[00:00] SuperJonotron: i'm not sure how that's supposed to work; my initial response would be install it as the root user, rather than as your regular user [00:00] SuperJonotron: but that seems less than ideal [00:01] that is actually how i have to install it but for some reason when you run the snap as sudo after that it doesn't work unless it's the user who calls it [00:01] SuperJonotron: you *may* need to pass -H to sudo? to use root's home [00:03] tbh, this feels like an odd corner case to me -- this doesn't sound like a normal application, if it rqeuires root [00:03] the software is bound to hardware, the hardware it needs access to requires root access [00:03] SuperJonotron: i believe normally you'd write/request an itnerface to the hardware [00:03] and then connect your snap to that interface [00:03] pass -H where in the process? in the method calling sudo or sudo -h [00:04] SuperJonotron: can you run the snap (as installed by the root user) as the root user? First run `sudo -s -H` to switch to the root user's home and then try to run the snap? [00:04] i'm rebuilding it now because I was trying something else, i'll test in just a sec [00:05] SuperJonotron: e.g., you can see for other hardware access stuff, there are interfaces at http://snapcraft.io/docs/reference/interfaces [00:05] my thinking is based purely on my understanding of snapcraft/snaps, not on any actual experience connecting to hardware, to beclear :) [00:07] even when installing the snap as root and going to it's home directory, it still say command not found [00:08] SuperJonotron: what ubuntu are you on? [00:08] 16.04 [00:08] SuperJonotron: echo $PATH as root? [00:08] SuperJonotron: is /snap/bin there? [00:10] it is not [00:10] so that's probably that problem [00:11] I do know the bash command that is requiring the root so if you know of any of those interfaces or an equivalent non-root of dmidecode i could make this snap happy [00:11] SuperJonotron: so you could try running /snap/bin/ as root [00:13] yup, that works but doesn't recognize dmidecode [00:14] SuperJonotron: right, did you put dmidecode into the snap? [00:15] SuperJonotron: just as quick check, do youhave /etc/profile.d/apps-bin-path.sh ? [00:15] SuperJonotron: that's what i believe suffixes /snap/bin into $PATH for all users [00:15] i definitely didn't put dmidecode into the snap [00:16] SuperJonotron: then your snap won't see it :) [00:16] SuperJonotron: your snap only has what is in the snap [00:16] not entirely sure wha that syntax would look like at the moment [00:16] SuperJonotron: e.g., you'd put stage-packages: dmidecode [00:16] http://snapcraft.io/docs/build-snaps/syntax [00:16] SuperJonotron: for a dmidecode part [00:17] SuperJonotron: you basically can't rely on anything in the host system being present in your snap (well, you can in some cases, but for clarity it's probably best not to) [00:17] yes, i do have that file [00:17] yeah, i've looked through the docs quite a bit [00:18] i just meant i don't know how to get dmidecode defined in the yaml immediately [00:18] I can start breaking things until i get it right [00:19] SuperJonotron: ok, not sure why root wouldn't pick that up -- you did `sudo -s -H` to switch to root? [00:19] yup [00:20] strange [00:20] let's leave it for now, separate issue :) [00:20] SuperJonotron: right, so you'd need to ensure that your yaml specifies dmidecode (and i guess any other dependencies you might have) are in the snap [00:20] switched to root, then did /snap/bin/ and can't pick up the dmidecode [00:20] yeah, that's the last one, just added it [00:20] everything else works fine [00:21] SuperJonotron: cool, I expect that shoudl work once it's built [00:21] i meant added it to the app, not the yaml [00:21] SuperJonotron: sorry? [00:21] SuperJonotron: can you pastebin your yaml? [00:21] added the dependency within the app, still haven't figured out how to include it in the yaml so it would exist in the snap [00:25] any thoughts on including that package into snappy? [00:25] SuperJonotron: sorry, i asked you to pastebin your yaml? [00:25] SuperJonotron: can't really help in the abstract :) [00:25] missed that, one sec [00:27] name: mysnap version: 1.0 summary: A