04
sergiusenszyga do it then ;-)22:04
sergiusensogra_ ^22:05
zygaogra_: correct me if I'm wrong but it feels like a core item22:05
zygabtw are you sure it's not in core?22:05
zygaextra users AFAIR use nss22:05
zyga(some fancy plugin to look at places other than /etc/passwd)22:05
jdstrandzyga: http://paste.ubuntu.com/16071638/22:05
sergiusenszyga if I don't add it $ vscode /snap/vscode/100001/code-oss: error while loading shared libraries: libnss3.so: cannot open shared object file: No such file or directory22:05
* zyga focuses on the pastebin22:06
kgunnzyga: will send you a mail with info later...got to attend to a house calamity atm :)22:06
zygakgunn: sure22:07
jdstrandsergiusens, zyga: not libnss3 is Network Security Services, an encryption library from the mozilla project and not 'Name Service Switch' from glibc name resolution22:07
jdstrandie, there is no daemon, etc, etc with libnss322:07
zygajdstrand: I see22:07
zygajdstrand: /o\22:07
zygajdstrand: as for the pastebin, I think we can special case 2. to just use the app name, for 1 I'd keep it as-is, it's precise and does the same thing22:08
jdstrandsergiusens: I wonder if you need libnspr422:08
* zyga will name all his new libraries libnss22:08
zygawith a random number22:08
sergiusensjdstrand it is in there22:09
sergiusensjdstrand I am manually adding the plugins now to see if that is it22:09
sergiusensadding libnss as a stage-package22:09
jdstrandzyga: so I'm inclined to agree, but note that complex alternations ({1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,...}) can be a bit unwieldy when a glob would be better22:10
zygajdstrand: hmmm, so the glob might make sense if slot.Apps != slot.Snap.Apps22:10
zygajdstrand: we can special case both cases so that we get a glob or no globs or alternation22:10
zygajdstrand: if you think it's worth doing22:10
jdstrandzyga: it is most correct and since we are thinking about it, it would be better to do it now. otherwise I would suggest you add a comment saying that snaps with a lot of apps in them could be better written with a glob rule22:11
zygajdstrand: you mean special case both of those?22:12
zygajdstrand: I can do it quickly22:12
jdstrandzyga: yes please22:13
sergiusensjdstrand zyga ok I stage-package'd libss3 and am further ahead, happy to see apparmor denials :-) http://paste.ubuntu.com/16071766/22:13
jdstrandzyga: which means we need a test for case 1 and case 2 in addition to your existing test (of course)22:13
jdstrandsergiusens: ok, two of those are the same as you gave earlier and I'll do a PR to add them (the /proc denials). the /etc/profile.d you need to invoke bash differently (ie with a bashrc) so it doesn't look there and /var/tmp is probably a noisy denial22:15
jdstrandsergiusens: it is probably trying /var/tmp then going to /tmp22:15
sergiusensjdstrand maybe, yeah, the first ones were there already, just put it to be thorough22:16
sergiusensjdstrand wrt bash, I not invoking it directly so no idea22:16
sergiusenswill need to debug some more22:16
sergiusensjdstrand the vscode electron process is running, but I don't see anything on screen; so that's my next issue22:17
jdstrandsergiusens: I mean, we could allow /etc/profile.d, but that would just open up whatever is in in there and cause trouble (eg, it adds apps-bin-path.sh which updates PATH which is not what we want)22:18
jdstrandsergiusens: I suspect that read on /etc/profile.d is not causing any issues and just noice (but I don't know that)22:18
sergiusensjdstrand --norc is the default if called as /bin/sh so I'll need to dive into the vscode code22:20
zygaok, done, pushing22:20
zygajdstrand: https://github.com/ubuntu-core/snappy/pull/1078/commits/c1c084f186a6e747db2d03109206d387760b73a422:23
croephawhat do you all use for a text editor on snappy? the best I can come up with is to run nano inside of docker with a mounted volume out to the host22:24
zygacroepha: I use vim on my host22:24
zygacroepha: I don't hack on snappy itself, I did when we had classic (we're getting classic back soon)22:25
zygacroepha: when I had classic I also used vim22:25
zygacroepha: don't settle for less :)22:25
croephawhat is classic?22:25
zygacroepha: on snappy you could say "snappy enable-classic" and you got a regular classic, deb-based ubuntu22:25
zygacroepha: we removed it before the 16.04 release to polish it before it is brought back22:26
croephaahh ok22:26
zygacroepha: it essentially lets you apt-get install anything you need inside snappy in a lxc container that is integrated with the host in a special way (more than a regular lxc ubuntu container would)22:26
zygae.g. apt-get install minicom and use the serial port directly22:27
croephazyga, ahh thats clever, kinda similar to what I was doing with docker22:27
wililupyOk, I installed the .snap on a current system, and when I look at /writables/snap/im-kernel, it has the symbolic links still set to the system I build the snap on... Is that normal?22:27
sergiusenszyga another one for you https://bugs.launchpad.net/snappy/+bug/157539922:27
ubottuLaunchpad bug 1575399 in Snappy "Half installation for a snap" [Undecided,New]22:27
zygajdstrand: I'm EODing, it was great to work on this but I'm falling asleep now22:27
wililupyI rebooted the system, but it didn't boot my new kernel snap...22:28
zygasergiusens: thanks!22:28
zygasergiusens: we're sure we have bugs in install/remove code22:28
zygasergiusens: and in undo code for those bits22:28
zygasergiusens: more testing == faster bug fixing :)22:28
croephaIm really new to snappy, just picked it up today, is one of the goals to basically make system upgrades bullet proof? like no issues where you did a dist-upgrade and now you have to get into recovery mode... stuff like that?22:28
zygacroepha: among other goals22:29
* zyga EODs for real, ttyl22:29
wililupylater zyga!22:29
wililupyIt's almost like snappy is seeing my kernel snap as a gadget...22:29
wililupyIt didn't ask to reboot after installing it to make it active...22:30
jdstrandzyga: thanks for the PR, I made a nitpick comment but feel free to ignore. thanks and good evening! :)22:31
sergiusensjdstrand last one http://paste.ubuntu.com/16072031/ (I hope)22:39
jdstrandsergiusens: seems you need 'plugs: [network]' but possibly 'network-bind' for 'shutdown' (assuming this is amd64)22:41
jdstrandsergiusens: do you know why it is using the shutdown syscall?22:41
jdstrandactually, strike that22:42
jdstrandshutdown is in 'network'22:42
jdstrandjust plugs network22:42
wililupyHas anyone successfuly built a kernel snap with snapcraft 2.8.4?22:43
sergiusensjdstrand oh, they weren't auto assigned22:51
sergiusensjdstrand this is totally weird http://paste.ubuntu.com/16072127/22:51
sergiusensok, back to the known issues22:53
jdstrandsergiusens: huh. sounds like a bug. I know I've seen times when they didn't get autoassigned...23:00
sergiusensjdstrand I've logged my fair share of bugs today!23:02
=== wililupy is now known as wililupy|afk
=== blr_ is now known as blr
=== wililupy|afk is now known as wililupy

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