=== chihchun_afk is now known as chihchun
=== chihchun is now known as chihchun_afk
=== chihchun_afk is now known as chihchun
zygagood morning06:55
dholbachgood morning07:06
fgimenezgood morning07:09
=== zbenjamin_ is now known as zbenjamin
=== chihchun is now known as chihchun_afk
dholbachmvo, hey hey07:40
dholbachcan you kick off a build of https://code.launchpad.net/~snappy-dev/+recipe/snapcraft-daily?07:40
dholbachwhy is it not set to automatic?07:40
dholbachor daily07:41
dholbachor whatever the setting is07:41
mvodholbach: sure, let me check why its not automatic07:43
mvofixed and triggered a new build07:44
dholbachmvo, https://code.launchpad.net/~dholbach/snapcraft/build-deps-syntax/+merge/26985307:48
* mvo hugs dholbach07:50
ogra_ppisati, mind to take a look at bug 1491094 ?07:50
ubottubug 1491094 in linux (Ubuntu) "wifi USB dongle fails to work with musb-hdrc error" [Undecided,Incomplete] https://launchpad.net/bugs/149109407:50
dholbachmvo, once it's merged we can try to build the branch again ;-)07:51
ppisatiogra_: BBB? or rpi2?07:51
ppisatiogra_: i guess the first07:52
dholbachmvo, merged07:52
=== chihchun_afk is now known as chihchun
ogra_ppisati, yeah ...07:56
ogra_ppisati, sadly rick seems to be back on US timezone, so it might take a bit to get a full syslog07:56
ppisatiogra_: do you know if he receive just a single warning or more than one?08:04
Chipacamo'in people08:06
ogra_ppisati, i had him re-plug it a few times and he said thats all that is printed in syslo08:09
ppisatiogra_: so just a single warning, i guess08:10
ogra_as i understand it it should use the rt2800usb ... but seems the kernel doesnt even get that far08:10
ogra_(we also tried plugging with the module manually loaded which according to rick made no difference at all ... )08:11
ppisatiand on a x6 box, it worked, right?08:12
ppisatitesting on a vivid kernel would be awesome08:13
ogra_i thinnk so ...08:13
ppisatilet me add some comments to the lp bug08:13
ogra_cool, thanks08:13
ppisatithis thing08:13
ppisati"Product: Љ"08:13
ppisatithis is particularli scary08:13
ogra_thats minicom i bet08:13
ogra_(getting the actual syslog would likely help a lot)08:14
ogra_ppisati, there is another bug where your input might help ... bug 148941208:15
ubottubug 1489412 in Snappy "RPi2: connection not realiable" [Undecided,New] https://launchpad.net/bugs/148941208:15
ogra_i'm notz really sure what to do with it ... turning off turbo mode on the NIC seems to help a little, but doesnt make the issue go away08:15
ppisatiogra_: so, about the smsc9xx there08:17
ppisatiogra_: i spinned a new kernel with min_free = 32k08:17
ppisatiand i rebased on the latest 3.19 stable08:17
ppisatiogra_: but i didn't update the meta yet08:17
ogra_ah, cool, i'll make sure to pull that in for 15.04.308:17
ppisatiogra_: when is the release?08:17
ogra_my script doesnt use the meta, it just pulls the newest binary kernel from your PPA08:18
ogra_not sure08:18
ogra_mvo, do we have some schedule for 15.04.3 ?08:18
ogra_or ETA08:19
mvoogra_: I'm not aware of one08:21
ogra_ppisati, so the debian way then :) when its ready08:21
ppisatii would like to have some testing before we mass deploy08:22
ppisatiyou know...08:22
ogra_(which means if you want something specifically, we can hold it back for that)08:22
ogra_well, we can indeed ask for testing and haven people use the 15.04 edge channel08:23
ppisatiabout the first bug, i have a match for that usb vendor/product id on a 4.2 wily kernel:08:29
ppisatidrivers/net/wireless/mediatek/mt7601u/usb.c:    { USB_DEVICE(0x148f, 0x7601) },08:29
ppisatibut i don't have that match on 3.1908:29
ppisatiDocumentation/usb/hotplug.txt:    USB_DEVICE (vendorId, productId)08:30
=== greyback__ is now known as greyback
dholbachhey rsalveti, I was playing around with snapcraft a bit more - do you think I should triage the list of open bugs right now and tag the ones which have to do with first user experience or something, so we can have all of them in one list? if yes, which name should we use?09:47
dholbachmvo, ^ too09:47
dholbachhttps://code.launchpad.net/~dholbach/snapcraft/require-ant/+merge/269864 is a quick fix I could do myself :)09:47
=== chihchun is now known as chihchun_afk
guest42315Snappy Ubuntu Core on Banana Pi BPI-M2 http://forum.banana-pi.org/t/snappy-ubuntu-core-on-banana-pi-bpi-m2/43210:00
ogra_guest42315, nice !10:23
Chipacapitti: script -qfc10:33
pittiChipaca: heh, nice trick! do you know how much overhead that has?10:39
Chipacapitti: about 2310:39
Chipacamaybe 2410:39
pittiI mean per line/read/block10:39
