[00:00] Anyway, here's my understanding: [00:01] The @click.version_option() command should be like @click.version_option(message.format(snapcraft.__version__)) [00:01] This get the version number and plugs it in the variable string [00:01] The same goes in the version module. click.echo(message.format(snapcraft.__version__)) [00:02] kyrofa, is this correct so far? [00:03] Essentially. The question is: where does `message` come from? [00:04] Yes [00:07] how do you ensure that the Mp3 or whatever you are sending is seen in servicesID ? [00:08] Hmm. Anyone run into this before? https://forum.snapcraft.io/t/raspberry-pi-2-edge-core-snap-seems-broken/3061/2 [00:09] Yes, that is essentially the question, kyrofa [00:09] gsilvapt, you've been right so far, can you continue? [00:09] I tried adding in version module, before click statement and didn't work. Tried adding to group section, not worked either [00:10] kyrofa, not really. The question is where should I define that variable [00:10] gsilvapt, take a look at snapcraft/cli/store.py [00:11] gsilvapt, notice the `_MESSAGE_REGISTER_PRIVATE` definition [00:11] kyrofa, yes, I'm trying to follow the example [00:12] gsilvapt, the module-level definitions (like the variable) are accessible from __init__ because we have an "import versioncli" there, right? [00:12] Which means you can access stuff via `versioncli.` [00:12] Uh, no sorry [00:12] Haha [00:13] Rather, the module-level definitions are available in the `.version` import [00:13] Note how in __init__.py we're using `from .version import versioncli` [00:13] All we need to do is also import the variable you define in there [00:13] It works the same way [00:15] kyrofa, so I added from .version import _MESSAGE_SNAPCRAFT_VERSION and I am able to use the command again without any errors [00:16] now I have to properly define the string to get prog name and version number [00:16] Right now the version command is always retrieving ($prog) and ($version) [00:16] There you go. Although since you've created it for use for other modules I wouldn't use an underscore at the beginning (that indicates that it's mean to be private) [00:17] Oh right, this is different in Python. A global variable does not take any underscore, right? [00:17] Well, it can, it depends on its usage [00:18] kyrofa, how about this case? [00:18] Python has very little rules, but lots of conventions [00:18] I would call it `SNAPCRAFT_VERSION_TEMPLATE` [00:18] I'm more used to work in Java and this stuff is specific anyway :P [00:20] Well, it's late again and I need to sleep. I hope we can work on this together some time soon. I think it is working now but the string is poorly defined. I need to tell Python where to get the variable values ($prog) and ($version) [00:20] %(prog) and %(version) instead of what I wrote before, lol [00:23] Yeah that's not too hard, happy to work on it later with ya [00:23] Thank you! Have a good night folks o/ [00:23] o/ [00:24] Night gsilvapt [00:35] PR snapcraft#1728 closed: beta [02:42] elopio kyrofa snapcraft#1781 is updated [02:42] PR snapcraft#1781: many: set rpath for elf files for classic === chihchun_afk is now known as chihchun === chihchun is now known as chihchun_afk [07:26] morning [07:27] sorry i'm later than usual, had to drive the kids to school & kindergarten [07:27] hey mborzecki, good morning! [07:27] mborzecki: no worries [07:28] mvo: thanks for the review & suggestions btw [07:29] mborzecki: your welcome, but I may have been to simplistic, I think there are more cyclic case [07:30] mborzecki: something like "a: after-b, b: after-c, c: before-a [07:30] mborzecki: which I guess (I need to look again) your code addressed but mine did not [07:30] zyga suggested trying to do topo sorting on a graph, so i tried to play with this yesterday, but the naiive approach that i did over the weekend beat the 'smart' in all the cases i tried [07:31] mborzecki: heh :) yeah, our inputs are pretty small [07:31] mborzecki: I will