[00:23] Hi [00:23] im looking into snappy these days but i want to install the boost library on snappy but how can I do that [00:24] anyone that can help me [00:25] noizer: you can use snapcraft to generate a snap that will consume the library [00:25] or you can use the classic mode while you are developiong something. [00:32] will boost then can be used of anny snap that i will deploy on it? [01:08] kyrofa, still around? I hope :-) [01:37] sergiusens, yeah hey :) [01:43] sergiusens, got your email [01:46] kyrofa, I just hacked ROS_MASTER to made up hostname and configured that in /etc/hosts [01:46] sergiusens, well, you just saved me some typing [01:46] kyrofa, just because I just read this " Great care should be taken when using localhost, as that can lead to unintended behaviors with remotely launched nodes. " [01:47] kyrofa, we should make our plugin do export VAR=http://$(hostname):#### [01:47] kyrofa, lets see how it goes :-) [01:47] kyrofa, how's everything with you, all good today? [01:48] sergiusens, I'm not convinced that will change things, but if it doesn't, I bet it's still hostname-related. I have more ideas for you [01:48] kyrofa, maybe start shooting them out :-) [01:49] sergiusens, today was a bit frustrating. rpi problems. But I worked through them, things are building [01:50] kyrofa, 16.04? Did unplugging the piglow thing work btw? [01:50] sergiusens, right so, when nodes communicate they provide contact info: "here's how to reach me" [01:50] sergiusens, thing is, by default they use their hostnames [01:50] So when setting up a ROS cluster, we always make sure each computer has the others in /etc/hosts [01:51] But an alternative is to use the ROS_IP variable [01:51] kyrofa, I have /etc/hosts setup now, more so because setting up ROS_IP on the snap would be painful [01:51] Which allows you to skip using hostnames all together [01:52] kyrofa, I tried ROS_IP on the plain ubuntu box, didn't work out that well [01:54] Yeah, but snappy doesn't have a real hostname right? It's "localhost"? [01:55] So you may want to set ROS_HOSTNAME on the device as well [01:55] kyrofa, I can set one [01:55] kyrofa, using snappy config [01:57] Oh interesting. It's odd that I didn't run into this, but yeah: the surefire way to make this work is to use hostnames all the way through. dev computer has a hostname that snappy can resolve, snappy has a hostname that dev can resolve, set the ROS_MASTER_URI to localhost on snappy, set it to snappy's IP on dev [01:57] sergiusens, I'm going to have to sit down and make sure we have a config that actually works for this use-case [01:58] kyrofa, right, so when you say "ROS_MASTER_URI to localhost on snappy" I think we want it to be the hostname and not localhost [01:58] from what I saw in the wiki === rharper` is now known as rharper [01:58] sergiusens, that should be fine. I've never done it, but I can't imagine it'll hurt anything [01:59] sergiusens, how was this working for you previously? Did you change your network setup somehow? [01:59] export ROS_MASTER_URI=http://rpi2:11311 [01:59] export ROS_HOSTNAME=rpi2 [01:59] kyrofa, that's my setting now on the rpi2 [01:59] sergiusens, if rpi2 is actually its hostname, I don't think you need ROS_HOSTNAME [02:00] ping rpi2 and ros-trusty (my hostnames) works from either side [02:00] kyrofa, I'm setting it up just in case [02:00] sergiusens, alright [02:00] sergiusens, yeah your setup sounds good. Settings on the dev machine? [02:01] kyrofa, export ROS_MASTER_URI=http://rpi2:11311 [02:01] ubuntu@ros-trusty:~$ vi /etc/host [02:01] oops [02:02] just the export :-) [02:02] sergiusens, yeah fire away [02:02] rostopic and rosnode still work with that setting [02:02] kyrofa, waiting for the squash now ;-) [02:03] sergiusens, nice. And no-- I'm still waiting for the owncloud build. Now you know why this has taken me so darn long [02:04] kyrofa, hah [02:04] sergiusens, compiling mysql died several times for lack of RAM. I started using swapfiles... 