pitti(what's that unit?)10:39
Chipacapitti: sorry, i was being mean. I have no idea of overhead, but I don't think it's much. It does create a file.10:40
pittiscript -qfc "sh -c 'sleep 30 & echo payload'" /dev/null |cat10:40
pittieven that works10:40
pittibut this unifies stdout and stderr, doesn't it?10:40
Chipacahm, probably10:41
Chipacais that bad?10:41
pittiah yes, it does10:41
pittiI need to keep them separate/intact, yes10:41
pittibut it's a nice trick anyway for other use cases, much simpler than using redirection and tail -f10:42
pittiand having to clean up tail10:42
=== chihchun_afk is now known as chihchun
pittiso thanks for the trick!10:42
Chipacapitti: and if you take it deeper?10:43
Chipacapitti: e.g. script -qfc '1>stdout 2>stderr commands here'10:43
pittiI played around with exec, redirections, fifos, bash's <(command) all morning; it's ridiculously difficult10:44
pittiChipaca: yeah, and then tail -f them both; that's what autopkgtest does on the "outside", as that only needs to be set up once per test10:44
pittibut once for every single command is heavier (and also it's brittle as hell), so I was hoping there was something simpler10:44
Chipacaah, you probably also want -e, to get the exit code of the process10:45
pittiit took me two months or so to get that "background tail -f" robust enough, and it's still far from perfect10:45
Chipacapitti: are you ssh'ing in, or how are you running these things?10:45
pittiChipaca: depends on the backend; "schroot", "lxc-attach", "ssh", or via a socket for Qemu's case (serial console)10:45
Chipacapitti: for ssh, a simple -t works10:47
Chipacaie ssh -t yadda yadda10:47
Chipacapitti: what you want is to have a tty10:47
pittiright, with PTYs there's no problem10:47
Chipacapitti: but it's simple enough, i think, in C10:48
Chipacapitti: so maybe you need to write a little C :)10:48
pittiyeah, in C it wouldn't be a problem at all10:48
pittijust that I can't assume that I'm able to compile anything on the target10:48
Chipacathen write C :)10:48
pittimy testbed could be a read-only phone or snappy without even a compiler10:48
Chipacapitti: it's standard library C; you can cross-compile it without breaking a sweat10:49
pittiChipaca: well, setting that up (lots of C code and making autopkgtest depend on lots of cross compilers and dynamically shoveling binaries to the testbed) sounds a lot harder than my ugly 15 lines of "kill leftover processes" shell hack :/10:50
Chipacaarm-linux-gnueabihf-gcc-4.9 -Wall -o blah blah.c && scp/put/cp/whatevs && etc10:50
pittimaybe perl10:50
pittiperl-base is still essential10:50
Chipacapitti: I'd expect you to say that two months ago, before spending two months on tail :)10:50
ogra_pitti, store the PIDs in a file and loop over it with kill when the parent dies  ?10:50
ogra_via a trap cleanup command ...10:51
Chipacapitti: perl works, also, i think10:51
pittianyway, quite a number of people have answered, and I found some related discussions on various forums10:51
pittiso at least I'm now reasonably sure that I wasn't missing something obvious10:51
Chipacaogra_: you'd have to strace -f to grab all the clones & forks10:52
ogra_Chipaca, well, or just start everything through a wrapper function10:53
Chipacapitti: "try" from libio-pty-perl's examples is probably a good place to start11:03
Chipacaanyway. to lunch.11:10
ogra_ogra@anubis:~/datengrab/images/snappy$ sudo ubuntu-device-flash core --channel edge --device device-pi2-0.16.tar.xz --oem pi2.canonical --developer-mode --install webdm -o kvm-rolling.img rolling12:04
ogra_[sudo] password for ogra:12:04
ogra_WARNING: this option should only be used to build azure images12:04
ogra_Determining oem configuration12:04
ogra_great ...12:04
ogra_now if it could tell me *WHICH* option "this option" is :P12:04
ogra_ah, obviously --device ...12:06
tbrdebug output is overrated12:12
rickspencer3hey, is there an easy way to change the channel that my bbb is on?12:26
rickspencer3I think I'd like to get daily updates on it12:26
ogra_rickspencer3, no idea if that works ... but try system-image-cli --switch (with a lot of -vvv to increase verbosity)12:29
* rickspencer3 tries12:29
ogra_i guess switching channels will become easier as soon as we dropped system-image12:30
rickspencer3ogra_, does it not need an argument for the channel to switch to?12:30
ogra_sure, like on the phone12:30
ogra_(up to you want you want to switch to)12:30
rickspencer3ogra_, so, I want to switch to 15.04 edge, is there somewhere that shows me how to formulate that argument?12:31
ogra_ubuntu-core/15.04/edge ...12:32
