[02:51] Does sudo snap install snap-codelabs no longer work or do I have something setup incorrectly. === mup_ is now known as mup [04:53] hi [04:53] I am porting snappy to bubblegum96 board, the Soc has a fixed layout that the bootloader.bin ’s offset is at 2097664 and the u-boot is at 3145728.It works when I use https://github.com/uCRDev/Bubblegum96-Snappy/blob/master/gadget/gadget/meta/gadget.yaml and ubuntu-image version 0.10. [04:54] Since the ubuntu-image is updated to version 0.12, it may cause the error like this: [04:54] sudo /snap/bin/ubuntu-image --channel stable --image-size 2G --extra-snaps bubblegum96-gadget_16.04-1.1_arm64.snap --extra-snaps bubblegum96-kernel_3.10.0_arm64.snap -o bubblegum96.img bubblegum96.model [04:54] Fetching core [04:54] Copying "bubblegum96-kernel_3.10.0_arm64.snap" (bubblegum96-kernel) [04:54] Copying "bubblegum96-gadget_16.04-1.1_arm64.snap" (bubblegum96-gadget) [04:54] bubblegum96-gadget already prepared, skipping [04:54] bubblegum96-kernel already prepared, skipping [04:54] gadget.yaml parse error: mbr structures cannot be larger than 446 bytes. [04:54] Use --debug for more information [04:55] ubuntu-image had fixed a bug so that causes my issue https://bugs.launchpad.net/ubuntu-image/+bug/1630769 [04:55] Bug #1630769: part with role='mbr' should be stricter about its parameters === chihchun_afk is now known as chihchun [08:29] good morning [08:47] PR snapd#2602 opened: overlord/snapstate: remove restriction on ResetAliases [09:04] Issue snapd#2603 opened: Disk free space left check [09:12] PR snapd#2604 opened: interfaces: add classic-dimension interface [09:16] PR snapd#2605 opened: overlord,overlord/snapstate: have UpdateMany retire/enable auto-aliases even without new revision [09:30] PR snapd#2498 closed: interfaces: add new upower-control interface [09:38] Issue snapd#2557 closed: interfaces: connect tun-based serial port [09:52] Issue snapd#2503 closed: chattr code (tests/main/chattr/task.yaml) fails on ppc64el [09:52] Issue snapd#2504 closed: interfaces-upower-observe snap test fails on ppc64el [09:52] Bug #1655592 opened: interfaces-upower-observe snap test fails on ppc64el [09:52] Bug #1655593 opened: chattr code (tests/main/chattr/task.yaml) fails on ppc64el [09:52] Bug #1655594 opened: expect based tests fails on ppc64el [09:53] Issue snapd#2502 closed: expect based tests fails on ppc64el [09:54] sergiusens: trying to do my classic snap, I added the .desktop file to the snap and suddenly snapcraft hates me again with: 'classic confinement requires the core_dynamic_linker to be set' [09:56] ogra_, ping [09:56] hey [09:57] ogra_, yesterday, thanks for answering my question regarding the slow build. do you mean to replace all of strings "http://ports.ubuntu.com/ubuntu-ports/ " in the source.list to " http://localhost:9999/ubuntu-ports"? [09:58] ogra_, currently the sources.list file is like http://paste.ubuntu.com/23780838/ [09:58] liuxg, yep ... then you need to use apt update once to initialize the proxy db ... [09:59] hmm [09:59] * zyga feels spring in the air :) [09:59] ogra_, how to initialize the proxy db? once it is installed, it is initialized? [09:59] that sources.list misses all basic entries [10:00] no, it is initialized with the first apt-get update [10:00] (just means it downloads the index files once from the archive) [10:01] Bug #1639284 changed: Cant start any snap application on Xenial [10:02] ogra_, ok.. thanks. this is the default sources.list file in my pi device. http://paste.ubuntu.com/23780856/ do you mean I need to add more stuff there? [10:03] I'm trying to remove the ubuntu-core snap (so I can install the core snap), and I can't work out how to do so. `sudo snap remove ubuntu-core` gives "error: cannot remove "ubuntu-core": snap "ubuntu-core" is not removable". [10:03] liuxg, no, that one looks fine .,.. your first one was missing the top part [10:03] https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1589210 suggests that running snap disable first allows removal, but that also doesn't work. [10:03] Bug #1589210: snap remove failed ubuntu-core snap package [10:03] ogra_, I think your solution is very useful since a lot of developers could not compile their target due to the slaggish network on-site. [10:03] Odd_Bloke: Just doing that myself. I think you are stuck 'apt purge snapd', trashing all your snaps and their data, then 'apt install snapd' [10:04] ogra_, yeah, I just cut some of part of it, it is my fault. thanks for clarification. [10:06] why is snapcraft.io not served with SSL? [10:06] sergiusens, this