1G... 2G... screw it, plugged in a flash drive, 16G swapfile [02:04] kyrofa, I've been considering scalingstack [02:05] sergiusens, do they have ARM? [02:05] sergiusens, I just learned about the alpha launchpad .snap builds, too [02:05] sergiusens, I asked to get in on that [02:05] kyrofa, I probably got the name wrong though [02:06] it is arm exclusive [02:07] Oh... hmm yeah that doesn't seem arm exclusive [02:08] 96%... [02:08] kyrofa, working! [02:08] sergiusens, yes! [02:08] kyrofa, going downstairs for a bit [02:08] sergiusens, okay :) [02:09] kyrofa, I've documented the recipe, so I'll share later ;-) [02:09] sergiusens, sounds good [02:14] sergiusens, my cell# is in the directory if you need me and I'm not around [02:23] 100%! [06:44] kyrofa hello I am just testing :-) === chihchun_afk is now known as chihchun [08:15] good morning [08:58] good morning, in the webchat snapcraft example I see an `apps` directive usage, but in the source there is no mention about it. Where is the magic? :) Thanks [09:01] pindonga: beuno: I'm dangerously close to get the xenial tests back to green. static, unit, integration and coveralls are working. [09:01] Just need to figure out why the examples started to fail yesterday. So ETA, one more day. [09:21] guys for a user providing some extra configuration to a framework is there any standard way? What I can imagine is having a config file in the framework's writable location and then have the wrapper script of the service read that config file to see what args to use when launching the service. [09:21] Then an extra binary to query/edit the config for the framework [09:33] Good morning all; happy Thursday, and happy Hugging Day! 😃 [09:50] basic question. All applications in ubuntu core snappy run inside LXD. So can it run gui apps like Google chrome ? === lool- is now known as lool === Odd_Blok1 is now known as Odd_Bloke === fgimenez_ is now known as fgimenez [12:02] Good morning [12:09] LefterisJP, might be a use for snappy config [12:09] LefterisJP, https://developer.ubuntu.com/en/snappy/guides/config-command/ [12:10] joeblew99, that's not correct-- they don't run inside an LXD [12:11] joeblew99, however, supporting things like .desktop files is work that ongoing [12:11] joeblew99, so yes, eventually, but only in betas right now [12:11] joeblew99, https://rainveiltech.com/posts/prototype-a-gui-friendly-snappy [12:19] kyrofa: looks interesting. So there should be a binary inside the app that can read and output yaml syntax. But seems there is no way to set specific config option from the CLI, just provide a new yaml file to replace the old one. [12:21] LefterisJP, yeah kinda. ogra_ may be your best resource here-- I know for certain he's used this. I'm not sure of its capabilities [12:24] all right will wait for his input then [12:25] one more question. I have been trying to turn a snappified directory project to a snapcraft project. How can I make a snapcraft, part of recipe for: 1) Clone repository 2)run specific commands on the repo to produce binaries 3) copy these binaries to the staging area. [12:26] Also the various profiles and other data that go in the meta/ directory for a snappified directory app, how can they be included when creating a snap from snapcraft? [12:27] LefterisJP, alright let me make sure I understand [12:28] LefterisJP, you have a project with some sort of build system (even if it's just scripts). You want snapcraft to clone that project, run the build system, and include the binaries in the .snap? [12:28] LefterisJP, that sounds exactly like what snapcraft does. Am I missing something? [12:29] LefterisJP, but your build scripts also produce the meta/ stuff? [12:30] LefterisJP, sorry... I'm feeling a little dense this morning :P [12:30] * kyrofa pours his coffee [12:34] hehe so let's see [12:35] I currently have 1 project which is bassically a readymade directory on which I snappy build [12:35] But I already have the binaries built and ready under bin/ [12:35] Okay, with you so far. And you don't use snapcraft because... cross compiling? [12:35] yes [12:35] Okay, continue [12:36] But if the user has the appropriate libraries, I can actually make it cross-compilable, since this is a go-project and the developers have made a very handy makefile (which resorts to go build eventually) [12:36] so I just want to clone the repo and run make with 2 different arguments [12:36] and then take the 2 produced binaries [12:37] and from then an on continue as I would with a normal snappy build [12:37] Hmm, and are these "appropriate