[01:49] elopio, I pushed an integration test [01:50] * sergiusens waves goodnight and calls it a day === chihchun_afk is now known as chihchun [05:59] good morning === shuduo-afk is now known as shuduo [06:56] kgunn: just want to update what i did yesterday and today. i finally made qt clock demo work with vm. the early error msg might be lead by network problem. I used command line to install mir.mvp-demo few times and got it installed at last. then I followed Darren Landoll's method to change mir-run and install qt clock demo and see it run perfect now. [07:05] good morning [08:04] oh, that's nice [08:05] hm? [08:05] fmt doesn't check the return value of its internal buffer handling [08:05] ie, do fmt.Println or fmt.Printf [08:06] morning all :) [08:06] * Chipaca reads further [08:07] ah, it's not a bytes, but a []byte with their own implementation of write that never returns error [08:07] ok, fair enough [08:09] ooh, and bytes.Buffer's write also has “err is always nil” [08:09] * Chipaca throws away some ifs === bigcat__ is now known as bigcat [09:03] Good morning all; happy Tuesday, and happy Brandied Fruit Day! 😃 🍒 [09:13] jdstrand: ogra_: mvo: https://code.launchpad.net/~chipaca/snappy/config-modules/+merge/274987 [09:13] jdstrand: ogra_: as it says in the branch, please do comment [09:13] looking for feedback [10:08] Chipaca, why is potato a boolean ? [10:09] * ogra_ would expect a list instread ... [10:13] ogra_: because you potato, or you no potato [10:13] :-p [10:13] ogra_: why a list? [10:14] (if you say the above with a pseudo-latvian accent, we could do the whole shtick) [10:14] Chipaca, because for i.e. firewalling i have to load a bunch of modules having a list makes it a lot easier for the admin [10:15] vs calling the same command over and over to load them all [10:15] * ogra_ commented on the MP [10:16] i'll reply in the mp for jamie's benefit [10:18] ogra@anubis:~/datengrab/images$ ls /lib/modules/3.13.0-65-generic/kernel/net/netfilter/|wc -l [10:18] 115 [10:19] there is a module for each and every sub-function of the firewall ... [10:21] ogra_: replied [10:27] me too :) [10:34] ogra_: thanks! [10:34] i'll leave it un-topapproved so jamie can get a word in [10:34] yep [10:34] @remindme 3h top-approve modules config [10:34] Chipaca: No such command! [10:34] needs the core-config change too anyway [10:34] yes [10:37] ogra_: no time like the present :-p [11:06] Chipaca, hey, mind looking at https://code.launchpad.net/~sergiusens/snapcraft/maven/+merge/274964 [12:23] mvo, if you don't mind https://code.launchpad.net/~sergiusens/snapcraft/1507814/+merge/275021 your input would be good [12:30] * Chipaca branches *all* the things [12:36] Chipaca: and I will start with the make it things instead of if a|b|c{ [12:39] mvo: is that a reasonable thing to do this week? [12:39] i mean, me no likey, but deadlines [12:40] Chipaca: good point, you seemed to suffer so much during the reviews that it looked like a good idea [12:40] Chipaca: let me time box it [12:40] Chipaca: and at least outline it [12:41] mvo: ok. I made my suffering explicit just so you knew i wasn't ok with it being like that forever, not for you to fix it now :) [12:41] mvo, apt has a big ft warning saying it shouldn't be used when scripting; is that no longer the case? [12:41] s/ft/fat/ [12:42] ah, apt install doesn't just search [12:43] sergiusens: oh, well, its a bit of a cry-baby [12:43] sergiusens: but it should be ok, no plans for changes in the cli for 16.04 [12:45] mvo, maybe a different MP? This doesn't look very nice http://paste.ubuntu.com/12876464/ [12:47] sergiusens: hm, ok [13:01] mvo, there, fixed; maybe approve now :-) [13:43] mvo, now that the meeting is over, approve my MP :-) [13:45] sergiusens: sorry I misclicked and did not notice [13:45] sergiusens: this was never "needs-fixing" [13:46] mvo, ah, well I abide to what the reviewer told me or it would be just bureaucracy to ask for reviews :-D [13:47] ogra_: can you help me making sense of this? "Debian supports the runtime modification of the Device Tree which is a feature required by BeagleBone-IO" [13:47] that's the dtoverlay you enabled for rpi, right? [13:49] elopio, where is that from ? [13:49] oh, BBB [13:49] ogra_: https://github.com/julianduque/beaglebone-io [13:49] well, thats a BBB feature [13:50] i doubt the Pi can support it [13:50] BBB does that through capemgr ... (which seems to have gotten a replacement in 4.2 that ppisati perhaps knows the name for :) ) [13:50] ogra_: davidm was trying to make a library to have standard GPIO across boards. [13:51] ogra_: Focusing on the 96boards, but might be something useful. [13:51] ogra_: I'm not looking after pi support for it. I just want to run beaglebone-io on bbb. But U didn't understand that requirement. [13:51] I don't have a /sys/devices/bone_capemgr.