ogra_note that this channel is on manual though12:32
ogra_only gets builds if we upload changes to the PPA or before a release12:32
rickspencer3manual is good for edge, I think12:32
rickspencer3ogra_, oh, is there a better channel if I want to test latest bug fixes, etc...?12:33
Chipacafwiw, channel switching is Not Supported12:33
ogra_wily ... ubuntu-core/rolling/edge ... that has daily builds but also the highest chance of breaking12:33
ogra_Chipaca, i think it worked for me last time on RPi ... but cant roll back or some such12:34
rickspencer3ogra_, so, the argument is ubuntu-core/rolling/edge ?12:34
Chipacai'm not saying it won't work; i'm saying if it works, huzzah, but if it breaks oh well :)12:34
rickspencer3and it will get automatically update each day, and maybe break a lot?12:35
ogra_i dont remember what was broken on my RPi test ... but iirc it initially worked ... and either didnt upgrade or didnt roll back anymore12:35
rickspencer3I guess I can just get another sd card and put a more stable version on it :)12:35
rickspencer3looks like it is trying to upgrade it :)12:35
rickspencer3[systemimage] Sep 02 12:35:29 2015 (1239) Upgrade path is 155 :)12:36
rickspencer3(note the ":)" is mine12:36
ogra_well, we should add more smileys to log output :)12:36
ogra_you could then also indicate error severity through them ;)12:37
ogra_:/ <- bad ... :( <- severely bad ... X( <- totally dead12:38
rickspencer3ogra_, that would be brilliant12:39
rickspencer3or just emojis12:39
rickspencer3says it is rebooting, I'll take that as a good sign12:39
ogra_well, UTF8 is often broken in serial terminal apps :)12:39
ogra_but yeah, emojis would rock12:39
rickspencer3emoticons it is, let's get the spec started ;)12:39
rickspencer3well, it didn't reboot itself12:42
* rickspencer3 reboots, braces for impact12:42
rickspencer3ogra_, so, it rebooted and my led blinking snap started!12:44
rickspencer3I take that as a good sign12:44
rickspencer3is there a way for me to confirm that I am now on the channel that I wanted to be on?12:44
ogra_rickspencer3, sudo snappy list -v12:46
ogra_abd system-image-cli -i12:46
rickspencer3ubuntu-core 2015-07-29 4       ubuntu*12:46
rickspencer3ubuntu-core 2015-09-02 155     ubuntu12:46
ogra_seems it didnt switch12:46
rickspencer3oh well, I guess I'll just set up a new sd card with the right channel later12:47
ogra_* indicates the active one12:47
rickspencer3at least it didn't hose my board12:47
ogra_you can force the swtich ... one sec12:47
ogra_fw_printenv |grep ^snappy_ab12:48
ogra_what does that print ?12:48
rickspencer3(BeagleBoneBlack)ubuntu@localhost:~$ fw_printenv |grep ^snappy_ab12:48
rickspencer3Cannot access MTD device /boot/uboot/uboot.env: No such file or directory12:48
ogra_hmm, you said this install is old ..12:49
ogra_seems it is really old :)12:49
rickspencer3ogra_, if this is going to become a yak shaving exercise, I suggest that we call channel switching "unsupported" and I can flash a new sd card at my leisure12:49
ogra_we *do* call it unsupported ;)12:50
rickspencer3well, right, that's my point ;)12:50
ogra_but you are running into issues with an ancient bootloader setup12:50
rickspencer3ok, sounds like a reflash is in order anyway, then12:50
ogra_it wouldnt even switch on normal upgrade i fear12:50
rickspencer3are there any docs on how I would use use snappy config?12:51
ogra_well, it isnt very user friendly yet12:51
rickspencer3I want to write a snap where you can define the gpio pin to use after installing the snap12:51
ogra_you need to create a yaml file and pipe it into the command12:51
rickspencer3ogra_, so, I could write go code that takes an argument in a form post12:52
rickspencer3uses that to write a yaml file12:52
ogra_i think sergiusens had some demo code for that12:52
ogra_piping from a here doc12:52
rickspencer3then pipes that yaml file in snappy config?12:52
ogra_oh, if you want to do it on image level you should be able to define it in the package.yaml of a custom oem snap12:52
rickspencer3ogra_, no, I want to do it on the snap level12:53
rickspencer3i.e. the snap is in the store12:53
ogra_(where you would also define your app snap as a requirement)12:53
rickspencer3then it can be configured to use the specified gpio pins12:53
ogra_ah, right, yeah , then you need to create a yaml file and run snappy config manually after install12:53
rickspencer3so the end user will have to edit the yaml file on the board and run the command12:54
ogra_nah, it will be more userfriendly12:54
ogra_but this is the status quo12:54
ogra_we have to many other construction areas atm ... config is only very basic yet12:55