snappy implementation description: Runs my snap confinement: devmode apps: niagara: command: bin/start.sh parts: niagara: plugin: copy files: appdata: app/ start.sh: bin/ [00:27] SuperJonotron: no [00:27] SuperJonotron: pastebin [00:28] SuperJonotron: paste.ubuntu.com, please [00:29] i've never used that before i pasted it in, now paste it back here? [00:30] SuperJonotron: ah sorry, it should provide a URL [00:30] SuperJonotron: paste that URL here [00:30] you can also do `cat snapcraft.yaml | pastebinit` if you have pastebinit installed [00:31] http://paste.ubuntu.com/23286932/ [00:32] SuperJonotron: just as an fyi, i think 'application' should be 'application:' (it might just work as-is, but is non-standard) [00:32] SuperJonotron: ok, so add under files: [00:32] i modified it slightly to remove specifics of the application, the original has it [00:32] SuperJonotron: stage-packages: dmidecode [00:32] at the same level of indentation as files: [00:32] SuperJonotron: ah ok [00:34] so like this? http://paste.ubuntu.com/23286946/ [00:35] SuperJonotron: it might be a list, so maybe [dmidecode], or [00:35] stage-packages: [00:35] - dmidecode [00:36] it was the second one [00:36] rebuilding now to test [00:36] SuperJonotron: so what that will do, is ensure that the ubuntu package 'dmidecode' is installed for that 'part' [00:36] SuperJonotron: so it will now be part of your final snap [00:39] ran as sudo /snap/bin/ [00:39] dmidecode: command not found still [00:39] SuperJonotron: did you install it again? [00:40] yup [00:40] just did it again to make sure [00:41] same results [00:41] SuperJonotron: can you pastebin `unsquashfs -l ` ? [00:41] SuperJonotron: how are you invoking dmidecode? (absolute path versus just cmd name?) [00:42] just command name [00:42] SuperJonotron: ok [00:42] what was app.snap in reference to [00:42] SuperJonotron: when you build a snap, a .snap file is created [00:42] i get a Read on filesystem failed because is a directory when i just put the app name [00:42] yup [00:42] give unsquashfs -l the path to that .snap file [00:44] it just goes to a blank terminal line '>' [00:44] SuperJonotron: unsquashfs -l /path/to/app.snap ? [00:45] yup [00:45] SuperJonotron: can you c&p the exact command you are running [00:45] just typed it again in case it was a type-o, worked that time [00:45] SuperJonotron: ok [00:46] SuperJonotron: and can you pastebin the current yaml file too [00:46] yeah, i can see the dmidecode pieces in there which i'm sure is what you wanted to see [00:46] SuperJonotron: well, i want to see the path, etc [00:47] http://paste.ubuntu.com/23286972/ [00:47] SuperJonotron: can you update your script to call /usr/sbin/dmidecode? (Or ensure that /usr/sbin is in PATH) [00:48] yeah, i'll have to rebuild the application before i can rebuild the snap so it'll take a few min [00:49] SuperJonotron: ok [00:55] no luck, sudo: /usr/sbin/dmidecode: command not found [00:56] SuperJonotron: wait, you're runing under sudo still? [00:56] SuperJonotron: please run *as* root, just to be sure [00:56] SuperJonotron: i probably have to go soon, getting late here, just fyi [00:57] i can't switch to root with su [00:57] don't have root password [00:57] SuperJonotron: sudo -s -H [00:57] SuperJonotron: as i said earlier [00:58] sorry, long day here as well, switching between too many things [00:58] SuperJonotron: np :) [00:58] switched to root with sudo -s -H [00:59] SuperJonotron: ok and now run /snap/bin/ [00:59] did, same error [00:59] SuperJonotron: can you pastebin it? [00:59] h/o, i still had sudo in the bash script calling that [01:00] let's see if rebuilding without fixes [01:01] if this does at least allow it to run, any thoughts on fixing this for non root user? [01:02] SuperJonotron: not off the top of my head probably better to ask during the day US time [01:02] can do [01:03] no luck [01:03] SuperJonotron: can you pastebin the exact output? [01:03] the whole output won't mean anything to you since it's just the applictions startup output [01:04] SuperJonotron: the error line [01:04] it just prints this out: sh: 1: /usr/sbin/dmidecode: not found [01:07] SuperJonotron: so there are ways to add a shell option to your snap [01:07] https://bugs.launchpad.net/ubuntu/+source/snapcraft/+bug/1615163 e.g. [01:08] Bug #1615163: shell boilerplate could be automatically added when building in devmode [01:08] SuperJonotron: and then you can run .shell [01:08] which will put in bash running in the env of your snap [01:08] you could see from there if you can run the commands you expect [01:08] i'll bookmark that to continue testing tomorrow [01:08] thanks for the help [01:09] SuperJonotron: i'll be around again then to help [01:09] awesome, thanks again === JanC is now known as Guest21552 === JanC_ is now known as JanC === cpaelzer_ is now known as cpaelzer [06:03] PR snapd#2113 opened: Add i2c builtin interface [06:14] good morning [06:18] hey dholbach [06:19] salut didrocks [07:28] svij, taskwarrior can be removed from https://github.com/ubuntu/snappy-playpen/wiki/SandPit - right? [07:28] dholbach: yes [07:28] great, doing that now [08:41] PR snapd#2114 opened: tests: add readme about spread's external backend [09:18] Hi all, I want to snap https://github.com/broadinstitute/wdltool [09:19] should I build it myself, or is there a way to tell snapcraft to pull the release .jar from github directly? https://github.com/broadinstitute/wdltool/releases [09:19] I don't have any experience with building .jar files myself [10:00] PR snapcraft#859 opened: Set GOBIN in go plugin build environment [10:04] ogra_: o/ [10:04] um === john is now known as Chipaca [10:04] heh [10:05] ogra_: you know a lot more sh than I. Is there a way to make /bin/sh behave wrt signals? [10:05] one sec [10:05] ogra_: case in point: a script runs a; b, you sigint, a exits on the signal, sh still runs b [10:06] bash behaves properly and will exit if a doesn't catch (or catches and reraises) the signal [10:06] Chipaca, http://paste.ubuntu.com/23288279/ [10:07] use trap ... [10:08] that seems to be the opposite of what I want [10:08] :-) [10:08] Anyone know if there is a way to set environment variables when invoking qmake? qmake FOO=bar doesn't seem to work, and I don't know the tool. [10:08] trap b INT [10:09] ogra_: maybe I'm stating my problem wrong. Give me a while and I'll show you a testcase. [10:09] (a should just exit anyway, then the trap runs b on INT) [10:11] Or maybe this qmake project is just broken :-( [10:12] stub, try prefixing qmake with the var instead ? [10:13] yer, but I can't do that with the qmake plugin so am working around it with a custom plugin [10:14] * ogra_ would just use the make plugin and call qmake from the build target with all env needed [10:14] I'm trying to ween myself off from that approach :D [10:14] heh [10:15] Although I am slightly tempted with a bash plugin, that lets you embed a shell script in the yaml. [10:16] Chipaca: WFM [10:16] $ sh -c 'sleep 5; echo a' [10:16] ^C [10:16] $ [10:16] yeah. test case coming soon. [10:16] stub, i think there are a few PRs on github with that approach [10:17] Chipaca: my theory is you have process A in a different process group for some reason (or is it session? I can never remember) [10:17] I could write a part that generates a snapcraft.yaml and calls snapcraft on it [10:17] whatever terminal signals are delivered to the whole of [10:17] * cjwatson waves hand vaguely at the copy of Stevens somewhere behind him [10:23] cjwatson: ogra_: so, given the following go program (could be in C, but I had the go one closer to hand): http://pastebin.ubuntu.com/23288428/ [10:23] cjwatson: ogra_: and http://pastebin.ubuntu.com/23288429/ [10:23] cjwatson: ogra_ there are four modes, with different behaviour [10:23] bash does the right thing as I understand it [10:23] sh does not [10:23] save 'em as /tmp/q.go and /tmp/q.sh & give it a try [10:26] the idea is, run q.sh, hit ^C when it prints the first text, observe [10:26] which particular arg demonstrates the difference? [10:28] cjwatson: with both catch and ignore, where the script should continue, it exits instead [10:32] why would the script continue at all your ^C goes to the shell [10:32] I think this is often what people want for scripts, in order that they can ctrl-c something they're running interactively and not have to repeatedly hammer ctrl-c to get the whole script to exit [10:33] bash happens to ignore SIGINT itself for reasons I haven't unpicked [10:33] oh [10:33] i wasnt aware of that ! [10:33] this is about WIFSIGNALED and WTERMSIG [10:33] but you can either do 'set -m' in sh to turn on job control, which has the effect of making it have the behaviour you're asking for [10:33] which it seems sh isn't looking at [10:33] * ogra_ actually hasnt used #!