look at this again with a fresh mind today [07:33] the benchmarks would suggest that we're loosing on doing extra allocations when creating the 'extra' data structures either dfs or khan may need [07:34] mvo: one thing that bugs me, is that if you add 'after' does that imply 'Requires=' in the service file? [07:37] mborzecki: I think it does not imply this, one is about ordering, the other about dependencies, they are often used together but orthogonal. i.e. you can express "if this thing.service is installed I want to start before. without the requires it also says: but I don't care if it is there or not" [07:38] this post mentions 'with' https://forum.snapcraft.io/t/snap-service-start-ordering/1470/13 this might map to Requires [07:38] mborzecki: yeah, I think there is language in there (as a next step) to express these dependeices as well [07:40] let's land this one first and we'll know if there's need for more ways to express dependencies [07:58] good morning folks [07:59] kalikiana: morning [08:06] PR snapd#4361 opened: devicestate: fix misbehaving test when using systemd-resolved [08:36] mvo: I uploaded a new version of the test-snapd-gnome-online-accounts snap (the original wouldn't run due to library incompatibility), which seems to have been held up by the automatic review process. Are you able to do anything about that? [08:41] jamesh: probably, let me check [08:43] jamesh: did you request a manual review? [08:43] jamesh: if not, please do [08:44] mvo: I've done that now [08:45] jamesh: approved [08:45] thanks [08:45] jamesh: yw, shall I pushlish it as well? [08:46] mvo: I think I've got it published right now [08:47] thanks for the help [08:47] great, your welcome [09:04] mvo: is edge building still broken? afaict current edge doesn't contain yet the configure-snapd fix [09:06] pedronis: let me check, I though I triggered a manual build the other day. auto-building is broken [09:06] pedronis: let me trigger it now [09:19] o/ [09:20] man, I overslept [09:25] I ended up poking at spread in qemu until way too late :-) [09:26] but, now i'm off to physio gym thing, so it evens out [09:26] Chipaca: what did you find? [09:26] o/ [09:26] ah, ttyl [09:26] zyga-ubuntu: i found that it's weird :-) [09:26] pfff, we all knew that [09:26] ;-) [09:26] enjoy; let's talk soon [09:26] zyga-ubuntu: i've got two straces, one that dies and one that doesn't [09:27] zyga-ubuntu: I don't see the difference [09:27] pastebin thenm [09:27] them* [09:27] +1 [09:27] o/ mvo [09:27] I have a pastebin I'm chewing through since yesterday, today I need to do some practical experiments on non-test machines [09:27] cachio_: fwiw, trying to reproduce the install-cache test failure you mentioned on ubuntu-core-16-64 seems to be ok for [09:27] well, on non-spread test machines running core [09:28] cachio_: I try harder [09:28] hey zyga-ubuntu [09:28] we are doing some very weird things on core [09:28] (lots of red lights in my mind) [09:28] the failure of those tests I mentioned yesterday, with the stale core detection, is deeper [09:28] even if we reboot the system is really broken IMO [09:28] zyga-ubuntu: oh [09:29] http://paste.ubuntu.com/26123959/ and http://paste.ubuntu.com/26123957/ [09:29] zyga-ubuntu: tell me more? [09:29] zyga-ubuntu: please [09:29] mvo: the pastebin I sent you pics of yesterday was from the console of a core system [09:29] mvo: it seems that we mount / twice [09:29] mvo: with a load of things in between [09:30] mvo: my understanding of how this works means that _all_ of the things we mounted before 2nd / are just wasted memory and hidden entries [09:30] mvo: I don't yet know why we do that [09:30] zyga-ubuntu: could it be a pivot root thing going on? [09:30] mvo: it might be but pivot root would leave / mounted elsewhere [09:30] you cannot pivot root / / [09:30] (no op) [09:30] zyga-ubuntu: well, we need to work on this for base-18 anyway, so a good time to cleanup [09:30] http://pastebin.ubuntu.com/26118789/ [09:30] mvo: good to hear [09:31] mvo: I'll poke at a real pi3 device and see if this is consistent [09:31] zyga-ubuntu: have you looked at the initramfs code? the writable path things makes it messy, I'm all for a cleanup [09:31] (and historic baggage) [09:32] ok, i'm off for real now. I'll leave you guys to ponder which one of those two traces is the one that kills the whole session. [09:33] to me they look identical [09:33] Chipaca: ha, good plan :) [09:33] (but yes, i know which is which) [09:34] it's as if one of them closed stdout, or stderr, or stdin, or whatever it is [09:34] hmm [09:34] * zyga-ubuntu hates pastebein with moronic need to auth [09:34] no wget [09:34] zyga-ubuntu: I'm looking at the pictures you pasted, I see core_ mounted twice on / is that what you are concerend about? [09:34] mvo: yes [09:34] mvo: IMO that is wrong [09:34] zyga-ubuntu: blame spammers [09:34] PR snapd#4362 opened: interfaces: update fixme comments [09:35] * Chipaca runs off before he gets sucked in again [09:35] zyga-ubuntu, ^ hey, one of the changes you requested - just an update of comments, no code change [09:35] zyga-ubuntu: yeah, let me wander into initramfs land again /me puts on asbestos cloth [09:35] mvo: yeah [09:35] zyga-ubuntu: did i ever tell you about debugging docker on arm64 where cloneflags was ignored so pivot_root was called in the default namespace [09:35] mvo: I'll gladly work on fixing this [09:35] zyga-ubuntu: that was quite confusing :) [09:36] mwhudson: haha, that must have been quite the alice in wonderland moment [09:36] especially as the host system didn't have an awful lot on it to distinguish it from the docker image i was testing with [09:36] zyga-ubuntu: we need to be careful this sh code is undertested [09:36] mwhudson: woah :) [09:36] i run docker run foo ... and docker itself disappears? [09:38] mvo: I'm acutely aware of that :/ [09:39] zyga-ubuntu: I know :) [09:40] zyga-ubuntu: I may have found the bug already and its amazingly stupid [09:40] zyga-ubuntu: my conservative self is not wanting me to say this, but: I really want to write the initramfs stuff in something sane that is not sh [09:40] !!! [09:41] that was quick [09:41] mvo: skunkworks project to rewrite that +1 [09:41] mvo: just not in go (we cannot) [09:41] zyga-ubuntu: no go because of the setns issues? [09:41] mvo: so, this is interesting: on bare metal I don't see that problem: http://pastebin.ubuntu.com/26123995/ [09:42] with threading and all that [09:42] mvo: not only setns but yes [09:42] mvo: 1thread requirement [09:43] mvo: the fact that / is mounted from initramfs and then core is mounted again from userspace is also a bit annoying but I don't know if it's worth fixing [09:43] http://pastebin.ubuntu.com/26123999/ <- core snap mounted twice [09:43] zyga-ubuntu: yes [09:44] pedronis: new core snap in edge should have the configure-snapd fix now, I did a bit of manual handholding [09:45] mvo: also interesting: http://pastebin.ubuntu.com/26124012/ the two snaps we mount from initramfs are mounted as _writable_ and not as auto-clear [09:45] mvo: the auto clear is probably irrelevant but the writable bit should have never been set [09:45] mvo: this allows an attacker to write to loop0 and change the squashfs bits [09:47] zyga-ubuntu: aha, that should be an easy fix, one minute [09:48] zyga-ubuntu: http://paste.ubuntu.com/26124029/ [09:49] mvo: +1 [09:50] mvo: every time I look at that pile if shell I feel bad [09:51] zyga-ubuntu: yeah, makes me want to scrub myself [09:51] mvo: can you fix inconsistent tabs/spaces [09:51] zyga-ubuntu: anyway, lets see what we can do to fix things [09:51] mvo: if you are going to make changes there [09:51] mvo: thank you, refresh to it now worked [09:51] mvo: I'll check other systems, I don't understand if the brokenness I saw is specific to spread [09:52] pedronis: yay [09:52] Issue snapcraft#1702 closed: Apply guidelines from design to error messages with reasons for errors [09:52] PR