rickspencer3ok, I'll look at sergiusens code when he wakes up if he can point me to it :)12:56
sergiusensrickspencer3: I don't have any code12:56
* sergiusens has been awake for 3 hours already12:56
sergiusensogra_: rickspencer3 you can also do snappy install [snap] [config]12:56
rickspencer3sergiusens, I meant "wake up" in a totally figurative way :)12:56
rickspencer3sergiusens, is there documentation that I can use anywhere?12:56
ogra_ah, thats new to me12:57
sergiusensogra_: it's in the --help (snappy install --help)12:57
rickspencer3so, my idea for the snap is that when the board is ready to be connected to over ssh, it blinks a led12:58
ogra_sergiusens, yeah i tried "snappy help" :P12:58
ogra_(no, i'm lying indeed)12:58
rickspencer3is there a way that a snap could easily tell if ssh is ready?12:59
rickspencer3i.e. from Go code?12:59
sergiusensrickspencer3: so you indeed want a config to configure where that LED is12:59
rickspencer3sergiusens, yeah12:59
sergiusensrickspencer3: you would need special permissions to check it at the service level12:59
sergiusensrickspencer3: the thing you can do today is poll the port and wait for the ssh signature13:00
rickspencer3there are 2 things I don't know: 1. how to determine if ssh is running, 2. how to allow the developer to config the gpio pin13:00
rickspencer3sergiusens, I can't do that with just the normal network permissions?13:00
rickspencer3seems like I could do a loop and just ping it over localhost?13:00
sergiusensrickspencer3: yes, we don't have any security in place for fine grained networking :-/13:00
rickspencer3sergiusens, yes I can do that, or yes I can't do that?13:01
sergiusensrickspencer3: ping over localhost or over the main interface13:01
Chipacaogra_: rickspencer3: cdparanoia has emoticons for status output13:01
sergiusensrickspencer3: yes you can!13:01
rickspencer3ok, that sounds easy enough13:01
Chipaca         :-)  Normal operation, low/no jitter13:01
Chipaca         :-|  Normal operation, considerable jitter13:01
Chipaca         :-/  Read drift13:01
Chipaca         :-P  Unreported loss of streaming in atomic read operation13:01
Chipaca         8-|  Finding read problems at same point during reread; hard to correct13:01
Chipaca         :-0  SCSI/ATAPI transport error13:01
Chipaca         :-(  Scratch detected13:01
Chipaca         ;-(  Gave up trying to perform a correction13:01
Chipaca         8-X  Aborted read due to known, uncorrectable error13:01
Chipaca         :^D  Finished extracting13:01
sergiusensChipaca: that reminds me of mac system 713:02
Chipaca    c.Check(logs, DeepEquals, []map[string]interface{}{{"a": 1}, {"a": 2}})13:15
Chipaca... obtained []map[string]interface {} = []map[string]interface {}{map[string]interface {}{"a":1}, map[string]interface {}{"a":2}}13:15
Chipaca... expected []map[string]interface {} = []map[string]interface {}{map[string]interface {}{"a":1}, map[string]interface {}{"a":2}}13:15
* Chipaca goes looking for somebody to wap over the head13:15
mvipAlexander -- this is Viktor from Screenly. Sorry, forgot your nick, but please just wanted to join the channel and say hi.13:17
Chipacaasac: ^13:17
mvipThanks :)13:18
mvipOh and i see mectors is hanging out here too :)13:19
mectorsmvip of course13:20
rickspencer3o/ mvip13:21
mvipNeed to brush up on my bitchx skills. Haven't spent much time on IRC in the last decade :)13:24
bartkusedvinasHello there. I need some help that I can't find yet on Google. How do I add to may #cloud-config DOCKER_OPTIONS :/ does anyone use it?13:29
bartkusedvinasIs it possible to write to /var/lib/apps/docker/ via cloud-config?13:30
mvipAre there any examples out there of running X11 apps in snaps?13:30
mvipbartkusedvinas: isn't /var residing on /, which is mounter r/o?13:31
bartkusedvinasyou are right, it is read only13:34
tedgzyga: There is a rumor that you're working (and close to completing) pip support for snapcraft. Is it true?13:35
mvipbartkusedvinas i'm by no mean a Snappy expert, but my guess is that it's mounted as r/o on boot. As such, you can not modify that file (unless you hackishly re-mount it *inside* cloudinit)13:35
=== chihchun is now known as chihchun_afk
tedgbartkusedvinas: Snaps themselves have a configuration interface, which the docker framework should use. I'm not sure if it is though.13:36
bartkusedvinasmvip: sounds right... tho, i bet that those who are using docker want to add DOCKER_OPTIONS='-H tcp://'13:36
bartkusedvinashowever i can't find what is the right approch to do so13:37
