=== lukedashjr is now known as luke-jr [04:59] morning [05:16] Good morning [05:16] zyga: hey [05:19] hey, how are you doing [05:19] it's gotten pretty cold :/ [05:20] zyga: yeah, it could rain a bit though, so far it's been rather dry [05:21] hmm somehow arch package of 2.39 is 4M smaller than 2.38 :/ [05:22] mborzecki: it was raining almost daily here [05:22] mborzecki: perhaps something changed from a copy to a link? [05:22] mborzecki: worth du-ing though [05:24] zyga: s-u-n is half the size [05:24] that is unusual! [05:24] perhaps we stopped importing something big [05:24] but half the size is very surprising [05:27] 12M -rwxr-xr-x 1 maciek maciek 12M 05-06 07:27 sun-2.38 [05:27] 6.1M -rwxr-xr-x 1 maciek maciek 6.1M 05-06 07:26 sun-2.39 [05:27] hm [05:31] wow, how the hell did image end up in s-u-n [05:32] zyga: image as in this one https://godoc.org/image [05:32] LOL [05:32] golang linking magic [05:32] how are you inspecting the binary? [05:34] zyga: looking at readelf output, i'm not aware of any other go friendly way [05:34] zyga: package is included in symbol names [05:34] I spent all of previous-work-day in readelf/strace/ltrace [05:34] we should add support for snap run --ltrace [05:34] yeah, ltrace would be nice [05:35] oh, and an up to date strace-static snap :P [05:35] yeah [05:35] maybe something to try to get to this week though I haven't set my priorities yet [05:36] man, resuming a virtual machine from hdd is super painful [05:36] (ram dumps) [05:40] zyga: http://paste.ubuntu.com/p/xypbbFqtd7/ quite a bit of packages gone in s-u-n 2.39 [05:40] mborzecki: perhaps go linking changed? [05:40] for real [05:41] I was joking before but maybe it's not a joke [05:41] zyga: same go version [05:41] perhaps something did import the image package before but not anymore? [05:41] dunno, it's a bit magic [05:42] zyga: hmm snapd in 2.38 also includes image :P [05:42] batteries and image processing included, that's the motto ;) [05:49] zyga: asserts pulls in golang.org/x/crypto/openpgp/packet which in turn imports image [05:50] what... the... [05:50] lol [05:50] so how did we end up dropping that from sun? [05:51] zyga: we dropped asserts, which probably was a dep of some other package [05:52] good find, [05:52] if you can automate enough it would be nice to have a tool that dumps, roughly, the imported go packages of a given binary [05:56] well, so probably nothing we can do about that import [05:56] at least i can proceed with the update now that we know where the 4MB went [06:02] mborzecki: indeed [06:23] hello mvo [06:23] hey zyga [06:23] zyga: good morning! [06:42] mvo: hey [06:43] ugh [06:43] I think I finally did it [06:43] and yay, I can sign into snapcraft forum again [06:44] hey mborzecki ! [06:44] mvo: did you know we can do image processing in snapd? :D [06:47] need to run a quick errand, back in 30 [06:52] mborzecki: image processing? [06:57] snapd 2.39 is currently held up due to https://pagure.io/fedora-infrastructure/issue/7762 [07:00] mvo: snapd was linking to the go imaging package [07:00] mvo: because reasons [07:00] mvo: mborzecki went investigating why snapd suddenly is 4MB smaller [07:00] that's why === pstolowski|afk is now known as pstolowski [07:01] mornings! [07:02] zyga: woah, nice job mborzecki [07:02] hey pstolowski [07:03] mvo: on similar spirit, did you know that opengl on the linux desktop requires libtinfo for terminal control sequences? [07:07] mvo: morning [07:09] hey pedronis [07:09] zyga: I had no idea :) [07:23] brb [07:35] re [07:41] re [07:41] pstolowski: pedronis: hey [07:41] o/ [07:44] mborzecki: pstolowski: zyga: hi [07:44] mvo: I made a couple more comments on your remodel PRs [07:45] pedronis: thanks, working on this now [07:45] mvo: also (still WIP) but this is were my own efforts or re-reg etc are heading: https://github.com/pedronis/snappy/commit/bd6908d1c5a87f7ceb1c51bf9b4b4bedb3f7a0af#diff-57a9337573ecc46801d40ff4e140872cR32 probably worth a quick look to get a sense of things [07:45] pedronis: ok [07:46] Eighth_Doctor: that's somewhat unexpected === lukedashjr is now known as luke-jr [07:46] Eighth_Doctor: but it's epel only? [07:46] yep [07:48] Eighth_Doctor: maybe some update didn't land for ppc64le just yet [07:49] probably [07:55] Eighth_Doctor: did you have any issues with snapd selinux transition in 2.39? (aside from the reboot to patch the mount units) [07:55] 🤷 [07:56] on the clean machines I have, no [07:56] but that doesn't tell me anything [08:11] mborzecki: https://fedoramagazine.org/use-udica-to-build-selinux-policy-for-containers/ [08:12] Eighth_Doctor: interesting, didn't we look at this a year ago or so? [08:12] we looked at CIL, I think [08:12] I don't remember if we looked at udica [08:16] or maybe i was browsing wrabcak's projects :P [08:17] mvo: tried to answer in 6775 [08:17] let me know if it's still unclear [08:20] mborzecki: heh [08:20] mvo: I'm mostly trying us to have to refactor the refactor later, but don't want to make a time sink for you either [08:20] s/to have/to not have/ [08:20] wow, I really haven't gotten any sleep at all [08:20] this is terrible [08:24] zyga: void is 111 instead of 000, i guess i need to update it in the fs otherwise things will blow up? [08:25] the package should update this... [08:31] mborzecki: the selinux-policy issue should be fixed in the next half hour or so [08:31] so I'll try building snapd then and go from there [08:32] mborzecki: pstolowski: I did some first pass review on some of your PRs [08:32] mborzecki: pstolowski: also I have a long chain of refactory PRs (related to remodeling and rereg) that need review [08:32] pedronis: yep, thank you, i'm working on the ensure timings stuff [08:33] pedronis: thx [08:33] pedronis: 6821 is the first one in your batch? [08:33] mborzecki: no it actually starts with 6810 [08:34] * pedronis has 6 open PRs and more coming [08:34] mvo should review those, but they will need 2nd reviews === Girt|Vacation is now known as Girtablulu [09:15] PR snapd#6824 closed: release: 2.39 [09:22] mborzecki: yes [09:23] mborzecki: the PKGBUILD change was insufficient? [09:23] zyga: pacman only issues a warning that the local permission bits are different [09:41] * dot-tobias says hi [09:41] mvo: left some comments under 6825, played with this locally and i think handling this from Go will be super tricky [09:43] mvo: offtopic-ish: I'd love to see snap run ltrace [09:44] perhaps doing per-tool hacks could be changed to a some sort of special hook (which could be a shell script) that allows us to add tools at ease? [09:54] mborzecki: thanks, in a meeting right now - does that mean the approach is doomed? [09:56] mvo: i don't think it's doomed, we just need to figure out a way how to make gdb, inferion and snap run not fight for the terminal [09:59] * mvo nods [10:51] mborzecki, mvo: https://forum.snapcraft.io/t/gpu-support-proposal/11247 [10:53] Brb [10:56] * pstolowski lunch [10:57] pedronis_: answered some of your questions under https://github.com/snapcore/snapd/pull/6750 maybe we could have chat with mvo after the standup [10:57] PR #6750: overlord/devicestate: update-gadget task handler with stubbed gadget callbacks [11:12] PR snapd#6830 opened: interfaces/dbus: fix unit tests when default snap mount dir is not /snap [11:13] super simple PR ^^ [11:20] PR snapd#6765 closed: tests: add security-seccomp to verify seccomp with arg filtering [11:30] mborzecki: yes, we should have a chat with mvo [11:31] mborzecki: this was fun btw: https://github.com/snapcore/snapd/pull/6828/commits/88b5794137a15266c7751ef865272402b26a8dd8 :) [11:31] PR #6828: many: use a fake assertion model in the device contexts for tests [11:34] pedronis: nice [11:42] pedronis: mvo: after standup then? [11:43] mborzecki: could work if it's not too long, we have a meeting after [11:43] pedronis: ok, otherwise tomorror morning maybe? [11:43] yes [11:45] ack [12:02] off to pick up the kids [12:04] mborzecki: thanks for keeping an eye on that PR === jdstrand_ is now known as jdstrand === ricab is now known as ricab|lunch [12:51] PR snapd#6830 closed: interfaces/dbus: fix unit tests when default