snapcraft#1640 closed: cli: add what, why, and how to fix to the common errors [09:52] pedronis, hey, can you take a look and weight on #4301? I've addressed the last bit from Gustavo (AttrGetter -> Attrer renaming), but aparently he didn't manage to re-review yesterday, would be good to land it as it's a prerequisite for others [09:52] PR #4301: interfaces: PlugInfo/SlotInfo/ConnectedPlug/ConnectedSlot attribute helpers [09:52] mvo: and btw, I don't see any use of pivot_root [09:52] mvo: is it hidden somewhere or are we just missing it? [09:53] zyga-ubuntu: I think I may misrember actually [09:53] zyga-ubuntu: now that I look at the code its less clear to me what is going on [09:53] pstolowski: will look in a bit [09:53] mvo: I'll check dragon and I'll get an amd64 image for qemu [09:54] PR core-build#24 opened: ubuntu-core-rootfs: mount os/kernel snaps RO [09:55] Issue snapcraft#1663 closed: patchelf with common rpath (not runpath) [09:55] PR snapcraft#1781 closed: many: set rpath for elf files for classic [09:55] mvo: can you test your PR in practice and see that losetup --json shows it's correct then? [09:55] mvo: a manual test if you can do that [09:56] mvo: I approved the PR already [09:57] mvo: we really have to rewrite mount units [09:57] mvo: on dragon I see three generation of mount files we write [09:58] mvo: a simple rewrite that fixes things for next boot would be good [10:05] mvo: uh, my mistake, I was blinded by odd double-sided print [10:05] mvo: stop looking for double / :) [10:05] mvo: so now I'm back to square one, what is wrong there :/ [10:06] zyga-ubuntu: rewrite fix is still needed? or should I also scratch that? [10:06] PR snapd#4363 opened: cmd/snap-mgmt: add more directories for cleanup and refactor purge() code [10:07] zyga-ubuntu: rewriting with the before=snapd.servie? [10:07] mvo: no, that is still needed [10:07] mvo: that should have been there forever [10:08] zyga-ubuntu: ok, but "just" that - or do we need more rewrite? [10:08] mvo: not sure I understand the difference; we need to be able to rewrite mount units and and we ought to rewrite mount units in the field to correct for past mistakes [10:10] zyga-ubuntu: we are in agreement, I was wondering if there is a new bug that requires us to rewrite them [10:10] aha, sorry for being confusing then [10:10] zyga-ubuntu: I will have a look. no worries, communication is hard :) [10:10] (it always is!) [10:10] mvo: the pastebin I sent you, I was holding it wrong, litteraly :) [10:10] I looked at head of both inside and outside parts [10:11] zyga-ubuntu: heh [10:11] zyga-ubuntu: we have 256 files in etc and var in my base-18 WIP snap :/ there is quite some work to get rid of those if we want to get rid of the writable-path magic [10:12] * mvo gets to work [10:12] mvo: where do you do this work? [10:12] mvo: I'd love a new repo for base-18 [10:13] zyga-ubuntu: lp:snap-base-18 [10:13] zyga-ubuntu: so far very minimal/experimental/WIP but its a start [10:14] bzr or git? [10:14] zyga-ubuntu: git [10:14] zyga-ubuntu: what is bzr? [10:14] hhaha [10:14] * zyga-ubuntu checks how to clone from launchpad [10:14] zyga-ubuntu: https://github.com/snapcore/base-18 :P [10:15] * mvo is afk for some minutes [10:15] mvo: so is it on launchpad or github? [10:21] mvo: I've started sending PRs [10:36] Hi guys, does anyone know how can I enable the driver from bluetooth on the Raspberry Pi with Ubuntu snappy core? [10:40] brunosfer: it should already be enabled, [10:40] zyga-ubuntu: github [10:40] ok [10:40] mvo: I send one trivial PR [10:40] mvo: how do you test/use this? [10:41] mvo: one more [10:42] zyga-ubuntu: well, I build a snap :) thats all there is for tests so far [10:43] zyga-ubuntu: good stuff, thank you! [10:43] mvo: ha, well, I didn't think of that :) [10:43] zyga-ubuntu: but yeah, its a good question, we need more automation and travis and all that here [10:43] mvo: snapcraft fails, did you run it via fakeroot? [10:44] zyga-ubuntu: yes, it needs to run