ogra_bartkusedvinas, the shipped cloud-init is only used for bringing up the system ...13:37
ogra_not for tinkering with snaps13:37
tedgmvip: I think that deb2snap was used to make some Xapps run, but I can't think of an example.13:37
tedgmvip: Is there are particular one you're trying to run? Many have Mir backends available as well.13:37
ogra_bartkusedvinas, well, you are likely not using docker standalone ... but run something in it13:38
ogra_bartkusedvinas, for that something you would create a snap that uses the docker framework13:38
ogra_and in that snap you would set such options13:38
mviptedg: Well, yeah, have some things in mind, but was just curious if there were any examples out there. `deb2snap` seems somewhat hackish.13:38
tedgmvip: It is :-)13:39
mviptedg: :)13:39
mvipJust out of curiousity, how's Docker configured on Snappy. Is it run on the host, or is it running within a snap?13:39
mvipBecause I saw a Docker snap when I was playing with it on the RasPi13:39
=== kickinz1|afk is now known as kickinz1
tedgmvip: I wrote up an examle using Qt/QML, but that is with the Mir backend: http://gould.cx/ted/blog/Creating_a_QML_snap_with_Snapcraft13:40
ogra_bartkusedvinas, i guess kickinz1|afk could help (if he werent afk) he maintains the docker framework snap and also the owncloud snap that uses it13:40
mviptedg: Oh cool. Thanks.13:40
kickinz1mvip docker is a framework snap13:40
mvipkickinz1 ah ok13:40
mvipstill getting used to the Snappy lingo13:40
ogra_ha, speaking of the devil :)13:40
bartkusedvinasthanks. i want to try snappy as alternative to coreos13:41
bartkusedvinaswith coreos i had the following cloud-config which would open access to docker https://gist.github.com/edvinasbartkus/8ed0e61a7c7980c087f113:41
bartkusedvinasso I would be able to use deploy tools to deploy to coreos docker's my containers13:42
ogra_yeah, this is the stuff you would do in a start script of the snap i talked about13:43
bartkusedvinasi see13:43
ogra_http://paste.ubuntu.com/12253305/ i.e. this is the start script used by the owncloud snap13:45
ogra_i assume you would have something similar in a coreos-docker.snap13:45
mvipkickinz1 so if Docker is a Snappy framework, i then presume it is running on the host and then snappy is responsible for communicating with Docker, right?13:45
kickinz1mvip, yes, docker is running on the host, there are some specific apparmor profile management.13:47
bartkusedvinasI was expecting to find some easy recipie like this https://gist.github.com/edvinasbartkus/a99dd8bf12f717c7c6f5 :))13:47
mvipkickinz1 got it. Thanks.13:47
bartkusedvinas@orgra_ thanks for the hint to the script. will try to improvise my way ;)13:48
nothalbartkusedvinas: No such command!13:48
kickinz1mvip, but for now you can connect directly to docker-daemon, be aware, that if you want to do some docker build, docker load/save, you need to have your files in $HOME/apps/docker.$VERSION/13:48
bartkusedvinas@nothal yeah, i am  aware :) i was expecting just to have such easy solution :P13:48
nothalbartkusedvinas: No such command!13:48
ogra_bartkusedvinas, well, i'm pretty sure such an easy setup is possible too ...13:49
mvipkickinz1 got it.13:49
ogra_bartkusedvinas, the point in snappy is that you are completely confined ... so you would never modify any of the system setup of docker but ship your own13:50
Chipacapitti: question for you sir13:50
Chipacapitti: what should "journalctl -q -u something-that-does-not-exist" print, given the -q?13:51
bartkusedvinas@ogra_ thanks13:52
nothalbartkusedvinas: No such command!13:52
ogra_you should leave out the  @ :)13:52
bartkusedvinasoh :) sorry for being newby13:54
bartkusedvinasand even spelling like a newbie :))13:54
pittiChipaca: you mean wrt. the -- No entries --? -q is documented as "Suppresses any warning messages regarding inaccessible system journals when run as a normal user.", but I see how one would want to shut up that message too :)13:55
ogra_no worries ...13:55
ogra_usually the ubuntu bots dont use @ but ! ... nothal is special in this channel13:55
Chipacapitti: the "No journal files were found." was more the thing i was expecting it to suppress13:55
Chipacaas it stands -o json outputs json except when it doesn't13:56
Chipacamaking it somewhat annoying :)13:56
pittiChipaca: I don't get that13:56
pitti$ journalctl -q -u something-that-does-not-exist13:56
pitti-- No entries --13:57
Chipaca$ journalctl -q -u potato13:57
ChipacaNo journal files were found.13:57
Chipaca-- No entries --13:57
dholbachsergiusens, is there a switch for snapcraft to force-rebuild everything? or how do you normally use it? (rm -rf parts stage snap; snapcraft)?13:57