libraries" available in the ubuntu repos for a given arch? Say amd64 or whatever you're running on [12:39] LefterisJP, or are they other things you need to clone and build? [12:39] no they are there [12:39] only need to clone this one repo [12:39] LefterisJP, ahh.... you just might be able to get this to work [12:39] LefterisJP, okay. Are these libs build-time dependencies then, not run-time? [12:41] yes, since it's go everything is statically linked. I already do this manually, and have it working on a Raspberry Pi. Just wanted to automate it with snapcraft [12:41] LefterisJP, excellent. Automation is good [12:43] LefterisJP, can you add a target to the repo's makefile to run both make commands that you need? [12:44] LefterisJP, or is this a third-part repo you're cloning? [12:45] I don't think this is an option [12:45] LefterisJP, alright, you need to do a bit more work then [12:45] I saw there is a make plugin, but that seems to run only make [12:45] (no args) [12:46] LefterisJP, with my limited understanding of what you're doing, I'd use `build-packages` to pull in those libs, and use the `make` plugin with `source: src` and in the src folder create a Makefile to clone your source and run the necessary commands [12:47] LefterisJP, essentially use the Makefile as a script [12:47] LefterisJP, keep in mind that the only targets the `make` plugin uses is `all` and `install` [12:48] And the `install` target is called with DESTDIR [12:48] hmm is there a way to use a "script" plugin? Just write a script and call it as part of the preparation of building the snap? [12:49] LefterisJP, not included with snapcraft. But did you know you can actually include your own plugin in-tree? [12:49] And use it in the snapcraft.yaml? [12:49] nope. That's funny though. There are so many plugins, I would expect one for a simple script to exist. [12:50] LefterisJP, I just use the `make` plugin when I want a simple script. How would you expect snapcraft to tell the script where to put stuff? [12:51] LefterisJP, all the other plugins have established ways to do that (make, autotools, cmake, go) [12:51] call it with an argument which would be interpreted as a path I suppose [12:52] LefterisJP, yeah, that might be a useful addition. Feel like logging a bug? [12:53] LefterisJP, regarding custom plugins: https://github.com/ubuntu-core/snapcraft/blob/master/docs/plugins.md [12:53] Sure, is it https://github.com/ubuntu-core/snapcraft? [12:53] LefterisJP, that's the src, but bugs are tracked here: https://bugs.launchpad.net/snapcraft/ [12:53] thanks for the link, will check it out right now. [12:55] LefterisJP, if you come up with a generic script plugin that you like feel free to propose it to snapcraft [12:56] Would love to. So that would be as a Github PR? [12:58] LefterisJP, indeed. Bugs are tracked at LP, everything else on github [13:01] LefterisJP, check out the HACKING.md as well as the CONTRIBUTION.md [13:02] LefterisJP, it'll need some test coverage as well (see the other plugins for examples) [13:04] noted, thank you. If I come up with something good in time I will send it back upstream. [14:31] kyrofa: there seems to be a problem in hangouts. Could you get in the call? [14:32] elopio, google has been annoying lately [14:33] elopio, want to try again later? [14:34] kyrofa: what if we try https://meet.jit.si/snapcraft ? [14:34] seems to work here. [14:40] Agh, elopio sorry got distracted. I'll be right there [14:44] elopio, that doesn't seem to be working for me either... [14:46] kyrofa: :) want to stand up here? [14:46] elopio, works for me! [14:47] elopio, your turn first [14:49] DONE: update integration tests working on snappy rpi2, coveralls reporting working again from travis lxd, attended the mqtt class and got the mosquito server and client working on rpi2 classic. [14:49] TODO: understand the examples errors in jenkins (when I put prints they stop raising an exception :@), make a snapcraft for mosquitto, more work on the jenkins snapcraft jobs. [14:49] kyrofa: your turn. [14:49] Awesome === jdstrand_ is now known as jdstrand [14:53] DONE: Built owncloud package for amd64 and uploaded to store. [14:53] DONE build owncloud package for armhf but then realized I couldn't upload to the store [14:53] TODO: Removing amd64 