python encodings error makes me sad [10:07] stub: Aha, fair enough; I knew I'd have to trash all snaps, didn't realise snapd itself had to go. Thanks! [10:08] stokachu we can skip lunch and fix it :-) Have you tried removing the stage/prime entries in your parts? [10:08] sergiusens: see my ping above? [10:08] also, can classic snaps access the system clipboard? [10:09] sergiusens, yea i cleared those out [10:09] icey nope, just logged on to irc after 6 days [10:09] sergiusens, we can look after lunch :) [10:09] sergiusens: it was 10 minutes ago :-P [10:09] how about this: sergiusens: trying to do my classic snap, I added the .desktop file to the snap and suddenly snapcraft hates me again with: 'classic confinement requires the core_dynamic_linker to be set' [10:09] icey you need the core snap installed [10:09] rather than the ubuntu-core snap, I'm going to have to do that uninstall dance aren't I? [10:10] icey yeah [10:10] I just filed https://bugs.launchpad.net/snappy/+bug/1655599 since this is technically a dataloss bug (if people have production data in their snap containment) [10:10] Bug #1655599: No migration of ubuntu-core to core [10:11] sergiusens: after removing everything, how do I replace ububtu-core with core [10:11] Bug #1655599 opened: No migration of ubuntu-core to core [10:11] oh really, uninstall snapd!? [10:11] https://bugs.launchpad.net/snapcraft/+bug/1650946 [10:12] Bug #1650946: unhelpful error when building a classic snap: classic confinement requires the core_dynamic_linker to be set [10:14] ogra_, one more question, do I need to change the sources.list in the ubuntu core snap or in the classic snap? sorry [10:15] in classic, i dont think it is even writable in the core side [10:15] sergiusens: :-/ '[Errno 2] No such file or directory: '/mnt/Media/projects/alacritty/parts/desktop/install/meta/gui'' [10:15] ogra_, ok. I got it.. yeah, that is currently what I am doing now :) [10:16] icey wait, let me get back to you on this, this is not how setup/gui works [10:17] ogra_, this is currently how sources.list looks like in my classic snap http://paste.ubuntu.com/23780885/ [10:17] sergiusens: sure, I'll grab you in a bit, copying the original source on it :) [10:17] that shoudl work fine [10:18] ogra_, thanks. I am now trying to apt-get, it is still slow. the first time apt update is slow, right? after that, it should be fine. [10:19] how do you clean a specific part again? [10:19] thought it was snapcraft clean pull [10:19] liuxg, yeah, it only pays off the second time you access a package indeed [10:19] liuxg, but from then on it will only download packages that are newer in the archive [10:19] stokachu: `snapcraft clean {STAGE}` just worked for me [10:20] ah stage [10:20] icey, thanks [10:21] ogra_, this is awesome. You saved the world:) [10:21] :) [10:22] liuxg, under http://localhost:9999/ you can also see some very basic statistics [10:23] (or rather http://:9999/ since you need a browser fro this indeed) [10:25] ogra_, this is what I am building now http://imgur.com/a/6UqDY, is it downloading some new packages? I have already done the apt-get update. [10:27] liuxg, it will have to download the index files every time to compare if something was updated [10:27] also, if it is your first build it indeed has to download the packages once, like i said above [10:28] ogra_, ok. thanks. I will check it. yeah, the first time build seems to be slow anyway. now, it says 30 mins [10:29] it needs to pull all debs into the proxy once indeed ... [10:33] ogra_, still, I think this is cool. for hackathon event, it is very needed. sometimes a single line change takes the build for 20 mins, basically, it makes the event useless. [10:34] well, on a hackathin i'D set up a dedicated machine and have everyone else point to it [10:34] *hackathon [10:34] and if you knwo what you are building you can do one build in advance and fill the proxy cache ahead of time [10:35] ogra_, but for developers, they may not know what are needed. Last time, I saw a developer using nodejs, the package took years to get downloaded. [10:36] yeah [10:38] ogra_, do you know how to set up a machine like what you said so that everyone can point to the machine? [10:38] you just do the same you just did above ;) and have people use the ip instead of localhost [10:40] ogra_, oh, you mean a Pi device? so everyone changes the sources.list to point to the same pi device? [10:40] or a laptop [10:40] ogra_, but a laptop is x86, right? most people work on arm. [10:41] whatever you have around ... the point is to