pittiChipaca: but this should be fixed anyway -- "-- no entries --" is quite useless for json output/parsing too13:59
zygatedg: hehe14:02
zygatedg: yes, let's talk14:02
zygatedg: I'm working on this hard14:03
zygatedg: and I'd love to have an accomplice14:03
tedgzyga: Do you have a branch pushed?14:04
zygatedg: no, I got into a few blockers, if you want we can sync and talk about that but I'm packed with meetings now14:05
zygatedg: we could invite you to a meeting in 25 minutes (monthly catch up on snappy)14:05
zygatedg: that could help as wel14:05
tedgzyga: Uhm, okay. I'm not looking for more meetings ;-)14:05
* ogra_ notes down ... ted ... looking for more meetings ... 14:06
tedgHeh, thanks ogra_14:07
zygatedg: so let's just talk here14:07
zygatedg: tell me about your approach14:07
zygatedg: and I'll tell you what I did and what I got stuck on14:08
tedgzyga: I was looking at adding a field to the python projects for adding a requirements.txt14:08
tedgzyga: Grab it, install it, party on. ← my approach :-)14:08
zygatedg: mmm14:08
zygatedg: have you run into the numerous issues in python3app implementation today14:09
zygatedg: it's pretty much all wrong now14:09
zygatedg: it bleeds the python stuff from the system into the snap stage area14:09
tedgNot sure what python3app is? python3-project ?14:09
zygaI never remember if it's -project or -application14:09
zygabut yes14:09
zygae.g. you pip install requirements and it ignores stuff you have locally installed14:10
tedgAh, I don't have anything from pip locally installed :-)14:10
zyganot from pip14:10
zygafrom the system14:10
zygaI have patched that heavily14:10
tedgSo it needs to restrict its search path. Sounds like an envvar?14:10
zygatedg: more than that14:10
zygatedg: you really need to run the python from the stage area14:11
zygatedg: there's no clean way to do it otherwise14:11
zygatedg: and you essentially need to do what virtualenv does to be correct14:11
tedgSure, but that's handled by setting up the path correctly.14:11
zygatedg: then there are some interesting issues related to that but that's14:11
zygatedg: path? yeah14:11
zygatedg: well14:11
zygatedg: no, just running the right python explicitly14:11
tedgEh, okay. That can work too.14:12
zygatedg: anyway, it's clear that you just run the right python with the right extra PYTHONHOME helpers and perhaps python -S (so that site is not imported)14:12
zygatedg: then you get empty python and pip install works14:12
zygatedg: I ran into some other issues though14:12
zygatedg: and that's where I'm stuck now (though just stuck with ENOTIME)14:12
tedgzyga: K, do you have a branch somewhere that I can look at?14:12
sergiusensdholbach: I'm rm'ing, but I do want to at least add a 'clean' statement14:13
zygatedg: I have it locally but it's all teribble as I was trying to figure out how to upstream it14:13
zygatedg: let me show you the diff in a sec14:13
zyga(really stuck in meetings)14:13
dholbachsergiusens, it looks like it's necessary to remove the sideloaded package in a snappy system as well when using snappy-remote and not updating the version number14:14
dholbachso I can't just build a new snap and snappy-remote it again14:14
sergiusensdholbach: oh, Chipaca fixed that, for rolling at least14:15
dholbachI love you guys14:15
ogra_sergiusens, oh ? including apparmor profile updating ?14:16
ogra_(i thinnk that didnt work in the past)14:16
zygatedg: can you tell me more about what test projects you are trying to bundle with pip?14:17
sergiusensogra_: dholbach the version is basically ignored and hashed, sideloading has no upgrade paths in any case, so it is all fine14:18
ogra_ah, cool14:19
ogra_in the past even for locally sideloaded snaps the apparmor profile wasnt updated when i didnt bump the version14:19
Chipacaogra_: the version in the file is ignored for sideloaded14:22
ogra_Chipaca, yup. got that now14:22
Chipacaogra_: it's a base62 hex string of the timestamp in ns14:22
Chipacabase 50, not base 62, sorry14:23
Chipacait's an apparently random string of a-yA-Y14:23
Chipacasuch that something installed later always compares as newer14:24
Chipacaas long as your clock isn't completely full of it, that is :)14:24
Chipacathis means rollback still works as expected14:24
* sergiusens reboots to enable VT on this laptop14:25
tedgzyga: Looking more at the feature genericly. I don't have a specific project I need to package.14:25
tedgzyga: To have something to play with I was looking at Jasper, but it's a huge thing to get working, but provides a nice example.14:25
