[00:10] sergiusens: if you're around tomorrow, I could really use your help with some cmake/pkg-config problems I'm having [00:26] in a snappy system, will apt still be used for core parts of the system such as kernel and drivers? [00:31] sgclark: d_ed: we've started sending out a weekly email higlighting snap work from the community, do you have a few bullet points about what you've been doing that I can include? === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [02:32] Shibe: no. === chihchun_afk is now known as chihchun [03:21] I give up. I'm trying to snapify a script that calls sox via play, and I can't get any progress past "play FAIL sox: Sorry, there is no default audio device configured". [03:22] The "pulseaudio" interface is defined and connected. I don't know whether or not I'm supposed to have PulseAudio installed in the snap (it doesn't seem to make a difference) or for that matter, why sox works immediately on a classic system and has no configuration inside a snap. [03:23] I have installed libsox-fmt-pulse. I come back to it every week and it's been five weeks and I'm stumped. [03:24] nhaines: does it fail even with --devmode ? [03:25] thomi: hey, I see in keybaise that you have an invitation available... :) [03:25] *keybase [03:25] elopio: I sure do - you want one? [03:26] elopio: it does! [03:26] thomi: yes please. [03:26] elopio: PM me your email address [03:27] nhaines: please report a bug. Sounds like sox needs configuration, which I have no idea how to do. [03:27] elopio: after 5 weeks of trying, neither do I! :) [03:28] But I hate to report a bug if it's really me just being dumb. :) [03:30] nhaines: that's what I do all day! [03:31] heh :) [03:32] What project do I file against in LP? === chihchun is now known as chihchun_afk [03:44] nhaines: snappy [03:47] Now to figure out how to protect my suuuuuper secret proprietary one-line sox command trade secrets! [03:55] Shibe: no. [03:55] okay, so how will drivers and kernels be installed/updated in snappy? [03:55] isn't each snap isolated from the rest of the system? [03:57] Shibe: drivers will be part of the gadget snaps and kernels will be part of the kernel snaps. [03:58] nhaines: is there any information about these "gadget/kernel snaps"? [03:59] https://developer.ubuntu.com/en/snappy/ under System Overview [04:00] Also probably https://developer.ubuntu.com/en/snappy/guides/porting/ [04:21] hello! [04:21] I'm trying to create a snap from a Python package. I've managed to create the snap, but when I try to run it, I have the following issue: [04:22] locale.setlocale(locale.LC_ALL, '') [04:22] File "/home/pierre/dev/rapid/parts/rapid/install/usr/lib/python3.5/locale.py", line 595, in setlocale [04:22] return _setlocale(category, locale) [04:22] locale.Error: unsupported locale setting [04:38] argh. Now I get another error when trying to re-build the snap. → "ImportError: No module named 'DistUtilsExtra'" although in my snapcraft.yaml, I have [04:39] build-packages: [04:39] - python3-distutils-extra [04:39] looks like snapcraft doesn't downloaded this package before trying to build... [04:39] (but it worked before and I don't know how) === chihchun_afk is now known as chihchun [05:30] https://bugs.launchpad.net/snappy/+bug/1593558 [05:30] Launchpad bug 1593558 in Snappy "sox not configured for pulseaudio when packaged in a snap" [Undecided,New] === JanC is now known as Guest67075 === JanC_ is now known as JanC [06:00] I've installed a snap as user1, but I can't use it as user2. When I want to install it for user2, I get a message that the snap is already installed. Is that a known limitation or should I file a bug? [06:01] hmm, it looks like /snap/bin is not in the PATH of the new user I created with adduser [06:07] trijntje: $ grep /snap /etc/profile.d/* [06:07] /etc/profile.d/apps-bin-path.sh:# Expand the $PATH to include /snap/bin which is what snappy applications [06:07] /etc/profile.d/apps-bin-path.sh:PATH=$PATH:/snap/bin [06:08] qengho: I've added /snap/bin to my PATH as a temporary fix [06:11] It happens both for users added with adduser and via the gui. Should I file a bug against snapd for this problem? [06:11] hey hey [06:24] https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1593578 [06:24] Launchpad bug 1593578 in snapd (Ubuntu) "new users don't have /snap/bin in PATH" [Undecided,New] [06:26] trijntje: What shell do new users have? Check /etc/passwd . [06:27] qengho: /bin/bash, same as the first user [06:28] trijntje: Is anything in their home clobbering PATH var? Also see /etc/skel/ . [06:32] qengho: no, only /etc/skel/.profile adds $HOME/bin to PATH. [06:32] echo $PATH [06:32] /usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games [06:37] dholbach: morning! [06:39] hi nhaines [06:41] trijntje: $ ls -l /etc/profile.d/apps-bin-path.sh [06:42] $ ls -l /etc/profile.d/apps-bin-path.sh [06:42] -rw-r--r-- 1 root root 101 mei 3 08:00 /etc/profile.d/apps-bin-path.sh [06:48] trijntje: That doesn't make sense. Something you said is wrong. At login, bash "runs" /etc/profile, which runs /etc/profile.d/*, which sets PATH to include /snap/bin . Put a "echo hi" in /etc/profile.d/apps-bin-path.sh to see if/that it's running. [06:50] qengho: you are right. If I do a real login, snap is in the path. I was just using 'su user' from a terminal, and if I do that its not in the path [07:01] dholbach: snapcraft frustrates me, as I've been unable to build absolutely anything. But I've destroyed whole development systems over and over in lxc containers while my laptop's 16.04 LTS install remains pristine, and *that's* been really pleasant. :) [07:11] qengho: what does this mean for the bug? Is it invalid/expected behaviour? [07:15] mhall119: * patched parts of KDE frameworks to support KIO in snappy [07:15] mhall119: thus fixing downloading/file browsing [07:17] trijntje: well, the bug-report title should be changed to mention "su". [07:20] nhaines, maybe if you post your snapcraft.yaml somewhere, somebody can help you? [07:20] nhaines, or you ask a more specific question? [07:24] qengho: I've added this to the bug report, thanks for your help [07:24] hey there [07:25] I'm trying to create a snap for this app: https://launchpad.net/rapid [07:25] I bzr branched the code (bzr branch lp:rapid), then created a snapcraft.yaml file, but I cannot build it [07:25] trijntje: In the mean time, $ su other - [07:26] it worked once (but then I couldn't run the program, which is another issue), and since then, whenever I try to rebuild, I get an error [07:26] That's all. sic [07:26] both snapcraft.yaml and error are here: http://paste.ubuntu.com/17426643/ [07:26] does anyone have a clue of what's going on? [07:27] ePierre: also use "--debug" as arg to snacraft. [07:29] qengho, ok. It provides a bit more info, but nothing interesting I think [07:31] dholbach: I did! It's in the bug report. :) https://bugs.launchpad.net/bugs/1593558 [07:31] Launchpad bug 1593558 in Snappy "sox not configured for pulseaudio when packaged in a snap" [Undecided,New] [07:32] cool [07:32] ePierre: "cd parts/rapid/install; pip3 install" # What is wrong with pip3 install there? How should you install? [07:32] ...if this were not a snap, what are install instructions? [07:33] dholbach: although my attempt at snappifying hollywood didn't go anywhere either, which is sort of a shame. :) [07:33] bring it up on the mailing list [07:34] Okay. I'm shy of mailing lists when the more probable cause is my own incompetence. :) [07:34] * nhaines tries to remember what the mailing list is caled this week. [07:34] snapcraft@snapcraft.lol [07:34] ;) [07:34] ogra_: :D [07:35] The list admin should really use the [snapcraft] prefix. [07:35] ogra_: you forgot the "lists." before snapcraft.lol! :P [07:35] ouch, indeed [07:35] nhaines, so that you cant read the subject in dekko you mean ? [07:35] yep, good idea :) [07:36] qengho, there is an install script: https://launchpad.net/rapid/pyqt/0.9.0a3/+download/install.py [07:40] ePierre: plugins are only smart about standard ways of installing software parts. You could make your own plugin that runs that. Extend python3 module, perhaps. [07:40] What is the proper way to create a snap for a program that can read files outside of the users home (eg /bigdata) [07:41] ogra_: do that I can find them in Thunderbird. Is there some strange Dekko bug? :) [07:41] good morning [07:42] nhaines, no, just a strange screen width of phones :P [07:43] prefixing subjects is a pretty bad habit since it pushes the actual info you want to the right ... if your screen isnt particulary wide the text you want to read might not be readable anymore [07:48] ogra_: that's what rotating the phone sideways is for, of course. ;) [07:48] sure ... [07:48] The actual info I want tends to be "why am I getting these emails?" [07:50] i get that by looking at the "To" address :) [07:51] ogra_: I don't have time for that. :) [07:52] then use a proper filter ;) [07:54] is it possible to have a snapped program read a file from outside the users home directory? 