* [13:52] s/U/I/ [13:56] elopio, right, thats a hack ... and it got replaced by some new mechanism in the mainline kernel afaik [13:56] (capemgr is a hack i mean) [13:57] ogra_: ok. Let me give it a try. [13:58] tedg, i wish him good luck with that :) (given that each and every board manages them different and in different layers of the system) [14:01] ogra_: Yeah, his goal was to align it with what users see on the board. i.e. on the header the first pin, because that could be any number of GPIO. [14:01] ogra_: cape manager is essentially back, now that it's mainline [14:01] tbr, ah, then i dont get why we dont have it in our 4.2 kernel [14:01] check with rcn? [14:01] probably different api/incantations now [14:02] ppisati, ^^^ ? [14:03] the official debian bbb images with 4.x kernel should support dynamical dtbo loading [14:22] ogra_: nope, there's no cape manager in 4.2 [14:23] if they pulled it in 4.3+, i didn't notice that [14:23] but i might have missed it [14:24] ppisati, well, i wonder how to dynamically load dtbs then [14:26] ogra_: i don't think there's any replacement fir that [14:26] ogra_: my bet is tht they are running with a patched kernel [14:26] well, see what tbr said above ... seems rcn-ee might know something here [14:26] if we need a patch we're indeed screwed [14:32] tbr: i don't see the capemanager in mainline, where is it? === chihchun is now known as chihchun_afk [14:46] ppisati: I'm not familiar with details, but I know overlays should work in 4.2 or 4.3. ask rcn-ee for details, he knows which patches he needs on top of which kernel version [14:46] or just look at the BBB scripts [14:47] tbr: overlays are a way to change a single field in a DT node, cape manager builds on that [14:47] tbr: and AFAICS, cape manager is not in mainline [14:47] ppisati: I didn't say that the 3.8.x kernel cape manager was 1:1 available in mainline now... [14:49] fgimenez: my only complaint about that waffle is that it's not free. Doesn't taiga give us something similar? [14:49] I do know though that rcn has been working on making things work in a *similar* fashion with his builds of mainline for BBB which seemed to have a fairly thin patchset (composition unknown) [14:51] elopio, mm it's free unless you use github enterprise https://waffle.io/pricing. maybe taiga also integrates with github, i'll take a look [14:52] fgimenez: gratis, no libre :) [14:54] elopio, yep :) iirc there was another project like this hosted in github itself, let me check [15:00] elopio, this one has an Apache License https://github.com/TWtablero/tablero, haven't tried it though [15:00] sergiusens: tedg: nexe creates a single file for nodejs projects. That seems like a nice shortcut :) [15:00] https://github.com/jaredallard/nexe [15:02] Seems like wrapping your JS in Python might be the definition of crazy :-) [15:03] Though certainly up for stealing code ideas. [15:04] tedg: but if he wraps the python in bash it's ideal right? [15:04] I think it should *compile* the Python to Bash. [15:06] all the cool kids are compiling interpreted languages. [15:46] Hi [15:46] I cant run apps under docker :( [15:47] This is my system information: http://paste.ubuntu.com/12877538/ [15:47] I cant start cgproxy, here is the status: http://paste.ubuntu.com/12877531/ [15:48] And this is my apps and their ports: http://paste.ubuntu.com/12877540/ [15:48] Docker cant run apps, why? [15:49] How can I run apps under docker? [15:50] hmm, i don't remember who worked on the docker snap [15:50] Rlyeh, you mean ther eis no owncloud running if you try to connet to it ? [15:50] works fine here [15:50] Yes [15:50] kickinz1|away, ^ [15:51] I'm in Iran, recently, docker applied some restriction for Iran and I cant login and pull images [15:51] However owncloud is a local image [15:51] But I cant run it [15:51] no, the snap pulls from docker i think [15:52] you should see info in syslog [15:52] and it should tell you there if it cant startz [15:52] *start [15:52] Snappy downloaded the image, it was about 22MB [15:52] thats only the env ... it then pulls the docker owncloud install on first startup i think [15:53] OK, let me check the log... [15:53] in any case, kickinz1|away is your man ... he m,aintains both packages [15:54] theoretically snappy install docker and then snappy install owncloud should give you a working owncloud install though [16:04] hmpf ... [16:04] i think wer need to do something about our initrd ... [16:05] booting with no root= arg set drops me into a shell ... while i would expect it to reboot on panic instead [16:06] mvo, i'm wondering, does dropping to an initrd shell ever make sense in snappy ? i wonder if we should rewrite the panic() function for this to force a reboot instead [16:08] why