use one central machine instead of everyone having his own proxy [10:41] ogra_, I got it. May I will have a try [10:41] the underlying arch doesnt matter ... packageproxy works on all arches ;) [10:42] and for all arches ... i.e. for x86 packages you would simply point to http://localhost:9999/ubuntu ... instead of /ubuntu-ports [10:44] ogra_, how does a x86 provides the debian packages for the arm architecture. [10:44] you are aware that the ubuntu archive is just a webserver ? [10:44] ogra_, yeah [10:45] the packageproxy just copies the content ... doesnt matter what arch [10:45] ogra_, ok. I got it. so it archives all of the debian packages whenever it is used to download from the archieve [10:46] for the main archive it serves them under /ubuntu ... for poirts.ubuntu.com it serves them under /ubuntu-ports [10:46] ogra_, so the proxy snap should be installed on the laptop machine, right? [10:46] yep [10:47] or on a central rpi if you have one spare ... or on a PC at home if you have proper bandwith to serve the archive from there ... [10:47] (or in the office if you have an external IP for the machine) [10:48] wherever you like ;) === chihchun is now known as chihchun_afk [10:48] ogra_, thanks. I got what you mean. this is truly useful for events. I had headaches to let developers develop onsite.. most of the time, they go home to finish the work. [10:49] PR snapd#2606 opened: overlord/snapstate: share code between Update and UpdateMany, so that it deals with auto-aliases correctly [10:56] hello all [10:56] ogra_: I'm building an image with edge, and I get snap 2.20.1, is that equivalent to 2.21? [10:56] obviously not [10:57] 2.20.1ubuntu1 is the latest release [10:57] ogra_: I see [10:59] ogra_: So when 2.21 is available, will it be in both edge and candidate channels? [10:59] yep [11:00] well, edge first, but it will move to candidate eventually indeed [11:01] Ok [11:06] PR snapd#2607 opened: many: move interface test helpers to ifacetest package [11:12] ogasawara: hey, do you remember the conversation about putting kernel snap projects on github.com/snapcore/snapd similar to how the gadget snaps are there now? [11:12] ogasawara: did you have a chance to discuss that in the kernel team? [11:14] zyga: yep, did you not see the email I sent you about it? [11:14] * ogasawara will resend [11:16] zyga: sent [11:16] PR snapd#2608 opened: don't install govendor each time [11:17] ogasawara: thanks, I got it now [11:38] PR snapd#2555 closed: many: implement 'snap aliases' === didrocks1 is now known as didrocks [12:11] ogra_, when I build my project for the second time, I get the error like http://paste.ubuntu.com/23781469/, what could be the reason for it? thanks [12:12] sergiusens: elopio the "build dir is empty" bug.. I'm seeing similar with nodejs plugin, not just rust. Plausibly same bug or new one? [12:12] hmm [12:13] liuxg, can you ping localhost ? [12:14] ogra_, yes http://paste.ubuntu.com/23781476/ [12:16] weird, works fine here [12:16] ogra_, inside classic http://paste.ubuntu.com/23781481/ [12:17] and if you just call apt update in classic ? [12:17] does it update ? [12:17] PR snapd#2608 closed: don't install govendor each time [12:18] or install a random package [12:18] ogra_, the snap is there [12:18] that wasnt my question [12:18] does apt update inside the classic env work [12:18] or can you apt install some random package [12:19] ogra_, it is like http://paste.ubuntu.com/23781493/ [12:19] looks like the proxy isnt running ... could it be that you ran out of diskspace or some such ? [12:19] ogra_, http://paste.ubuntu.com/23781496/ [12:20] ogra_, http://paste.ubuntu.com/23781501/, I have a 16G disk [12:22] systemctl status snap.packageproxy.approx.service [12:22] ? [12:22] (outside of classic obviously) [12:24] ogra_, http://paste.ubuntu.com/23781511/, it is running. I just disable and enable it again. [12:25] well, i have never seen it not work and i use it since about two years on various machines ... pretty weird [12:26] ogra_, http://paste.ubuntu.com/23781527/ [12:27] anything in syslog about why it fails ? [12:27] (thats an extremely dumb app, it can not fail, this is super strange) [12:27] PR snapd#2609 opened: store: request no CDN via a header using SNAPPY_STORE_NO_CDN envvar [12:30] is 9999 taken by anything else ? [12:31] ogra_, no, I do not think it is taken. no other software uses the port [12:33] do you perhaps have a stale lockfile in /var/snap/packageproxy/3/lockfile.lock ? [12:34] ogra_, I am now doing it again. [12:34] doing what again ? [12:35] ogra_, http://paste.ubuntu.com/23781554/ [12:35] ogra_, you are right. [12:35] remove it [12:35] thought this is extremely odd ... it is really a very dumb script, i havent seen it fail in years [12:36] the only thing i could imagine is that you either run out of diskspace (which you dont) or ram to make it fall over [12:36] ogra_, how could that happen? I must win the first prize :) [12:37] ogra_, now, it seems to be right. it starts to snap ... [12:40] ogra_, thanks.. this time, the build is really fast :) [12:40] ogra_, we should let the rest of the developers know this trick! [12:41] are you building something that could use a lot of ram during the build process? [12:42] ogra_, I do not think so. it is just dump plugin and compile some c files. [12:42] liuxg, well, if you see anything like the above again, please let me know [12:42] ogra_, by the way, if someday I do not want to have the cached files, how can I remove them? [12:42] ogra_, sure, I will observe it. [12:43] either snap remove packageproxy ... or just rm -rf /var/snap/packageproxy/3/var/cache/approx/* [12:43] ogra_, alright. many thanks [12:43] if you want to toggle it dynamically, just use a default sources.list and copy them back and forth as you need [12:45] ogra_, yeah, that makes sense. In fact, i did not keep the old one :) I will find it back. [12:45] it is in the core env ... if you exit classic [12:45] remember ? it is readonly there [12:46] ogra_: is there any kind of autoclean of the cache? [12:46] ogra_, yeah, it is the same. [12:46] ogra@pi3:~$ cp /etc/apt/sources.list . [12:46] ogra@pi3:~$ sudo classic [12:46] (classic)ogra@pi3:~$ sudo cp sources.list /etc/apt/ [12:46] and you are back on the default sources.list ;) [12:48] didrocks, there is /var/snap/packageproxy/3/config.yaml ... you can adjust such stuff if needed [12:48] nice! [12:48] ogra_, right. I got it. thanks [12:49] didrocks, i didnt port all config options yet but largely http://manpages.ubuntu.com/manpages/xenial/man5/approx.conf.5.html applies [12:51] ogra_: good work! thanks for the reference :) [12:51] old stuff ... i need to update it again so that snap set and get work ;) [12:52] i created it in 15.04 snappy originally ... :) [12:52] adjusted it for newer changes) [12:52] *(only adjusted it... [12:52] PR snapd#2607 closed: many: move interface test helpers to ifacetest package [13:13] PR snapd#2601 closed: overlord, store: move confinement filtering to the overlord (from The Store) [13:17] fgimenez: ping [13:18] fgimenez: Hey. Do you have a doc/spec or any good description about re-exec? It's planned for 2.21, want to be prepared to check it. [13:23] rvr: give me some minutes and I send you something [13:23] mvo: Thanks [13:24] hey rvr, nope sorry, none that i know, thx mvo! === hikiko is now known as hikiko|ln [13:29] PR snapd#2610 opened: interfaces/browser-support: add @{PROC}/@{pid}/fd/[0-9] w and misc /run/udev [13:46] PR snapd#2599 closed: interfaces: add new-style interfaces === ben_r_ is now known as ben_r [14:09] PR snapd#2602 closed: overlord/snapstate: remove restrictions on ResetAliases === hikiko|ln is now known as hikiko === andyrock_ is now known as andyrock [14:16] PR snapd#2611 opened: interfaces: use fewer dot imports === ben_r_ is now known as ben_r [14:42] popey: can you share your nodejs snapcraft.yaml ? [14:42] I've been playing a lot with nodejs and haven't seen anything like the error on the rust one. [14:47] elopio: it's super basic. http://paste.ubuntu.com/23782053/ - output http://paste.ubuntu.com/23782054/ === ben_r_ is now known as ben_r [14:53] popey: agh, seems like exactly the same rust issue :( Can you please report a bug? I'll fix it for tomorrow's release. [15:01] is the snap store accepting classic confinement snaps, yet? [15:02] ryebot: I believe it should but I heard that people had some issues [15:02] zyga: ack, thanks [15:02] elopio: ok [15:04] PR snapd#2612 opened: cmd/snap, daemon, overlord/snapstate: tests and fixes for "snap refresh" of a classic snap [15:15] elopio: https://bugs.launchpad.net/snapcraft/+bug/1655678 [15:15] Bug #1655678: nodejs plugin fails to build - empty build dir [15:16] thank you. [15:22] PR snapd#2611 closed: interfaces: use fewer dot imports [15:23] PR snapd#2613 opened: interfaces: add new interface API [15:27] jdstrand: ^^ [15:27] jdstrand: take two on the idea [15:47] jdstrand: hey [15:49] jdstrand: debugging those i386 seccomp failures on xenial (for the old tests) [15:49] jdstrand: I get killed by lack of 201, looking at libseccomp source code that is geteuid32 [15:49] jdstrand: what