'my-app /path/to/data/file' [07:56] not at the moment [07:56] zyga: ping [07:57] well ... if you install with --devmode it might [07:57] morphis: hi [07:57] zyga: do you have time to take a last look at the modem-manager interface? [07:58] morphis: little but I can have a quick look now [07:58] ogra_: I'll give devmode a try, thanks [07:58] zyga: we would like to get it merged, jdstrand already gave his +1 [07:58] abeato: ^^ [07:58] morning, I finally succeeded in building a custom kernel (which is correctly configured to working with ubuntu-core) for raspi with IIO support \o/ [07:58] Yay! [07:59] I finally could run a command from a snap which tries to access iio-devices [07:59] and it fails ! Yay! [07:59] \o/ [07:59] I'm using lsiio which access sysfs [08:00] my goal is to create a bug on launchpad which will make the devs create an appropriate interface - unless there is already one?! Hence my report here before posting a bug. [08:02] (I will also try to write an article/report on how I did to create an IIO-kernel for raspi based on 4.7-rc3 - it wasn't torture but not far from it. Thanks to ogra_ for supporting me all the time [08:05] ogra_: putting 'devmode' in the snapcraft.yaml file did not work, but installing the snap with --devmode did. Thanks for the advice [08:08] seb128, good morning. I tried your suggestion last night http://paste.ubuntu.com/17426218 and that made dconf happy - thanks! The app started, but then apparmor blocks it as it's trying to write to /var/cache/fontconfig/ [08:08] qengho, thanks! I'll dig a bit more and perhaps discuss with the owner of the software about it [08:10] dpm, great for dconf! unsure about the fontconfig issue though [08:10] no worries, I'll keep investigating [08:12] dpm, first time I see that issue, you use the standard wrapper? [08:13] seb128, parts of it. It's the qt5-launch wrapper, to which I had to add some bits from the gtk wrapper for a qt app that uses gtk theming [08:13] dpm, is XDG_CACHE_HOME set? [08:14] seb128, yeah, that's one bit that I took from the gtk wrapper: https://github.com/dplanella/qt5conf/blob/with-gtk/qt5-launch#L53 [08:16] dpm, sorry I don't know offhand [08:16] np, thanks anyway [08:17] trijntje, yeah, the entry in snapcraft.yaml just makes sure that a snap from the store blocks when you install it without --devmode ... [08:21] ogra_: I see. I was still able to install it though, I just got the same 'permission denied' error when I tried to read the file [08:25] trijntje, yeah, but you installed it locally ... the a snap install from the store should behave differently [08:26] we dont actually stop you from shooting both your feet when you sideload stuff ;) === chihchun is now known as chihchun_afk [08:27] ogra_: that makes sense. I'll mostly be creating snaps for myself and some colleagues, so foot-shooting is no big deal ;) [08:27] just wear these heavy duty shoes then ;) === verterok` is now known as verterok === chihchun_afk is now known as chihchun [08:47] dholbach: dpm: did you get issues with snapcraft cleanbuild? I'm trying from snapcraft-example godd, run "snapcraft cleanbuild" [08:47] I'm getting a lot of E: Failed to fetch http://archive.ubuntu.com/ubuntu/pool/main/f/file/python3-magic_5.25-2ubuntu1_all.deb 500 Internal Server Error [08:47] in my new lxd container [08:47] it doesn't refresh? [08:48] it worked for other projects earlier [08:48] didrocks, it worked for me yesterday, I've not tested it today. I think someone mentioned this on gitter yesterday, and `lxd init` helped them [08:48] let me try godd [08:48] thanks :) [08:48] keep me posted [08:49] it wfm with godd - looks like snapcraft 2.11 can get all build-deps - trying with trunk [08:50] hum... it install snapcraft 2.11 in the lxc container [08:51] so yeah... it works [08:52] hello, I tried the "pulseaudio" interface on my snap, and it didnot seem to work. I have blocked message from app armor [08:52] I'm going to try lxd init [08:52] already ran :/ [08:52] hum, unsure what's happening [08:58] didrocks: there's an open bug by Alan Pope reporting the 500 [08:58] * tsimonq2 finds it [08:59] oh great! [08:59] popey: you have the same issue as well, did you get it fixed? [08:59] didrocks: I am not here, but I did fix it [09:00] nuke lxd/lxc config from orbit and