as root currently, I think fakeroot will work [10:44] zyga-ubuntu: the LP builder runs it as real root [10:44] * zyga-ubuntu tries fakeroot snapcraft [10:45] zyga-ubuntu: what is annoying is that in order to get to an empty /etc we need to undo the alternatives system i.e. awk which is provided via an alternative [10:45] (and more, w, pager) [10:46] hmm, we need real real root [10:46] zyga-ubuntu: ok, I need to look into why, iirc because of chroot, right? [10:46] zyga-ubuntu: *maybe* fakechroot works [10:47] mvo: no, because of chroot [10:47] mvo: anyway, that's fine, [10:47] zyga-ubuntu: fakechroot may work [10:47] mvo: just trying to see where this is going and what's in the snap [10:48] zyga-ubuntu: I hope it goes into a minimal base that can boot and that works with an empty /etc/ /var so that we can do factory reset easily and don't need the writable-path hacks [10:48] why dont you just grab an ubuntu-base tarball from cdimage and modify it as needed ... [10:48] ogra_: that would also be an option, right now it is using debootstrap to get that [10:48] (with a bit of fiddling oyu can even do that completely without chroot calls) [10:49] mvo, yeah i was guessing that ... [10:49] ubuntu-base is pretty much a minbase debootstrap though [10:51] mvo: there's some things that should be easy to kill [10:51] ogra_: indeed, interessting [10:51] mvo: etc/selinux [10:51] zyga-ubuntu: yeah /etc/{apt,dpkg} [10:51] mvo: /etc/apt [10:51] yeah [10:52] mvo: /etc/kernel/install.d [10:52] zyga-ubuntu: and some stuff we need to look at and upstream, i.e. if adduser.conf has different defaults than adduser itself we should ensure that gets fixed [10:52] just dont kill /usr/share/doc//copyrights.gz ... [10:52] ogra_: make that a fuse https FS [10:52] ;-) [10:52] ogra_: yeah, no worries :) [10:52] :) [10:53] ogra_: its mostly based on the current livecd-rootfs scripts [10:53] ogra_: the new skeleton [10:53] ogra_: but using a different "base" [10:53] yeah .. many of these scripts can also be dropped [10:53] we (I) didnt do a good job of throwing away old stuff thats not needed anymore [10:54] and some bits that live in ubuntu-core-config should rather move to the build scripts ... and vice versa ... someone should do an audit of all this [10:55] ogra_: yeah, the work on this is (slowly) starting with base-18. unfortuantely quite a bit of work but at the same time exciting because of the cleanup potential [10:55] yep [10:56] /etc/bindresvport.blacklist <- why do i have this file in /etc :( [10:56] i'm still curious how it will work out ... but looking massively forward to have bases for commercial projects ... they will massively solve issues i'm running into since i started in the new team [10:56] anyway, [10:57] ogra_: fingers crossed, but I'm excited aboutt the opportunities, all sorts of crazy bases [10:57] yeah [11:02] mvo: some more stuff in 3 [11:13] https://github.com/snapcore/base-18/pull/3 [11:13] PR base-18#3: Document how to build the project locally, trim /etc, ignore build artefacts [11:14] mvo: do we need /etc/init.d? [11:15] mvo: /etc/ld.so.conf.d/libc.conf also looks like something to remove [11:16] mvo: /etc/logrotate.d/{alternatives,dpkg} [11:16] mvo: /etc/profile.d [11:16] mvo: /etc/rc[0-6S].d [11:17] mvo: anyway, I think it would be fantastic if we had a way to kvm-boot this for testing [11:17] mvo: take a kernel snap, base-18 and gadget and boot it some way for very fast iteration [11:25] * kalikiana coffee break [11:36] okay, that was a lie, I ended up making tea [11:38] * zyga-ubuntu wonders when we will see snaps on windows 3.11 http://assets.metacade.com/emulators/win311vr.html [11:38] mvo, hey [11:40] mvo, The changelog for 2.30 rc2 doesn't contain the 2.29.4.2 changes [11:40] mvo, is it ok? are those changes included? [11:48] cachio_: thanks for catching that, I will double check [11:49] mvo, apart of that I have seen some issues during the execution [11:49] zyga-ubuntu: thanks! yeah, I think that must