[04:32] PR snapcraft#877 opened: [WIP] Add new "source-commit" field for VCS sources === vigo is now known as vigo|coffee === chihchun_afk is now known as chihchun [06:09] good morning === hikiko is now known as hikiko|off [07:59] o/ [08:33] Can I install Ubuntu Core on the Dragonboard eMMC ? [09:25] Bug #1637445 opened: The serial vault fails to connect to serial-vault.canonical.com [09:55] pitti, hmm, i cant remember the bugnumber but i am sure there was a bug that seeding libnss-resolve fixed [10:03] ogra_: it certainly came up as part of debugging something, but it's really by far not ready in xenial [10:04] sadly there is no info why exactly it was added but it looks like it was related to cloud-init ... http://paste.ubuntu.com/23392138/ [10:05] i. e. someone thought "oh, yakkety uses resolved with this, so let's try this", but this was probably bug 1620559 or so [10:05] Bug #1620559: /etc/resolv.conf is empty on snappy

[10:05] mvo added it to livecd-rootfs .... i later moved it to the metapackage during a cleanup [10:05] yeah, I think this was just chinese whispers/a misunderstanding [10:08] Are there any "real world" tests for snapd? By which I mean, not just formal verification but "try x, see if it worked" [10:09] My practical example would be a tiny Go test snap using the lxd-client interface I added [10:10] Which is what I've got, but manually [10:11] balloons: Incidentally, as you were interested in it, https://github.com/snapcore/snapd/pull/2225 is in a state that you can try out if you'd like [10:11] PR snapd#2225: Implement lxd-client interface exposing the lxd snap [10:12] hi kalikiana_, take a look at snapd's spread test suite, there are lots of real world usage examples https://github.com/snapcore/snapd/tree/master/tests/main [10:12] (I removed my attempted access to the executables, it's socket-only now) [10:14] fgimenez: Oh, thanks, it seems I didn't look properly, I saw the _test.go files all over the places but didn't reaize there was another suite entirely [10:16] kalikiana_, yep, both suites, unit and integration, are run with each PR, let me know if you need any help writing the spread test for the case you mentioned [10:50] I'm looking for some input on dbus confinement. KDE has lots of apps that (semi-implicitly; as in: devs might not even know dbus is involved) access the session bus and claim a service name on it. This is how we implement unique application behavior (e.g. user starts foo, foo claims org.kde.foo, when foo is invoked again by whatever means it will first check if org.kde.foo is claimed and if so send a raiseWindow to that instead of starting a second [10:50] instance of foo). I am wondering if it would be possible to have a common interface which allows a snap to connect to the session bus and interact with a specific service name there (i.e. its own). === Guest91977 is now known as ahasenack === ahasenack is now known as Guest60864 [11:28] didrocks: yo man. I wonder if the main project name would be available in the snapcruft plugins? The other question I have in mind is of it is possible to move out the tar.gz created during the build to the same place as the final .snap is put. [11:29] bzoltan: If your build tool is defined to "install" [11:29] I'm trying to build a snap of a Python app, I'm having trouble with one of it's libraries which I have as a part: but fails to "snapcraft build" because of what seems to be peculiarities in the setup.py. Is there a one-liner I can use to reproduce what the Python plugin is doing? I want to add debugging and logging to the src and retry the build [11:30] all that, then it could easily. [11:30] sparkiegeek: Does the "--debug" param on "snapcraft" give you enough? [11:31] qengho: how to define it to "install"? [11:32] qengho: not quite - it shows me it's running ['/bin/sh', '/tmp/tmpcuv22g6y', 'pip', 'wheel', ...] but I don't get to see inside that tmp file, nor see the environment it's running in (PWD, os.environ etc.) [11:32] bzoltan: hey! there is no restriction, so move it to .. ;) Unsure about the main project name, let me check the code [11:33] qengho: I've been cleanbuild --debug'ing and get dropped into the LXC - great - but struggling to iterate - if I use system python it Just Works™ but am failing to use the python in parts/install [11:33] bzoltan: What is it you want to move, exactly? [11:34] bzoltan: so, you can do: [11:34] config = snapcraft.internal.load_config(project_options) [11:34] then config.data['name'] has the snap name [11:35] qengho: I have a plugin what wraps up a part of the stage for further use. I made a simple plugin to ride the build ... it creates the tar.gz to the parts [11:35] project_options