version from the store, uploading new armhf version to replace it [14:54] TODO: Got emailed several papercut bugs last night. Turning them into real bugs and fixing the low-hanging fruit [14:54] elopio, ^^ [14:55] thanks. [15:16] @fgimenez I think I need to go and pick up the car of a friend that's traveling, soon. So I might miss our standup. [15:16] elopio: No such command! [15:16] anything you would like to discuss today? [15:16] ahh, telegram is affecting me. [15:17] elopio, ok :) i have a couple of PRs and another one is coming, no rush at all anyway [15:19] fgimenez: I'll get to them soon. [15:19] elopio, thx also there was an error in the integration tests yesterday caused by with https://github.com/ubuntu-core/snappy/blob/master/integration-tests/tests/autoupdate-msg_test.go#L53, could you take a look? if we can't rewrite it maybe it should be dropped, let me know what do you think [15:19] fgimenez: from my side, I had to trick travis using the same path in the container and in the host. Now the coveralls works again. [15:20] I tried everything I could think about to send the report from the container, and got every possible error. [15:20] elopio, great, it's working so \o/ [15:21] fgimenez: well, now the problem are the examples tests. They raise and exception, but when I try to print the stderr, it's not raised. [15:21] damn it with travis. [15:22] elopio, let me know if i can be of any help [15:22] fgimenez: about that test, I would prefer to discuss with mvo about possible solutions. It was one of the first tests added after a UI change, so I wouldn't like to remove it. [15:22] I'm fond of it :) [15:22] fgimenez: is it failing often, or was it just one time weirness? [15:22] I can put it in a loop. [15:23] elopio, only once in a lot of executions [15:26] this is a great use for jenkins. [15:26] just loop running one test to get more info. I'll try to make the job soon, but I still owe a couple of things from yesterday. [15:27] like the dockers in the hub. I'll finish that first. [15:27] elopio: we make it a long operation by creating a fake os update first [15:27] elopio: well, longer - and also throttle the download of the fakestore [15:28] ahh, that's right. Now the update takes a lot less. [15:28] ok, going for the car. bbs. [15:39] that was a lot faster than what I thought. [15:40] and now I have a car for the rest of the week. No idea what to do with it, though. [15:41] elopio, hahaha === chihchun is now known as chihchun_afk [16:02] kyrofa: can you give me a hand debugging this? https://paste.ubuntu.com/14590836/ [16:02] how can I see what causes make to exit != 0? is there a verbose flag or something? [16:03] elopio, that's the most unhelpful error message I've ever seen :P [16:05] jdstrand: thanks for your work on the squashfs-tools \o/ [16:05] elopio, with cmake you can to VERBOSE=1 [16:05] elopio, but the actual error should be above what you pasted... [16:07] mvo: sure. Now I need to see about different archs. I see some code about little endian vs big [16:08] jdstrand: iirc its nowdays (v4) all one endianess, that might be from the old v2 or v3 squashfs format. but mind you, I'm not an expert :) [16:08] mvo: oh, btw, kgunn needed a new os snap for the latest ubuntu-core-security when you have a chance. I don't hink he is blocked on it per se (perhaps he can comment) [16:08] mvo: we'll see! :) [16:08] jdstrand: sure, happy to update. amd64? [16:08] I think so [16:08] kgunn: ^ ? [16:08] ta [16:08] elopio, can I see more of that log? [16:08] elopio, is that in travis somewhere? [16:09] kyrofa: I tought python was swallowing the stderr, but it seems it just doesn't print anything. [16:09] ubuntu@xenial:/home/elopio/build/snapcraft/examples/ros/parts/catkin-tutorials/build/build_isolated/beginner_tutorials$ VERBOSE=1 make -j2 -l2 || echo $? [16:09] 139 [16:09] kyrofa: yes, one second. [16:10] kyrofa: https://travis-ci.org/ubuntu-core/snapcraft/jobs/103794957 [16:10] search for Failed to process package 'beginner_tutorials' [16:15] elopio, wow, that really is all it gave you [16:15] elopio, I have no idea what happened there. Yeah, try the verbose [16:17] kyrofa: but did you see my line with verbose above? It exits with 139, but prints nothing. [16:17] elopio, ... [16:20] elopio, I'm