am i on this channel??? wtf [16:10] ogra_: there is a open bug about this and pitti has some suggestions what we can do to make it do a reboot instead [16:10] Ogra_, I didn't see anything wrong, but can you please check my log? http://paste.ubuntu.com/12877687/ [16:10] well, we just need to replace initrd's panic function [16:10] Rlyeh, syslog, not dmesg [16:10] dmes will only have kernel messages [16:11] nothing from services usually [16:11] By what command? syslog is not valid [16:11] sudo less /var/log/syslog ? [16:11] or some such [16:25] oga_, you were right, the package was not pulled: http://paste.ubuntu.com/12877810/ [16:25] Damn it, why did they blocked Iran? :( [16:37] ogra_: do you recall offhand why we have pycurl on the image? [16:38] i think that predates me [16:38] but let me bet .,... zyga ? [16:42] Chipaca, hmm, was snappy itself not initially python ? [16:42] yes [16:42] could be a leftover from that [16:42] we've got pycurl, urllib3, and requests [16:42] to download snaps [16:43] Chipaca, I know [16:43] Chipaca, system image [16:43] sergiusens: it uses all those? [16:43] oh [16:43] Chipaca, it uses pycurl [16:43] ah, k [16:43] and requests? [16:44] Chipaca, is that directly seeded [16:44] dunno :) [16:44] likely not [16:44] i just happened to ls /usr/lib/python*/dist-pacakges [16:44] and saw more than i was expecting [16:44] including jinja (!??) [16:45] Chipaca, urllib3 is a dep for requests; I don't think anyone is using urllib3 directly [16:45] ah, ok [16:45] we just need to figure out who uses requests :-) [16:45] ogra_, where is the package manifest for the latest build? [16:46] just ubseed it and see what fails ;) [16:46] http://cdimage.ubuntu.com/ubuntu-core/daily-preinstalled/current/wily-preinstalled-core-amd64.manifest [16:46] ogra_, it isn't directly seeded, is it? [16:46] wily [16:46] http://cdimage.ubuntu.com/ubuntu-core/vivid/daily-preinstalled/current/vivid-preinstalled-core-amd64.manifest [16:46] vivid [16:47] Chipaca, ogra_ so cloud-init depends on python3-requests ;-) [16:47] and no, i dont see it in the seeds [16:47] oh man === Guest83_ is now known as Norbell === greyback__ is now known as greyback [18:57] ogra_: I suspect there is something with the resize script, if I remove system-ab my system hangs forever in what appears to be the resize script. is there a way to enable debug mode, i.e. see what its doing? [21:00] sergiusens: so y;day we were discussing how snapcraft shouldn't look outside it's proj directory for pulled libs...and i ran into something just want to chat thru [21:00] so for mir, i got snapcraft to the point of actually compiling on my wily machine (where boost is installed on the system) [21:01] so went to lxc for vivid, started there...didn't get past cmake... [21:01] well, mir has a FindBoost.cmake that searches for boost based on $boostroot or some such, if it can't find it goes to /usr where it finds it... [21:02] in this case, should i simply create a plugin extension specific to mir, to pass in a Boost_DIR= as part of the cmake .. call ? [21:03] also...kind of interesting that the cmake file went and looked for stuff outside the project dir, not sure how snapcraft would fight against that... [21:08] kgunn: We can't fight against it other than when you push it to LP it won't work. [21:09] kgunn: The cmake plugin has a parameter I believe for build flags, let me check. [21:10] tedg: i mean that's not awful...but if i hadn't tried it in lxc, i wouldn't have realized [21:10] kgunn: Yeah, it's an array, configflags, those will get passed to cmake [21:10] I think we need to figure out a warning or something. Not sure how we'd implement it. [21:10] But yeah, I've had that problem. [21:14] Hi, is there documentation on making a snappy package for a docker image? [21:32] I think that kirkland wrote a blog post on that... let me look. [21:32] another thing is, is there a way to install ruby? :/ `search` isn't showing anything. [21:34] gchristensen: You wouldn't install ruby in the base system, it would be included in the application that uses it. [21:35] Ah, there it is. gchristensen: http://blog.dustinkirkland.com/2015/07/prime-time-docker-juju-and-snappy.html [21:36] ok, thank you for the help and information tedg. I don't think snappy is a good fit for this project, though. [21:36] cheers, good luck -- I look forward to checking it out again in a year or so. [21:44] tedg, kgunn the best way to circumvent that is to put snapcraft under apparmor and restrict access to /include and /usr/include ;-) [21:45] not the best way, just a quick way [21:45] Yeah, I figured that wouldn't be flexible enough. As we'd want the libc headers from there, for instance. [21:45] Perhaps if we could make it only give a warning or something like that. [21:46] right