/bin/bash in years [10:34] or probably do something with trap [10:34] to get sh to ignore SIGINT [10:34] ogra_: wrt why: imagine the process you launched changed the behaviour of ^C (e.g. because you launched emacs) [10:35] whats emacs ? [10:35] :P [10:35] no, it has nothing to do with WIFSIGNALED/WTERMSIG; the terminal-generated signal is being delivered to *both* processes (the shell and the Go program) and it's a question of what the shell does with the SIGINT signal that it receives itself [10:35] ogra_: an embedded operating system for 128-bit arm boards [10:35] and i thought a backend for organist :) [10:35] *organists [10:36] cjwatson: but bash isn't ignoring the ^C, it's correctly stopping the script if the process dies from the signal [10:36] cjwatson: i.e. in the other two cases [10:37] in any case, as you say "set -m" makes sh do the (imho) right thing wrt this [10:38] cjwatson: thanks! [10:42] Chipaca: I misspoke when I said "ignore"; bash installs a custom handler for SIGINT, I haven't quite unpicked exactly what it does [10:43] it's nevertheless true that the signal is delivered to the whole foreground process group [10:44] I'd suggest playing with things under "strace -I4 -f" to see more detail of what's going on [10:44] cjwatson: doesn't sh also have a handler, to wait until after the running command finishes to exit? [10:44] cjwatson: ok :-) [10:46] yeah, I guess it has slightly different semantics. I suspect this is not well-specified: the closest I can find so far is http://pubs.opengroup.org/onlinepubs/9699919799/utilities/V3_chap02.html#tag_18_11 [10:47] the relevant effect of set -m here is that the subprocess gets run in its own process group [10:48] job control is really mainly for interactive shells though; I don't recall whether there's weird behaviour from trying to use it in scripts [10:48] you'll probably get some spurious output at least [10:49] hmm, i there anything special i need to do to make ubuntu-image use core instead of ubuntu-core ? [10:49] ok. Once I pop this out of academic-ish and back in to "why does systemd-activate not exit when I ^C if I'm running it via ssh" we'll see [10:49] it's been fun digging into this anyway [10:50] i installed snapd and snap-confine from proposed and use the ubuntu-image trunk tree to build [10:50] cjwatson: i got here via https://www.cons.org/cracauer/sigint.html fwiw [10:50] ogra_: ubuntu-image ships its own snap [10:50] ogra_: snap-the-binary [10:50] ogra_: I suspect you're using an old ubuntu-image [10:50] hrm, so how do i build rom trunk now [10:51] *from [10:51] Hi all, I'm trying to make a snap by downloading a .jar directly, what plugin should I use? I tried 'copy' but i get a "no handler to manage source" error [10:51] Chipaca, i dont, i directly run out of trunk to get the very latest [10:51] ogra_: how does trunk compare to the edge snap? [10:51] not sure [10:52] ogra_: me neither [10:52] mvo didnt leave any docs, i thought using the proposed packages and latet trunk would suffice [10:52] and I'm not about to ask mvo after the week he's just had [10:52] *latest [10:52] yeah [10:52] same [10:52] Chipaca: looks like https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=683671 too [10:52] ogra_: when you say trunk, where're you getting it from? [10:53] github upstream trunk [10:53] cjwatson: heh [10:53] the same thing that barry uploaded to the archive yesterday [10:54] ogra_: and what snap is that using? [10:54] Chipaca: of course there's also the underlying bug that your test shell script isn't using set -e ;-) [10:54] Chipaca, whatever is on the host indeed [10:54] cjwatson: sssh :-D [10:55] ogra_: so your snap is probably pre-core [10:55] well, i hopees that one was in proposed [10:55] *hopd [10:55] ogra_: pre-core is like deathcore but with simpler chords? [10:55] sigh [10:55] this kbd didnt