start again [09:00] it was lxc/lxd networking at fault, not snappy [09:00] * popey disappears in a puff of aether [09:00] popey: always the big restart button! Many things, and good week-end! :) [09:00] * tsimonq2 chases after popey but to no avail, he is gone (:P) [09:01] popey: have a good weekend :) === chihchun is now known as chihchun_afk === chihchun_afk is now known as chihchun [09:31] When I try to build a snap from a Python3 project, what happens in snapcraft exactly? Cause in the project I'm trying to build, it requires a few packages to be installed (e.g. python3-gi), and they are installed on my system, and they are also in build-packages: section, but somehow the setup.py script doesn't seem to have access to them when I launch snapcraft [09:37] ePierre: you want them in stage-packages [09:37] ePierre: they are required at runtime, not just at build-time [09:37] ePierre: gi is gobject introspection [09:37] ePierre: that requires the actual shared objects (random .so files) [09:37] ePierre: the .gir files [09:37] zyga, indeed... it seems to build properly now [09:37] ePierre: and then python[23]-gi itself === chihchun is now known as chihchun_afk === qengho_ is now known as qengho [10:34] dpm: the access to /var/cache/fontconfig is non-fatal and almost certainly not why your app didn't work (consider the fact that it wants to write to a directory it doesn't have access to anyway-- that dir is root owned and your app runs as you) [10:34] it is a weird fontconfig thing [10:37] jdstrand, I figured out the /var/cache/fontconfig issue, as you say, it was a red herring. It turns out that the app ships a fonts.conf file that specifies that directory as the first to use in a list. I tried to alter that list with a local.conf file, which didn't work, and then figured out that the app was not starting because of a missing 'network-bind' plug, which snappy-debug hadn't flagged in previous tests [10:37] after it did, I added the plug and it then worked [10:41] Hi - I am trying to package a java server application with snappy for the first time. My jenkins server builds the things and I have a url for that (hardcoded for now). [10:41] What I would like to do is to download it (with auth) [10:41] is there a wget plugin? [10:41] or would that be the pull plugin [10:41] jdstrand, by the way, am I using snappy-debug correctly? It does not seem to work as expected -> http://paste.ubuntu.com/17428561 === chihchun_afk is now known as chihchun [10:43] dpm: you said 'does not start the app', but it does-- it is tailing the log and will hang there until it has something to report [10:43] dpm: you can run it as non-root, but if you do you run the risk of hitting kernel rate limiting for log output [10:44] dpm: so, in earlier tests it might have been rate-limited or stopped before it hit the denial [10:44] (the app stopped before it hit the denial) [10:44] ack [10:45] jdstrand, so regarding the usage of snappy-debug, then it's better to launch the app in another terminal to at least see the UI? [10:45] dpm: there is no ui. think of it as a fancy 'tail -f /var/log/syslog' [10:45] well, there is no gui [10:46] there is cli output [10:47] jdstrand, right, I just thought that that would launch the app's UI rather than holding it [10:47] Is there any ETA for Firefox and LibreOffice snaps appearing? [10:48] Mikaela: AFAIK there's already a LibreOffice snap [10:49] Mikaela: http://news.softpedia.com/news/libreoffice-5-2-beta-2-now-available-as-a-snap-for-ubuntu-linux-other-distros-505263.shtml [10:49] that's from two days ago [10:50] nice [10:51] Is there a place where the plugins are documented? [10:53] e.g. the curl plugin has no docs doing snapcraft help curl [10:54] dpm, dholbach: ^ [10:55] Maybe i should take a step back and find out if I am using this for the right purpose? I have the server app that is already built. I want to create a snap package that can be run anywhere. For this all i really need is to download the zip, unzip it in a location and setup the properties. After that it only needs to run [11:19] zyga: you had time for the modem-manager interface? [11:22] morphis, not zyga, you know he is slacking all day :) [11:22] ogra_: I know, but it looks like nobody else can then approve my PR :-) [11:35] pandaadb, https://developer.ubuntu.com/en/snappy/build-apps/plugins/ [11:35] there's no `curl` plugin, it's a part: https://wiki.ubuntu.com/Snappy/Parts [11:36] hey - thanks [11:36] anytime :) [11:36] so downloading a zip somewhere is not what I am meant to do? I am starting to think I am using snappy for the wrong purpose [11:38] Oh okay - so using a "curl" part simply downloads curl into my snappy package so that curl is available [11:38] morphis: I'm still reading it [11:38] morphis: I'll land it shortly [11:38] morphis: it's just very long :) [11:39] zyga: awesome! :-) [11:39] morphis: I think you will be happy to hear about that you may soon get auto-connecting network-manager [11:40] morphis: internally plugs and slots will auto-connect within one snap [11:40] zyga: you make my dreams become true :-) [11:41] pandaadb, yep and downloading a zip sounds reasonable (https://github.com/ubuntu-core/snapcraft/blob/master/demos/tomcat-maven-webapp/snapcraft.yaml does it too) [11:43] ah tar-content can download anything [11:43] zyga yo [11:43] snapcraft help tar-content does not recognize the plugin though for help (add authentication and such) [11:45] zyga, is SELinux support for snaps coming soon? [11:45] I'd rather not have to throw SELinux into permissive mode for this [11:47] Son_Goku: hi [11:47] Son_Goku: I'm not working on it yet but I can help someone that is interested and experienced in selinux [11:48] Son_Goku: the model supports it from day one [11:48] I work in both the Fedora and Mageia communities, so I'm happy to help if you need anything [11:48] Son_Goku: that's great! [11:49] and protip, Fedora COPR will soon offer Mageia as a release target [11:49] Mageia 6 anyway [11:49] since we're gradually moving onto the same tooling as Fedora for package management [11:49] Son_Goku: I think the first thing that should be doable is a way to run snapd in a mode where some interfaces are available and some are not [11:49] Son_Goku: and when an interface is not yet implemented for selinux, snaps using those have to be installed in devmode [11:49] Son_Goku: on fedora we could enable seccomp pretty easily I think [11:49] yes [11:49] that should cover most of it [11:50] Son_Goku: the first step towards that would be to make snap-confine allow seccomp while not yet using apparmor [11:50] Son_Goku: that should be pretty simple, it's roughly one or two lines and an optional compile on one file (have a look, it should be very fast to implement) [11:50] Son_Goku: for selinux itself I'd like to have something simple we could implement to prototype the idea [11:50] Son_Goku: I think the network interface (the thing that lets snaps talk to the network) is the first thing we could try [11:51] unfortunately can't touch things requiring a CLA [11:51] it gets too messy for me to contribute to stuff like that [11:51] :( [11:51] best I can do is help you along [11:51] do you not already talk to NetworkManager? [11:52] all variants of Fedora use NetworkManager as the default management interface [11:52] well, except Cloud, which uses systemd-networkd [11:53] Son_Goku: as it is very well defined, a lot of the snaps need just that and we could try that [11:53] Son_Goku: why not? is your employer against that? [11:53] Son_Goku: it's not about network manager [11:53] Son_Goku: there are two interfaces: network, which lets you create AF_INET sockets and a few other things [11:53] Son_Goku: and network-manager, which lets you talk to n-m over dbus [11:53] Son_Goku: those are totally separate [11:53] Son_Goku: if you cannot work with the CLA in any way then I guess you'll have to wait until someone else steps up and can sign the CLA or that your situation changes [11:53] it's not that I can't, but because the CLA involves copyright assignment, it has to be reviewed [11:53] certain types of CLAs are easier to deal with than others [11:54] and lawyers scare me :) [11:55] in the meantime, it seems like it'd be a good idea to learn Go, though not really a fan of the language [11:56] Son_Goku: again, network and network-manager are totally different [11:56] Son_Goku: network lets you run a program like wget [11:56] Son_Goku: it lets wget use a few syscalls [11:56] Son_Goku: and perhaps read a few typical files (/etc/resolv.conf) [11:56] Son_Goku: are you familiar with the particular CLA used here? are you sure it's impossble to sign? [11:56] (for you) [11:57] not sure actually [11:57] but I was familiar with the ones used in older projects I tried to contribute to [11:57] upstart, etc. [12:01] why isn't snapcraft part of the snap project? [12:03] Son_Goku: the same