zygatedg: I think it would be healthy to set up some projects as canaries, that you can build "complex" stuff14:27
zygatedg: thanks, I'll have a look at that14:27
zygatedg: as a sanity check, add checkbox-ng to that list, it's very important for an ongoing project we're doing14:27
zygatedg: one thing I think needs more help is building C extensions14:28
zygatedg: the requirements file needs to be coupled with a list of debian packages you need on the outside to build your stuff14:29
zygatedg: stanity check for that is lxml214:29
tedgzyga: How does PIP express those dependencies?14:30
zygatedg: it doesn't at all :D14:30
zygatedg: but it's very very important14:30
zygatedg: without this you won't get important things (requests)14:30
zygatedg: and it's sad but that's what it is14:30
tedgGuess we'll have to take those on a case by case basis. But it means having to parse the file by ourselves :-/14:31
ogra_hmm, funny ... why does the rolling initrd on my Pi not attempt to resize at all14:39
ogra_probably because i should update to the new initrd in the device tarball :P14:40
ogra_super irritating that /usr/share/initramfs-tools/scripts/local-premount/resize exists in my rootfs :P14:41
ogra_(but not in the binary initrd)14:42
elopiofgimenez: sorry, I stayed in bed a little later today.15:03
fgimenezelopio, np! :) we can move our meeting a bit later15:03
elopiofgimenez: thanks for the subunit reviews. I'll make the changes.15:07
fgimenezelopio, ok thanks, when you have time take a look at https://code.launchpad.net/~mvo/snappy/lp1480248-test-reenable/+merge/26917615:19
dholbachtedg, do you know what's up with https://bugs.launchpad.net/snapcraft/+bug/1491301?15:34
ubottuLaunchpad bug 1491301 in Snapcraft "qmldemo example doesn't work" [Undecided,New]15:34
elopiosergiusens: or somebody, do you know how to make something like a prerequisite branch in git?15:34
elopiofgimenez: on it. I pushed the two subunit branches.15:37
sergiusenselopio: I'd ask the lp guys, cjwatson maybe knows15:39
elopioI'll wait for them.15:39
tedgdholbach: Hmm, no. I'll blame sergiusens ;-)15:40
tedgLet me look at it.15:40
elopiofgimenez: wow, nice catch on that branch. Now I understand the mess.15:40
elopiomvo: sorry, my suggestion on https://code.launchpad.net/~mvo/snappy/lp1480248-test-reenable/+merge/269176 was wrong. I'm testing Federico's proposal now.15:40
dholbachthanks tedg15:41
dholbachanother example is broken as well: https://bugs.launchpad.net/snapcraft/+bug/149130315:42
ubottuLaunchpad bug 1491303 in Snapcraft "tomcat-maven-webapp example doesn't work" [Undecided,New]15:42
mvoelopio: no problem15:42
elopioev: about my request for git, if we want to save the master configuration in a repository, we will need to run git on the master.15:43
elopioit's run by the plugin, so we don't need to access the shell manually. But it's not useful in this case to install git in a slave.15:44
fgimenezelopio, thx maybe we could try to refactor things there15:44
elopiofgimenez: I think it's clear. The mess is because the behaviour is different in uboot, but your new comments make that ok.15:45
mvosergiusens: I plan to look into u-d-f tomorrow for bigger rootfs, I would also like to increase /boot while at it to something like 256mb, what do you think? we no longer need system-a,b15:53
sergiusensmvo: sounds good, you can also look at my branches in lp:~snappy and reuse those if you want (there's uflash and an import of mostly everything into lp:snappy)15:56
mvoogra_: if you have some you could have a look at https://code.launchpad.net/~mvo/ubuntu/wily/initramfs-tools-ubuntu-core/new/+merge/269945 at some point, just to see if there is anything obviously wrong (except for the fixmes :)15:58
mvosergiusens: oh, do you think that should be the way forward? to use uflash instead of u-d-f?15:58
sergiusensmvo: I do, I really do; it will be easier to separate old world order from the new world order16:01
sergiusensmvo: http://blog.sergiusens.org/posts/Recovering-Ubuntu-Core/16:01
sergiusensmvo: would need some polishing, but that's where I left things before the Lex sprint16:02
sergiusensmvo: I did some general snappy cleanup there (moving away from helpers) so it is not all that easy to follow; but if you want, I can get these branches up to speed with the latest trunk16:03
ogra_mvo, hmm ... dont you create a bunch of loops there ?16:04
mvosergiusens: ok, I check tomorrow16:05
mvoogra_: probably, not sure that can be avoided16:05
mvoogra_: because the os is on /writable16:05
ogra_mvo, the rootfs snap should not live in the writable partition16:05