honestly clueless. I've never seen that [16:20] elopio, what are you testing here? [16:21] elopio, the only time I've seen anything similar was yesterday, when make was segfaulting for me [16:21] that's what I hate about this. I'm sure I saw it before, at some time. But I've seen so many travis errors this days that I'm confused. [16:21] kyrofa: I'm trying ot build from a xenial lxd. [16:22] at least now I can reproduce it on my local xenial lxd, so I don't have to wait for travis' output. [16:22] elopio, I'm so tired of ROS barfing on xenial [16:23] kyrofa: oh, but there are 7 tests failing with the same. [16:23] sorry, I picked you on ros because I thought you would magically tell me the solution :) [16:23] mvo: jdstrand yep, amd64 [16:23] elopio, hmm. You can reproduce on xenial though? Are they all `make` failures? [16:24] kyrofa: I'm trying that, but the costa rican archive is out of sync. So trying another mirror and goes even slower... [16:24] will know soon. [16:25] elopio, AH! mvo [16:25] elopio, mvo just installed make on my xenial lxc. IT SEGFAULTS! [16:25] elopio, I bet you $100 that's your problem [16:25] elopio, oh I'm so happy I'm not insane [16:26] kyrofa: ohhhh! so a real upstream make bug! [16:26] mvo, yeah! [16:26] mvo, dang. That's so upstream I don't even know where I'd log such a bug [16:26] oh come on, now also the main archive gives a hash mismatch. [16:27] kyrofa: what command are you running to get the segfault? [16:27] elopio, make -v :P [16:27] elopio, everything seems to segfault except make -h [16:27] oh, wow. [16:28] Yeah, terrible [16:28] it segfaults when I run it as root, but when I run it as user it doesn't say anything. [16:28] pindonga: I give up. Even make is failing on me. [16:28] elopio, yeah I went back to good-old trusty [16:28] trusty trusty [16:28] elopio, :( [16:29] kyrofa: so, this happens only on xenial lxc, right? on your real xenial machine it works? [16:29] elopio, I fist experienced in on my rpi2 in classic, so no [16:29] Well wait [16:29] That's actually lxc too [16:29] right? [16:30] mvo, ^^ [16:30] elopio, I don't actually have a xenial machine [16:30] kyrofa: not lxc, more like a chroot, it using unshare lightly but mostly a chroot [16:40] kyrofa: same version in real machine and in container. 4.1-2. It only fails in the container. [16:41] elopio, very interesting! [16:45] kgunn: I put an updated ubuntu-core snap into the edge channel, once I tested it I will promote to the other channels (I guess you will still get it because all images default to edge iirc currently) [16:45] mvo: ta [16:50] yw [16:56] it works from gdb. [16:56] mvo: any idea who to ping about this make segfault? === chihchun_afk is now known as chihchun [17:00] https://paste.ubuntu.com/14591130/ [17:30] elopio, easy review when you have a minute: https://github.com/ubuntu-core/snapcraft/pull/246 [17:31] kyrofa: was the change in the indentation on purpose? [17:32] elopio, yeah, it's not valid markdown [17:32] elopio, and we have another PR that wants it done but it hasn't been updated in a while [17:32] elopio, so I figured I'd just tackle it here [17:32] so it was a trap! [17:33] simple, but I have to read it all :) [17:33] elopio, yeah, stupid git [17:34] don't worry, I'm on it. [17:34] elopio, if you trust me, I promise all I added was the build-packages [17:35] kyrofa: ok, I will trust you but just because I'm hungry ;) [17:35] elopio, mmm === chihchun is now known as chihchun_afk [18:28] elopio: did you got an anser already? do you have some more info? so it happens on amd64 in a lxc xenial env, right? [18:32] kyrofa: I managed to automate it as I had planned with a nice local plugin for snapcraft. [18:32] kyrofa: What does not work now is the framework's policy [18:33] kyrofa: Before I had the framework policy in meta/ [18:34] Lefteris`, you'll need to specify it in the snapcraft.yaml. Did you? [18:34] same level as package.yaml, but now with snapcraft how do I do it? [18:34] (no I did not write anything in snapcraft) [18:36] in short everything I had in meta/ when I snapify a directory ... where should it be if I use snapcraft? [18:41] kyrofa, here? [18:41] enoch85, yessir! [18:44] hmm there is an integration test having `framework-policy: dir` inside a snapcraft