codebase or the same group on github? [12:04] snapcraft is in ubuntu-core rather than snap-core [12:05] does snapcraft only work with ubuntu? [12:05] Son_Goku: it is a combination of reorgs to the code [12:05] currently, yes [12:05] Son_Goku: currently snapcraft does currently only work on ubuntu but this is being changed [12:05] Son_Goku: it should be able to build snaps on any OS [12:05] its just python ... might not be hard to port to other distros [12:05] Son_Goku: probably as a snap itself [12:05] O.o [12:06] hey - i am trying to pull a jdk. I read this: If you only need to embed a Java runtime, add a part with the jdk type. [12:06] how does a part with a type look like? [12:06] if it's a snap, that'd make it hard to consume things like rpms and debs, wouldn't it? [12:06] why [12:06] It keeps telling me to use a plugin but i would like to only add the jdk [12:06] Son_Goku: not really, it would be easy to run it anywhere snapd runs [12:06] as long as it knows where to get them and how to unpack them [12:06] pandaadb: I think the word "type" is really "plugin" today [12:07] oh okay. So that one tells me that source is mandatory [12:07] oh geez [12:08] it needs arch translations [12:08] Son_Goku: Snapcraft is one of many possible tools to make snap packages. For a while, snap packages were made with "cp" and "vi". [12:08] ew [12:08] vi!? [12:08] Yes. UCB. not vim. [12:08] Son_Goku: snaps are just squashfs filesystems with meta/snap.yaml file [12:08] Son_Goku: they can be made in any way you want [12:09] that's more gross than vim [12:10] Okay. [12:11] but wow [12:11] so it's a loop with stuff inside plus a declaration file [12:11] Son_Goku: Anyway, you asked why. Snapcraft came after snap, as an idea to help snaps, not as The Only Ordained Way To Make Snaps Forevermore. [12:11] yep [12:12] I don't see debs as a plugin in snapcraft [12:12] so how does it consume debian packages/repos? [12:12] Son_Goku: Anypart can include debs. [12:12] it pullse the binaries from the ubuntu archive [12:12] *pulls [12:12] and unpacks them so you get your depending libs and all [12:12] Son_Goku: "stage-packages: [...]" [12:13] oh, so it's integral to it rather than being a component of it [12:13] right [12:13] it could use rpms [12:13] or fooples ... or shrobs [12:13] :) [12:13] whatever helps you to fulfill your requirements to run your snap [12:14] (even a tarball full of binary blobs you pre-built would work) [12:16] ogra_: package names will not match, so it will matter what kind of system you're building on as to whether a snapcraft.yaml "stage-packages" will work. I think it's safer+better to say "this always gets debian packages". [12:16] sure ... but nothing stops you to do non-standard hackery [12:19] hm. when using the jdk plugin, i need to define a source (i think it is ignored). But when doing snapcraft snap, it doesn't build my snap anymore :/ [12:21] pandaadb, did you run snapcraft clean first ? [12:21] no, let me try that [12:22] it seems to be updating my repositories and then in the end tells me my security is a "danglink link" i think [12:23] ah, you need to make sure to not have links that point to somewhere outside the snap [12:24] ah that's maybe because I had to add security certificates to my local java [12:24] and they are not really needed for the snap i am building, but rather so i can talk to our shitty maven repos [12:24] * pandaadb tries to remove link === hikiko is now known as hikiko|ln [12:24] Or is there maybe a way to have the jdk part be completely independent? [12:26] pandaadb: No. If you need it, you must pick it and ship it in the snap. [12:26] oh by that i meant that it should just download everything and not try to link to something local [12:26] I think maybe the issue is that I am using the oracle jdk for development, but the snapcraft is downloading the openjdk of a different version [12:27] the simlink doesn't even exist (the one it is looking at) [12:27] i guess that's what dangling means [12:29] seems to be a known issue: https://bugs.launchpad.net/snapcraft/+bug/1500505 [12:29] Launchpad bug 1500505 in Snapcraft "snapcraft produces snaps with dangling external symlinks" [High,Fix released] [12:31] ogra_ actually, I'd suggest that it should be able to handle either debian or rpm packages [12:32] Son_Goku, well, as qengho said above, then your snapcraft.yaml cant be universall [12:32] since the names for stage packages will differ [12:32] the