survive 1,5 years ! [10:55] half the keys dont work [10:56] ogra_: no, we've not started the SRU dance of the FF'ed snap yet [10:56] Chipaca: sigint.html is interesting, thanks [10:56] Chipaca, well, given that yakkety is alreeady frozen i'd have expected that all needed bit are in there by now [10:56] and likewisse in proposed in xenial [10:57] ogra_: i don't know [10:57] (literally) [10:57] (else you wont be able to build images in yakkety) [10:57] ogra_: that is: I don't know if the last SRU was cut with the core change or not [10:57] yeah [10:57] ogra_: it's been crazy :-) [10:57] yakkety is overrated :-p [10:58] well, devs use it i guess [10:58] (i dont though ... admittedly :P ) [10:59] the prob is that mvo actually patches the ubuntu-image snap ... but not with all patches i fear [10:59] this is the first cycle since 2008 that i haven't switched at or before the alphas, i think [10:59] (because our focus was on 16.04) [10:59] iirc his version still builds 4GB images or everything [10:59] *for [11:00] ogra_: if it's any help, afaik barry also can upload the snap [11:00] well, he wont know either which patches are used and which arent [11:00] the wondeful world of binary only packages :P [11:02] * ogra_ tries to find a new kbd in the house ... half the char of my password dont work ... igh [11:10] How can I create a snap using an existing jar file? I keep getting "no handler to manage source" [11:20] trijntje: have you read up on the java parts? [11:21] trijntje: http://snapcraft.io/docs/build-snaps/parts#snapcraft-for-java-maven-or-ant seems to have some info [11:22] if that's not enough, hang around for a while and somebody who knows snapcraft and/or java might be able to help [11:23] Chipaca: I don't want to build the .jar myself, I just want to download the .jar and put it in the snap. The project I want to use requires sbt [11:24] trijntje: I think there's a "copy" plugin, maybe that works? [11:26] Chipaca: when I tried that it said it had been replaced by 'dump', but I can't get dump to do anything at all [11:26] Preparing to pull wdltool [11:26] Pulling wdltool [11:26] no handler to manage source [11:27] and the examples are not help either https://github.com/ubuntu/snappy-playpen/search?utf8=%E2%9C%93&q=dump [11:27] trijntje: and "snapcraft help dump" doesn't help? [11:28] nope. I can't even find where the it supposedly pulled the .jar to [11:29] trijntje: the good news is, we're needing people new to snapcraft to find these issues [11:30] trijntje: the bad news is it's a little bit early for the people who could help you [11:30] its just frustrating that there are no simple examples that make clear how that plugin works [11:31] like this: https://github.com/ubuntu/snappy-playpen/blob/af7154cc9cac61d55b17c4aa9517c4d7c003a1a0/jtiledownloader/snapcraft.yaml From this, I guess dump automatically unzips stuf as well? === hikiko is now known as hikiko|ln [11:34] trijntje: i wouldn't know :-) [11:35] Chipaca: yeah, me neither ;) Do you happen to know the best time to get help here? [11:36] trijntje: usually I'd say starting about now (or in half an hour tops), but it's been a crazy week so people might be taking it slower today [11:43] hrm, so i have an image building now ... but that seems to hav cloud-init enabled and not configured [11:44] (which means the boot takes about 5min with spilling cloud-init errors to the console) [11:45] sigh [11:45] augh [11:45] Chipaca: thats good to know, thanks. I'll hang around for a bit ;) [11:46] i really dontz know what mvo did here :/ [11:47] * live-build/ubuntu-core/hooks/22-disable-cloud-init.chroot: [11:47] - removed, the gadget.yaml now contains an "cloud.conf" file [11:47] that will create a /etc/cloud/cloud.cfg or a "cloud-init.disabled" [11:47] file (if coud.conf is missing) [11:47] oh man [11:48] would be helpful if he actually had published that gadget.yaml anywhere to the world [11:48] (and also ... this broke all dailies ... ) [11:52] Chipaca, do you by chance happen to know if he publisshed that gadget somewhere ... seems to have been sideloaded into the image he published [11:53] (there is nothing in the store nor in the gadget upstream tree) [11:53] ogra_: the pc gadget in the