is available in your plugin [11:35] didrocks: \o/ [11:37] bzoltan: this is just from reading snapcraft source code, might be wrong ;) [11:38] didrocks: it is config = snapcraft.internal.load_config(self.project) [11:38] didrocks: perfect :) [11:38] bzoltan: sweet! :) [11:39] sparkiegeek: Just to peek, I suggest "sudo -e /usr/lib/python3/dist-packages/snapcraft/internal/common.py" and adding subprocess.check_call(['/bin/cat', f.name]) after the two f.flush()es. [11:40] qengho: hah! I'm doing something similar right now :) [11:40] qengho: thanks === chihchun is now known as chihchun_afk [12:00] pitti, hrm hrm [12:00] http://paste.ubuntu.com/23392478/ [12:05] hi [12:07] i try to make a snapcraft package but in my project I need to provide pypandoc (a python wrapper of pandoc) but the python3 installation try to use a bdist_wheel that is not supported by pypandoc. Is there a way to tell to snapcraft to not for the usage of wheel with the python3 plugin ? [12:08] foxmask: does it fail, or just look ugly when it tries something not supported? [12:09] qengho: its ending like this [12:09] Failed to build pypandoc [12:09] Oh man. [12:09] even if i can use it with import python [12:10] "ERROR: Failed to build one or more wheels === Guest60864 is now known as ahasenack === ahasenack is now known as Guest42345 [12:37] ogra_: is there a way to install Core on snapdragon' internal storage ? [12:38] or boot from sdcard is the recommended and the only way ? [12:38] the latter [12:38] zyga: ping [12:39] theoretically it would be easy (you can do it on the beaglebone for example) but for dragonboard you would have to re-build the first stage bootloader to initialize the MMC instead of the SD [12:40] ogra_, I'm writing some test cases for console-conf but there are somethings not clear yet, hope you can help me :) [12:40] hmm, I was curios to check if the DB internal storage performs faster than my sdcard. I'll wait, maybe in future we'll have some kind of wizard to achieve that. [12:40] Is my Pandaboard useless? [12:44] ogra_, for example when we setup the network config, are all the changes suposed to be saved even if I reboot? [12:45] can anyone answer where snap(1) man page is? [12:46] because it's referenced in the systemd units [12:57] ppisati, ooooh !!! look what i just hit on the pi3 http://paste.ubuntu.com/23392621/ [12:57] vigo, i think when you did hit the "done" button they should be saved in a file in /etc/netplan/ [12:59] ppisati, so it smells like you are right about a powering issue ... the above actually happens when console-conf (or netplan) tries to bring the wlan0 device down and up again [13:00] Sorry if I'm "reposting", but I think I got disconnected: I'm the maintainer of an open source SDK for distributed systems (Safir SDK Core), and I've been thinking about whether Snap is suitable for packaging it. The product consists of a bunch of executables and programming interfaces in C++ (libs and headers), Java (jars) and C# (assemblies). A developer is meant to download the SDK and start using my interfaces in his/her programs. I've been readin [13:02] hello [13:02] how can I get git repo of ubuntu core [13:02] ogra_, great thanks, so that means that once you have set the network config it should appear already configured after reboot right? [13:03] DonOregano: A snapped program can't write outside of some very specific places, so perhaps it's not a good fit for a tool like a compiler, but it might be nice to make the software that your SDK makes to be easily snap-able. [13:04] Ok, that was kind of what I was suspecting. [13:04] vigo, theoretically even without reboot, but that seems broken atm [13:04] ogra_, excellent now I can write it :) thank you [13:04] ogra_: how did you get that? [13:05] DonOregano: How can we help make your software-maker generate snaps? [13:05] ppisati, i ran "sudo console-conf" on an already configured snappy pi3 ... to reconfigure for wlan [13:06] ogra_, this is one of the bugs related? https://bugs.launchpad.net/ubuntu/+source/subiquity/+bug/1636419 [13:06] (i.e. i can only finish the install using eth0 currently ... then i reboot and do the above to turn off eth0 and turn on wlan0) [13:06] Bug #1636419: Network settings aren't set in Dragonboard [13:06] qengho: Well, that is not so much the issue. Whatever the user/developer that uses my SDK does with packaging is not likely to be SNAP-friendly, since it is mainly defense industry management systems, so easy deployment to "people on the internet" is not really their cup of tea.... [13:06] vigo, no, that should work (it surely does here ) [13:07] qengho: I was just hoping to find an easier way of getting my SDK to my users... Especially