output is universal, though [12:33] well, the idea is also that you can just grab an upstream snapcraft.yaml and it will just build [12:34] so declaring a "from " is hard? [12:34] so while other package managers will be possible to hack in, i guess snapcraft as a snap (which would then just use debs) will likely become a default [12:34] :/ [12:34] butu why would you care at all [12:35] you do snap install snapcraft ... feed it your snapcraft.yaml and out comes a snap [12:35] what you care for is your source and how to get it to users and running, no ? [12:35] Okay - so my snap did not package, because jdk needs a source. But since I didn't get why, i added the string "some source". This is obviously not a location, which is why my snap broke. I set it to "." and now it packages [12:36] :) [12:37] pandaadb: Are you getting source from somewhere and compiling it, or wrapping up some existing debianubuntu package in a snap? [12:37] Son_Goku, the execution environment should nnot matter at all in the end [12:38] qengho, i am trying to package up my java app (build with gradle) and make it run [12:38] the next step will be to write a gradle plugin as I am not using maven [12:38] and then i will attempt to build it from source [12:38] I care about the source because it can matter [12:38] Ah, then you can put it next to snapcraft.yaml and say "source: pandaadbdir" [12:39] Is te root directory $SNAP or is it @SNAP_DATA ? [12:39] oh okay - that is what the source is for [12:41] found it, it's SNAP [12:41] pandaadb: SNAP is snap/. SNAP_DATA is a /var/ something. SNAP_USER_DATA is a /home/~user/ something. [12:41] I need to provide my own .pro file to build this thing (the one that comes with it is "broken" - I will look at fixing it later). I had added it in to my git repo in parts//src/ - but then when I tested it from scratch, of course git wont clone in to a non-empty dir. So is a suitable hack to have another step in my yaml to copy the file over after the checkout has happened? [12:41] perfect :) thanks a lot [12:41] pandaadb: install snap "hello-world" and run hello-world.sh . [12:41] s/checkout/clone [12:42] thanks! [12:42] Son_Goku, so use source for everything (you can do that) and no binary packages at all [12:43] this is awesome. Remind me of docker. But with docker if feel so restricted to the container I built. If I made a mistake or i want to change something inside my container, i am kind of screwed [12:43] but that pulls in ubuntu sources, right? [12:43] I can't use srpms or something else [12:43] it is just quite some effort and will take extra build time ... but technically nothing stops you from building *all* the bits from source [12:43] it pulls the sources you point to [12:43] (it will use ubuntus GCC ... but thats it) [12:44] i'm talking about upstream tarballs (you can just point to an url) or git trees [12:45] nothing requires that you actually use stage-packages at all [12:47] can snappy packages be run with sudo? i think my server can not start because it does not have permissions [12:49] if it is a server it already runs as root [12:49] did you define a daemon line in your snapcraft.yaml ? [12:49] oh okay, that may be the certificate dangling stuff then [12:49] https://github.com/ubuntu/snappy-playpen/blob/master/tinyproxy/snapcraft.yaml [12:49] ogra_, no - gradle creates a wrapper that uses java_home and just runs it with the classpath env [12:49] see how i did it for tinyproxy there [12:50] ( you need "daemon: simple" in the "apps:" part) [12:51] what does the plugs do? [12:51] willcooke: Hmm, you could abstract it a bit. Make a ./Makefile next to snapcraft.yaml , and massage it together in your build-step? pull, pull, cp, and build in the makefile. So, plugin:make and source:. ? [12:51] d_ed: are your KDE patches in upstream, or are they only in an Ubuntu or Neon archive? [12:51] willcooke: that^ will not work in out build servers, though. [12:52] only pull stage has access to network. [12:52] qengho, ah, good point. [12:52] Can the copy plugin help me I wonder. [12:52] * willcooke reads [12:52] kyrofa, sucessfully built using your new qmake plugin, thank you!! [12:53] willcooke, excellent, thank you! [12:53] willcooke, it'll probably go through some more renditions, but now I know those variables are the ones you need [12:53] willcooke, so you should be good once it lands [12:53] cool! I will comment on the PR [12:54] I'd appreciate it [12:54]