store should be like that [11:53] it definitely isnt [11:53] * Chipaca looks [11:53] ha not been changed since 4 weeks [11:53] *ha [11:53] *h a s [11:54] * ogra_ curses this kdb ... thats the last expensive one i ever bought ! [11:55] ogra_: sorry, i don't follow [11:55] Chipaca, the gadget in the store has not been changed in 4 weeks [11:55] ogra_: the test suite creates an image using the gadget from the store [11:55] (ignore my rasmbling about my broken kbd) [11:56] ogra_: the resulting image disables cloud-init because it doesn't have a cloud.cfg [11:56] Chipaca, well, that might be ... but the pc gadget in the tore doent have any magic for cloud init [11:56] ogra_: correct, the lack of magic means no cloud-init [11:56] no [11:56] that's what you just pasted from that changelog [11:56] th magic *disables* cloud-init [11:57] the thing you just pasted says "if no cloud.cfg, disable cloud-init" [11:57] the gadget in the store doent have *any* cluld handling [11:57] ince it is to old [11:57] which means cloud-init gets disabled as wanted [11:57] :-) [11:57] well, it doent [11:57] hah [11:58] doe that come from an ubuntu-mage ack ? [11:58] ARGH [11:58] ogra_: is this still with ubuntu-image master? [11:58] does that come from an ubuntu-image hack that waa not publihed ? [11:58] yeh [11:58] i built a local nap from mater now [11:58] oh man [12:00] Hello, is `$ snap-compose` on the roadmap? [12:00] * ogra_ wonders where these chars go when they are missing :P [12:01] or should we install other snaps from the snap we install? [12:01] (if that's even possible) [12:02] technically it is, via the snapd-control interface i think [12:02] but thats no auto-connect one [12:03] what is snap-compose? [12:04] docker-compose [12:04] as an analogy [12:04] what is docker-compose? [12:04] For a pre-configured setup based on snaps [12:05] oparoz: is the setup at the gadget level, ie you want a system with a bunch of snaps, or at the snap leve, ie you want a snap with a bunch of building blocks? [12:06] I'd like to make some snaps more flexible, so instead of shipping the database, per example, I'd like to be able to create a config file which comes with one option, but if you want another one, you modify the config file to load the other one and then re-launch the "bundle" [12:06] Chipaca, because tight coupling doesn't always make sense for server apps [12:07] ogra_: ubuntu-image fails with an error from mkfs.ext4 for me [12:07] ogra_: the one from master, that is [12:08] Chipaca, you need the foundations ppa [12:09] oparoz: i'm probably too obtuse right now. Are you assuming you can make snaps explicitly depend on each other? Or do you want to have, say, a blog engine that talks to either postgres or mysql? [12:09] sudo add-apt-repository ppa:canonical-foundations/ubuntu-image [12:09] Chipaca, the latter [12:09] and add "- lib" and e2fslibs to snapcraft.yaml [12:10] oparoz: and what's stopping you from doing that today? [12:10] i'm sure mvo has a good bunch of local patches for his snap [12:10] sadly not in his github branch [12:10] Chipaca, nothing is stopping me, but then I have to install every single snap and connect them [12:10] oparoz: (i'm still trying to understand the problem you're having, hence all these questions, sorry) [12:11] oparoz: so you want something to orchestrate your snaps? [12:11] Chipaca, exactly [12:11] ah! [12:11] :) [12:11] Sorry [12:12] oparoz: some people were looking into that, but there's nothing promised yet afaik [12:12] Chipaca, OK, that's what I wanted to know. Thank you [12:14] ogra@localhost:~$ ls /snap/pc/current/meta/ [12:14] gadget.yaml icon.png snap.yaml [12:15] so there is no cloud.conf in the released image in the gadget === hikiko|ln is now known as hikiko [12:16] it is definitely the 4 weeks old one from the store ... so there must be some ubuntu-image bit missing [12:18] ogra_: I'm sorry, i'm still missing something [12:18] error ? [12:18] ogra_: your problem is that the generated image runs cloud-init? [12:19] Chipaca, yes ... and there must be code somewhere that disables it for mvo's [12:19] ogra@localhost:~$ ls /etc/cloud/ [12:19] cloud-init.disabled [12:19] i.e. something that puts this file in place [12:19]