to be able to get new users :-) [13:08] vigo, not sure we have a bu for that yet .. console-conf calls "netplan apply" but that doesnt seem to actually restart the network interfaces in the new configuration without reboot ... perhaps pitti can tell you if thats actually supposed to happen [13:08] *bu [13:08] bah [13:08] *bug [13:10] ogra_: netplan apply simulates an unplug/replug for interfaces that are down; networkd doesn't touch interfaces which are already up [13:11] DonOregano: Three features DI folks would like are 1) roll-backs to known good versions, 2) internalized dependencies so updates of A don't affect B, and 3) compartmentalized execution, so an adversary breaking one pieces doesn't mean he can break more. [13:11] pitti, well, if i enable wlan in console-conf (logging in via ssh through eth0) and call ifconfig afterwards, the wlan0 device isnt up [13:11] pitti, eth0 is gone and wlan0 is up after a reboot though [13:11] apply is supposed to bring it up then, so if that doesn't work, it's a bug [13:11] DonOregano: That updating part is easily turned off. [13:12] ogra_: (works here, though) [13:12] i. e. needs precise logs and config files [13:12] pitti, ok ... not sure if its consoleconf, netplan or the driver though [13:12] ogra_: someone added a blacklisting to one driver, though, bug 1630285 [13:12] Bug #1630285: mwifiex_pcie crashes after several bind/unbind

[13:13] genqho: Yeah, I agree! And I can see that Snaps would be useful for a "finished product" based on my SDK. I may look in to that as a suggestion for one of my users. But that wasn't really what I was after today. [13:13] ogra_: so if it's that driver, that would explain it [13:13] pitti, yeah, no PCIe on my boards here :) [13:13] well, at least half of it; if it's down and you have a config then it should still be brought up [13:13] ok, that's the only one I'm aware of [13:13] pitti, i see it in a Pi3 ... [13:13] which is bradcom wlan [13:14] genqho: Today I was after getting out of maintaining separate packaging for Ubuntu, Debian, RHEL/Centos and Arch, etc. [13:14] ogra@pi3:~$ lsmod [13:14] Module Size Used by [13:14] brcmfmac 282624 0 [13:14] brcmutil 20480 1 brcmfmac [13:14] cfg80211 548864 1 brcmfmac [13:14] pitti, but that could indeed also die due to bind/unbind loops [13:15] DonOregano: Cool. We'll be here when you're ready. IF the sdk always builds in the user's $HOME, it could fit, but I can imagine build bots in /data and such not working. [13:16] qengho: But if snaps are not suited for exposing APIs in C++/Java/C# to developers then I will stay where I am. [13:17] qengho: Thanks. I'll keep checking out where things are headed with Snaps. I do like the idea... [13:17] DonOregano: Snaps aren't that specifically restricted. It could do all that, but it would have an opinion about where it is allowed to write, which is a feature, and possibly hated by some fraction of your users. [13:18] DonOregano: So, not now. Yes, try again later. [13:18] qengho: Hehe. Yeah, I would really like to avoid aggravating my users with "newfangled stuff" :-) [13:20] ogra_: can you open me a bug for that? my rpi3 is still dead, but i can start looking at it (and then we keep track of it) [13:23] ppisati, bug 1637505 [13:23] Bug #1637505: power error under ubuntu core when running console-conf

[13:24] Bug #1637505 opened: power error under ubuntu core when running console-conf

[13:43] zyga: https://github.com/zyga/snapcore-fedora/pull/10 [13:43] PR zyga/snapcore-fedora#10: Refresh patches for snapd spec [13:58] hi, there, how is one supposed to debug a snappy program? [13:59] http://paste.ubuntu.com/23392842/ is what i get at the moment === Ciblia[m] is now known as Riotela [14:04] http://askubuntu.com/questions/783979/how-do-i-debug-snaps [14:05] not sure that helps me [14:05] let's try [14:24] tsdgeos_: afaik, debugging something in a snap is mostly the same as debugging something not in a snap, it's just that the logs and locations of various things are all different; and you can't debug symbols from .debs necessarily [14:25] dobey: well, i can't gdb it [14:25] so debuggin it is not "as you would do it without a snap" [14:25] unless people don't use a debugger to debug stuff nowadays :D [14:26] tsdgeos_: why can't you gdb it? [14:26] you can, but you need to ship gdb inside your snap [14:26] dobey: http://paste.ubuntu.com/23392842/ [14:26] ogra_: lol [14:26] very useful :D [14:26] then you can use "snap run --shell $app_in_the_snap" [14:26] ogra_: or use the classic snap or be running on a classic system? [14:27] that gives you a shell you can run gdb in [14:27] dobey, the classic snap only works on all-snap systems, but yeah, thats