ogra_thats a prob16:05
ogra_how would i do a factory reset ? wiping writable would wipe everything16:06
ogra_the rootfs should be a readonly img in /boot16:07
ogra_or better in system-boot16:07
mvoogra_: thats what we will do with the recovery, there will be a os/kernel in /boot to always go back to16:07
ogra_in fact i'd even say the rootfs should be a readonly img in system-boot/a or b16:08
ogra_mvo, yeah, but still that mount loop of bind mounting stuff on top of itself isnt good16:08
mvoogra_: we also have this mandate to stay within two partitons16:09
ogra_mvo, yes, readonly bits should stay in system-boot16:09
ogra_so you end up with system-boot and writable16:09
mvoogra_: if we put the entire os into system-boot we need a much bigger partition and it can no longer be vfat16:10
ogra_oh ?16:10
mvo(at least until we also move to squash/ext.img)16:10
ogra_the readonly img is biger than 1G ??16:10
* ogra_ cant imagine that 16:10
ogra_it should be a few 100 M16:11
ogra_if at all16:11
mvothe core is ~90Mb and a kernel ~120Mb16:11
ogra_yeah, thats far from the 2G limit for vfat16:11
mvoand we want to keep at least two each, so we need ~500mb16:11
ogra_right, still plenty of space16:11
mvoright, but it also means we need to move to squashfs in one step16:11
mvowe can unpack on vfat16:12
ogra_why would you unpack ?16:12
mvobecause right now the kernel and the os are regular snaps and snaps are unpacked (currently) until we move to squashfs16:12
mvowe will get there, but not right now16:12
ogra_well make it unpack a single img file then16:13
ogra_you want to loop mount that anyway16:13
ogra_no matter what the format is16:13
ogra_i dont think the loop you create currently is safe at all16:14
mvothat could be a way, yes. I would prefer to keep snaps the same, this is what I dislike about the approach16:14
mvoogra_: unsafe in what way?16:14
ogra_dunno, i have to think about it but it makes me twitch heavily16:14
mvoogra_: ok16:15
ogra_i did something similar for the first iteration opf the phone and stgraber pointed out a lot of flaws with that16:15
* ogra_ will have to dig up that conversation 16:15
beunoah, you might be16:16
mvoogra_: that would be nice, I obviously do not want unsafe stuff, but I also want to understand whats unsafe about it16:16
mvoogra_: I can ask him directly too if thats helpful (but after dinner)16:16
ogra_probably even better having him take a look at the code16:17
mvogreat, I will ask him16:17
ogra_it makes me twitch heavily :)16:17
mvothanks for the feedback!16:17
mvowell, we will get to the squashfs eventually :) it will all be good. and to the recovery and all the other missing bits16:17
ogra_(even though i did something similar with the loop mounted img files on the phone )16:18
mvobut yeah, if its not safe, its not safe and we need to skip the step16:18
ogra_i dont see a need for squashfs here16:18
ogra_its a nice extra but not really necessary16:18
ogra_the img files can even be ext216:18
mvoogra_: well, s/squashfs/binary-images/16:18
ogra_(and probably should, for speed)16:18
mvostgraber: so if you have a moment at some point, a quick look at https://code.launchpad.net/~mvo/ubuntu/wily/initramfs-tools-ubuntu-core/new/+merge/269945 would be cool and how horrible (or not) the approach is16:19
* mvo gets dinner16:19
ogra_mvo, one thing that comes immediately to mind is that i can indeed edit content in /writable/system-data/os/ubuntu-core.sideload/${snappy_os}/*  at any time, it isnt actually readonly only the bind mount on top is16:25
ogra_so i have always full write access to the original files16:25
* ogra_ commented too 16:45
* tedg thought some of these nicks weren't LP bots, apparently was wrong.16:46
sergiusensmvo: still around?20:00
mvosergiusens: yes, but almost ready for eod20:07
sergiusenspindonga: did you log a bug about the forking issue in the systemd unit?21:48
sergiusenspindonga: can I have the number? I can't find it :/21:48
* sergiusens sucks at bug searching in lp21:48
* pindonga thanks firefox for remembering seen urls21:49
ubottuLaunchpad bug 1490574 in Snappy "add support for raw systemd unit files" [Critical,Triaged]21:49
sergiusenspindonga: ah, the title was ignored by my eyes :-) I thought we just wanted to support forking processes21:50
pindongasergiusens, we probably do21:51
pindongait's just I didn't enough it when I filed the bug :)21:51
sergiusenstedg: btw https://code.launchpad.net/~sergiusens/snapcraft/ubuntu-core/+merge/269938 feel free to slaughter it ;-)21:56

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