was that tool that resolves syscalls to numbers? [15:50] jdstrand: I'm asking because this cannot be right, we allow that specific syscall in the common.sh profile [15:50] ah [15:50] found it [15:50] sorry :) [15:52] kyrofa, sergiusens: somehow the dependency handling for my snap is really different when using snapcraft to build and switching between classic and devmode confinement [15:53] kyrofa, sergiusens: my app depends on libboost which I pull in a build-depends via libboost-system-dev for example but with classic confinement libboost-system doesn't end up in prime [15:53] with devmode or strict it does [15:53] is there some fundamental difference other than that confinement is disabled with classic? [15:58] morphis, hmm... there is a difference, but I wouldn't expect it to cause that [15:59] morphis, it uses the linker of the core snap and uses rpaths to the libs within, but that shouldn't stop boost from getting in [15:59] kyrofa: so classic snaps are still running against the core snap or not? [16:00] and fixed it :) [16:00] morphis, you're on the verge of my knowledge here, but zyga will know more [16:01] zyga: any ideas? [16:01] I suspect that since snapcraft uses the rpaths, all those bind mounts don't happen [16:01] So while the core snap is not its execution context, it still runs reliably [16:01] But I'm making stuff up [16:03] kyrofa: possible, lets see what zyga says [16:03] morphis: classic and devmode are not just confinement, the snap is built entirely differently [16:04] morphis: all the code in your snap needs to be built from source [16:04] morphis: and it is built against what is in the core snap [16:04] zyga: means if I use boost, I have to build boost on my own? [16:04] morphis: all the executables are linked with stuff from /snap/core/current [16:04] morphis: yes [16:04] so I can't reuse any deb packages? [16:04] morphis: the dynamic linker is also used from the core snap [16:05] morphis: not that I know of [16:05] so if I have a app which depends on mesa, sdl, boost, protobuf, sdl, alsa, ... I have to build all of them manually? [16:05] yes [16:05] wow [16:06] that makes it a knockout thing [16:06] I don't think anyone tried it against gui apps [16:06] since core doesn't contain anything GUI-ish it is a difficult task I think [16:06] so what is the reason for this? [16:06] what is the reason for what? [16:07] that they need to be built? [16:07] because they are linked against the core snap [16:07] zyga: when I don't use classic confinement I use the linker from the core snap too, correct? [16:08] morphis: no [16:08] morphis: not directly [16:08] morphis: it depends on where you run [16:08] true [16:08] so on Ubuntu Core I do but on desktop I don't [16:08] morphis: if you want to understand the details I can discuss this with you but believe me when I say this is the only way it can work [16:08] morphis: you have to build it [16:08] zyga: I believe you, just trying to understand [16:08] morphis: ok [16:08] morphis: build this and ldd it [16:09] https://git.launchpad.net/checkbox-converged/tree/build-me [16:09] hmm [16:09] wrong link [16:09] https://github.com/zyga/hello-classic [16:09] morphis: ^^ [16:09] morphis: ldd this and look at what the elf file contains [16:10] morphis: this is the only way to make it work without controling the mount namespace [16:10] hm [16:10] I see [16:11] morphis: yeah, I bet it is going to be tough; I would love to see better support in snapcraft where one can lessen the cost of those builds [16:11] morphis: smarter caching [16:11] morphis: distcc/ccache [16:11] morphis: and working complex examples [16:11] zyga: its quite of useless if you have to rebuild everything and add it to your snapcraft.yaml [16:11] which makes it kind of complex for not tiny applications [16:12] morphis: well, not useless [16:12] morphis: just time consuming [16:12] morphis: sergio built a lot of nice things (git, vim) this way [16:12] morphis: and this is not a design choice, just technical challenge [16:12] morphis: this is a hard problem [16:13] morphis: you can cheat [16:13] yeah it is [16:13] morphis: you can build it to look for big libraries in /usr/lib [16:13] morphis: but that's not guaranteed to work [16:13] morphis: only /snap/core is guaranteed [16:20] morphis: btw, I could use a review on the new interface API if you have some time [16:20] zyga: sure! [16:20] zyga: but not today [16:33] PR snapd#2587 closed: interfaces/mount: add snippet types [16:34] popey: could you try with snapcraft from master? [16:45] elopio: uhm, where's that? [16:46]