possible too [14:27] apt install gdb and attach to the pid [14:27] (dont try to install the classic snap on a classic system ... wont work) [14:28] ogra_: right, i'm saying that's an option if you're doing that. if you're on a classic system, gdb from the host should be usable. the main issue is getting the env set up and running the right thing [14:28] i dont think gdb from the host will work, apparmor and seccomp will get in your way [14:28] i mean, running gdb from the host, not from within a shell in the confined snap [14:29] yes, me too :) [14:29] ie SNAP=/snap/foo/8 gdb /snap/foo/8/bin/baz [14:29] anyway, since you need debug symbols anyway, i'd just build a debug snap with gdb included and use snap run [14:30] is there a way to put the /init that resided in the initrd into a debug/verbose mode? i built a kernel using snapcraft, and its behaving differently than the vanilla binary does... it boots to systemd and I can get a debug shell, but its like the system doesn't fully come up, and if I boot with init=/bin/bash then the environment is pretty different upon boot, my custom kernel has a bunch of /tmpmnt_.... left on the root, while the vanilla version [14:30] looks like it swapped root correctly.... [14:31] Croepha, put "break=bottom (or =top or =premount) on your kernel cmdline ... that gives yoou a shell iside the initrd [14:31] if you ctrl-d it moves on with the boot [14:33] ogra_: ok cool thanks, i'll give that a try [14:34] the only thing I can think of, unless there is an error, is just to extract both the initrds and diff them [14:35] they should hopefully be 100% identical (apart from the modules shipped) === chihchun_afk is now known as chihchun [14:42] PR snapd#2228 opened: Allow fwupdmgr refresh [14:43] wow, looks like break=top did the trick, it stopped on an error [14:44] zyga: ping [14:51] maybe it doesn't have anything to do with the contents of the initrd... [14:52] im getting a "No such file or directory" fir /tmpmnt_writable/system-data/var/lib/snapd/snaps/core_324.snap [14:53] my vanilla version has a "ubuntu-core_758.snap" [14:53] that sounds like your ubuntu-imae is very outdated [14:53] ubuntu-core is dead, new images should all be using "core" [14:53] ahh ok [14:54] (we still build ubuntu-core but only for classic installs until there is a proper migration path) [14:54] so snap refresh your ubuntu-image snap ;) [14:55] * ogra_ goes afk for a while [14:55] pitti, what I asked ogra_is if once the network configuration is done, it's saved in a way that reboot will keep changes? [14:55] ogra_ Thanks! [14:56] has anyone seen zyga today? [14:56] pitti, will it save the access point name so there is no need to enter it again after reboot [15:03] vigo: no, it should be saved somewhere in /etc/netplan/*.yaml [15:06] ogra_: do we have separate image for RPi3? [15:06] bzoltan, yes, the one with pi3 in the name [15:07] ogra_: where does it live? Not listed here - http://cdimage.ubuntu.com/ubuntu/releases/16.04/release/ [15:07] https://lists.ubuntu.com/archives/snapcraft/2016-October/001463.html [15:07] pitti, yes, that's what ogra_ told me, so we must re-enter the whole network config if console-conf did not finish? [15:07] you should really read your mails [15:07] vigo: I don't knwo that part [15:08] ogra_: danke [15:08] np :) [15:11] elopio: I'm not sure how to write unit tests for https://github.com/snapcore/snapcraft/pull/870 [15:11] PR snapcraft#870: sources: Add RPM source [15:12] I have integration tests but I don't have a unit test for it [15:12] the integration test works, btw [15:13] pitti, after wlan0 config I pressed done, and reboot before typing the e-mail. This way seems to work fine because after reboot it remembers the wi-fi ap, IP etc [15:13] so fine for me :) [15:15] hey, If I set a package in stage-packages, is there a way to ensure that the recomends for that hare pulled too? [15:18] Trevinho: I don't think so. Why not an explicit depends, if it's a hard requirement for you? [15:31] qengho: well, of course I can, but having a flag that allows to do that would be nicer :-) [15:31] than listing all the packages that might change with time... [15:32] woah, i updated ubuntu-image and my output file is way smaller than before, either this is a really cool feature, or my stuff is borken now [15:36] ogra_: is there a way to create custom core image with certain pre-configured network profile? [15:37] bzoltan: im sure there is a better way to do it, but what I have been doing is mounting the image and adding a script to run on a first boot to set the network config [15:38] PR snapd#2229 opened: interfaces/sytemd: enable/disable generated service units [15:38] Croepha: that would