yaml [18:44] will go with that :) [19:59] mvo: sorry, late reply. I reported this: https://bugs.launchpad.net/ubuntu/+source/make-dfsg/+bug/1536727 [19:59] Launchpad bug 1536727 in make-dfsg (Ubuntu) "make crashes in xenial lxc" [Undecided,New] [20:06] elopio, even easier than the last one: https://github.com/ubuntu-core/snapcraft/pull/248/files [20:18] mvo: hmm, just pulled an image using core rolling --oem=generic-amd64/stable --channel edge...and system-image-cli -i shows it's from jan 12, expected? [20:18] thot it might be updating soon.. [20:22] kgunn: yes, we stopped updating using system image, please use http://people.canonical.com/~mvo/all-snaps/ for now, unfortuantely the ubuntu-device-flash in xenial is not updated there is a branch review pending still [20:23] np, thanks mvo [20:23] kgunn: here is a bit of the background https://lists.ubuntu.com/archives/snappy-devel/2016-January/001400.html [21:17] hi [21:17] how do i download xenial version of snappy? [21:17] or is that not available currently? [21:17] circ-user-edG3L, use http://people.canonical.com/~mvo/all-snaps/ [21:18] kyrofa: k thx [21:18] kyrofa: are you at the ubucon? [21:18] circ-user-edG3L, note that it's under heavily development, though [21:18] circ-user-edG3L, unfortunately no. Too close to having a baby to leave! [21:18] kyrofa: oh. congratulations!. [21:18] circ-user-edG3L, haha, thanks [21:19] kyrofa: is mir spec for xenial ready yet? or still being modified?... is there a timeline for mir/xmir ? [21:19] circ-user-edG3L, kgunn is your man there [21:19] kyrofa: k thx [21:19] hello [21:20] kgunn: is mir spec for xenial ready yet? or still being modified?... is there a timeline for mir/xmir ? [21:20] dbuellough: hi [21:23] circ-user-edG3L, did you just see sergio and manik's talk? [21:23] still running :) [21:23] kyrofa: im watching it now [21:23] circ-user-edG3L, the live stream is broken :( [21:23] kyrofa: https://www.youtube.com/watch?v=1C8T4Mp9aSM [21:23] kyrofa: it works [21:24] circ-user-edG3L, terrible audio here [21:25] kyrofa: :{ [21:25] I can confirm that audio is bad [21:33] Trying to publish a snap on the store. It fails with "ERROR: manifest malformed: hooks/mir-server is empty...". How do I fix this? [21:33] My snap --> https://myapps.developer.ubuntu.com/dev/click-apps/share/4255e10a7324e99961bb6a15b0c4368692e318870c3e527339100387379a08c96802720a73162eabafce/ [21:42] circ-user-edG3L: i assume you're asking here so specific to snappy? [21:43] we have it working, just worked thru sec profiles for mir recently with the sec team...so waiting for those to flow thru [21:50] kgunn: yes. [21:52] kgunn: ok. so spec is finalized? [21:55] kgunn: or still tbd? [21:55] kyrofa: I recorded the demonstration if you want to see it [21:57] circ-user-edG3L, yes please [21:59] kyrofa: can i have your email address to send? [21:59] kyrofa: im uploading to youtube [22:01] circ-user-edG3L, just paste the url here if you don't mind, I'm sure other people would like to watch as well === devil is now known as Guest26391 === Guest26391 is now known as devil_ === devil_ is now known as devil__ === devil__ is now known as devil_ [23:48] hello! [23:49] aatchison, http://people.canonical.com/~mvo/all-snaps/ [23:50] utlemming, hey, I'm getting reports from aatchison that the vagrant images don't work. Are you still maintaining those? [23:51] sergiusens: interesting... [23:51] sergiusens: yup, looking [23:52] utlemming, also, out of the blue, do we have rolling/all snaps images for vagrant? [23:53] sergiusens: I haven't checked since last week, but if that is the default for u-d-f, then yes [23:53] I'll paste bin the results [23:53] utlemming, oh, it's not; it needs a non backwards compatible u-d-f [23:54] sergiusens: then we do not [23:55] http://paste.ubuntu.com/14593617/ [23:55] utlemming, can we use a different u-d-f to set that up? Calling a different image creator? [23:58] aatchison: windows/mac? [23:58] sergiusens: can you send me the instructions on that? Better yet, file me a bug with the instructions and I can try to get to it. [23:58] utlemming, Ubuntu 15.10 [23:59] ok [23:59] aatchison: where did you procure your version of vagrant....this looks like a bug in Vagrant/Ruby, not in the image itself. [23:59] aatchison: double checking