be the goal, right now the initramfs stuff cannot boot bases - just core [11:49] cachio_: tell me more, you mentioned yesterday that the install-cache test fails ? [11:50] mvo, yes [11:50] and the error seems to be genuine [11:51] mvo, similar issue for catalog-update [11:52] mvo, https://paste.ubuntu.com/26120827/ [11:52] this is a reexec in pi3 for all the errors that I saw [11:53] some of them are test errors, I am working on fixing them [11:55] cachio_: for the cache-install failure, does that happen just on arm or on any core device? [11:55] mvo, it was in any core device [11:55] both tests failed in any core device [11:55] the install-cache and the catalog-update [11:56] mvo, I have to go to the doc now [11:56] I'll be back for the standup [11:56] cachio_: thanks, I need to dig into this then, I was not able to reproduce it from a release/2.30 git checkout and running spread from there, sounds like I need to create an image first [11:56] cachio_: sure, see you. and thanks for the update [11:56] mvo, np [11:57] hello everyone [11:59] hey sergiusens [11:59] sergiusens: what's the weather like where you live? [11:59] i bet it's better than here [11:59] zyga-ubuntu: why do you ask? [12:00] mvo: ? [12:00] just curious [12:00] zyga-ubuntu: I was wondering if you want to fly there ;) [12:00] zyga-ubuntu non deterministic [12:02] zyga-ubuntu, definitely better than yours :) [12:03] zyga-ubuntu https://www.facebook.com/photo.php?fbid=10159702028205367&set=a.10154038137770367.1073741826.844960366&type=3&theater [12:04] it might rain, it might not; it might get cooler (like 18), it is 25 now, and it might also go up to 39 [12:04] what did you do to your knee ? [12:04] depends on if i decides to rain [12:04] ogra_ https://en.wikipedia.org/wiki/Chondromalacia_patellae [12:05] ouch [12:09] sergiusens: nene, eso no se hace! :-( [12:15] hmm tarvis jobs mostly timing out today? [12:18] mborzecki: sad travis day [12:18] I'm working entirely locally [12:21] btw sergiusens I tried out the slot feature in the calendar. Can you see any difference on your end? I'm not sure if it handles regular appointments correctly. You should see at least one 'unavailable' slot tomorrow. [12:25] PR snapd#4362 closed: interfaces: update fixme comments [12:28] PR snapd#4301 closed: interfaces: PlugInfo/SlotInfo/ConnectedPlug/ConnectedSlot attribute helpers [12:30] #4354 needs a 2nd review [12:30] PR #4354: tests/lib: introduce helpers for setting up /dev/random using /dev/urandom in project prepare [12:58] hi. does the automated github build service only build from the master branch, or do all branches get built? [12:59] mcphail: hey, do you mean on build.snapcraft.io? [13:02] yes, zyga-ubuntu [13:02] lunch time [13:04] mcphail: ah, I don't know what's the state of that, sorry [13:04] zyga-ubuntu: ok, thanks [13:10] ogra_ is there an amd64 Ubuntu Core image for download? [13:10] sure [13:10] sergiusens: yes [13:10] sergiusens, http://cdimage.ubuntu.com/ubuntu-core/16/ [13:10] pick your channel [13:11] sergiusens: http://paste.ubuntu.com/26125012/ [13:11] kalikiana slots are interesting new concept in the calendar, but I think it is more of a thing for another person to click to get a slice of your time [13:11] note that this is zero padded for better use with kvm though (unlike the other arches) [13:11] sergiusens: make boot; make login [13:11] kind of like a doctor's appointment thing [13:12] ogra_ zyga-ubuntu thanks, I won't be needing any of the automation, it is for a potential demo at the IoT meetup I am presenting at today [13:13] sergiusens: this is not much automation, just remembers the URL and the long command [13:16] mvo: btw, why are you using >100 hook numbers? [13:18] mvo: sad [13:18] zyga@kaedwen:~/snap-base-18$ grep -R bindresvport.blacklist . [13:18] Binary file ./prime/lib/x86_64-linux-gnu/libc-2.26.so matches [13:18] Binary file ./prime/lib/x86_64-linux-gnu/libc.so.6 matches [13:26] zyga-ubuntu: well, we can patch that [13:26]