make the job for me :) Do you have that script to share? - sans credentials of course :) === lazyPowe_ is now known as lazyPower [15:43] morphis__ (cc niemeyer): notice my updated comment: https://github.com/snapcore/snapd/pull/2228#issuecomment-256955020 [15:43] PR snapd#2228: Allow fwupdmgr refresh [15:44] morphis__: so I backed out my approval after thinking about it [15:44] jdstrand: aye, didn't looked much into it yet, just got that from Tim [15:45] morphis__: if you guys go the plugs route, then I'll need to adjust the snap declaration (which I'm happy to do) [15:45] jdstrand: I am not really sure why Tim added it on the plug side [15:46] * jdstrand nods [15:47] bzoltan, I haven't tested this on uptodated core yet, still trying to get my custom kernel working, but this is what I did: https://gist.github.com/croepha/d006ec1f62dd4dcb1774c5116d1cebff [15:48] also, I kinda quickly just through that together so, hopefully it has everything you need and isn't messed up... let me know if you need help [15:50] bzoltan: basically sudo touch ${DIR4_}/custom_image/system-data/var/lib/console-conf/complete prevents the normal ubuntu config from running [15:51] bzoltan: and cat < PR snapd#2230 opened: Add an interface that allows clients to use media-hub over dbus [16:05] Croepha: cool, thank you. I will try that [16:23] PR snapcraft#878 opened: Added a fix for cases where modpbrobe append options to the line, ie.… === Guest42345 is now known as ahasenack === ahasenack is now known as Guest79971 === AdmWiggin is now known as tianon [16:35] So I've added a package to my snaps build-packages. How do I get snapcraft to pull that? Snapcraft pull doesn't seem to work and I can't tell if the package is broken or if it was never pulled... [16:45] it doesn't seem to be pulling any of the "build-packages" is there a way to debug this? [16:47] Croepha: the file system layout is rather different with the RPi3 latest image -> http://pastebin.ubuntu.com/23393529/ [16:48] josharenson, build-packages are installed on the host-- you realize this? [16:49] bzoltan, the systemd/network dirs might not exist before first boot [16:49] Bug #1637445 changed: The serial vault fails to connect to serial-vault.canonical.com [16:49] Croepha: Ahh... so I better play with the once booted images. [16:50] that might be a good place to start, but if you actually just create the dirs after building the image that works too [16:53] kyrofa: yeah, I think its actually just, yet another, path issue [17:07] Bug #1637596 opened: Configure hook in tried snap with ecryptfs: permission denied [17:21] well, even after updating ubuntu-image i still have a problem, it looks like with my custom kernel, there is a problem with mounting the fs for boot.... IO charset iso8859-1 not found... going to compare the kernel configs now [17:23] Croepha, you wand vfat builtin ... (and the right nls codepage for iso8859-1) [17:29] Croepha: yeps, just creating that wlan0 file did the trick. Thanks for your help. [17:29] bzoltan: ok, no proble, [17:29] no problem* [17:29] ogra_: lame question... how can I turn this beast to be in a dev mode. I would like to apt stuff there [17:30] bzoltan: are you use that DNS is working, i had to start a systemd-resolve service tooo [17:30] you cant ... you can install the classic snap to have a classic "chroot" [17:30] sudo snap install classic --devmode --edge [17:30] then: sudo classic [17:30] that sets it up and gives you a shell where you can use apt [17:31] ogra_: cool, thanks [17:31] ogra_: I guess I need a classic env in order to snapcraft projects for armhf locally [17:32] yep [17:34] * ogra_ vanishes again [17:37] ogra_ so the kernel I built has CONFIG_VFAT_FS=y CONFIG_FAT_DEFAULT_CODEPAGE=437 CONFIG_FAT_DEFAULT_IOCHARSET="iso8859-1" which seems to match the vanilla kernel [17:38] im going to double check to make sure I didn't like bundle the wrong kernel or something really stupid [18:04] it looks like nls_iso8859-1 is a module in the vanilla kernel, and I guess I just forgot to include that one [18:04] building a new snap now, hopefully this time it works [18:50] Bug #1637611 opened: UTF-8 local via ssh is poorly handled === davmor2 is now known as davmor2_HOLS [19:40] yes === hikiko|off is now known as hikiko === JanC_ is now known as JanC [21:52] what's the current blessed way of defining the license for a snap? the docs still talk about a `license:` key in snapcraft.yaml, but snapcraft says that's deprecated and if it's included you get an error on second and subsequent builds [21:55] oh, i found the discussion in the mailing list. docs are lagging [21:55] https://lists.ubuntu.com/archives/snapcraft/2016-September/000978.html