snap mount dir is not /snap [12:52] pstolowski: thanks for looking at my PRs [12:57] np [13:52] drat, the forum does not support SVG images [13:59] :( [14:07] zyga: github does (kind of), maybe if you link to it? [14:07] I will link to a repo with the .dot files, for now the image is enough [14:07] mborzecki: fyi https://forum.snapcraft.io/t/hello-world-cuda-analysis/11250 [14:08] mborzecki: isn't that image nice? :) [14:08] zyga: that's from strace logs? [14:08] mborzecki: yep [14:08] mborzecki: + some tooling [14:09] mborzecki: + love :) [14:09] zyga: nice! [14:09] mborzecki: I have much more, just finally nailed how I want the FS access to look like === ricab|lunch is now known as ricab [14:50] * zyga updated https://forum.snapcraft.io/t/hello-world-cuda-analysis/11250 and goes for lunch [15:05] hey ijohnson [15:05] hey zyga [15:06] ijohnson: I've added https://forum.snapcraft.io/t/hello-world-cuda-analysis/11250 and I will soon make a post like that about openGL (which will be much much longer as there is far more data and complexity there) [15:07] zyga: nice, I saw your post there! BTW, not sure if you're aware but I also created a snap with cuda samples from the SDK that bundles libcuda inside the snap, not sure if that would be useful to you or not [15:08] ijohnson: I think I saw that a while ago, the snap itself is interesting but I followed a simple cuda tutorial on the nvidia website, I mainly itererated on the tooling and analysis of what is going on at runtime so that I can form some kind of vision about how to support GPUs in snapd better [15:09] ijohnson: I have 4 GPUs on my desk and I've been experimenting with various drivers for a few days [15:11] zyga: I may be mistaken, but using CUDA with multi-GPU setups may use different accesses (not libraries but like /dev) - if you're able it might be worth trying to setup your machine with multiple nvidia GPUs that are compatible (I think as long as your GPUs aren't more than 5 years old they should work) with multi-GPU CUDA and then you can try to run the multi-GPU samples from my snap [15:12] ijohnson: there are more /dev/nvidia{1,2,3} entries for sure, I only have one quadro card but your remark is spot on. I will look for a few more cards to buy before Lyon [15:13] I have a discrete GPU in my desktop and another discrete in my laptop but unfortunately haven't been able to get ahold of 2 discrete cards I can plug into my desktop [15:13] ijohnson: the magic of olx.pl (like craigslist, I guess) and my mobo-on-desk setup :) [15:16] nice :) [15:33] * cachio lunch === msalvatore_ is now known as msalvatore [15:41] Hi guys, what are usually the times needed for a classic confinement approval request? === pstolowski is now known as pstolowski|afk [16:36] I'll EOD now [16:36] albertosottile: hey, it usually takes a few days, sometimes longer when holidays or a longer backlog intervene [16:37] zyga: thanks, we have been waiting for 5 days but there was a weekend in between, so it might be that [16:44] PR snapcraft#2555 opened: extensions: block direct use of private extensions [16:46] jdstrand: FYI https://forum.snapcraft.io/t/hello-world-cuda-analysis/11250 [17:27] zyga: nice writeup [17:57] Can someone help me creating a snap for a wine application? [17:58] I used this https://github.com/mmtrt/notepad3 as a basis, but i'm stuck at the copying of the file, it says can't find the file (in the original script it uses wget) [17:58] should I start over from scratch? [18:00] PR snapd#6810 closed: many: do without device state/assertions accessors based on state only outside of devicestate/tests [18:04] I rebased on #6817 on master, it's ready for reviews (it's not too big) [18:04] PR #6817: overlord,overlord/devicestate: do without GadgetInfo/KernelInfo in devicestate [18:57] PR pc-amd64-gadget#10 closed: Add mmx64.efi (MokManager) to support mokutil [18:57] PR pc-amd64-gadget#11 closed: Add mmx64.efi (MokManager) to support mokutil [18:58] PR pc-amd64-gadget#10 opened: Add mmx64.efi (MokManager) to support mokutil [18:58] PR pc-amd64-gadget#11 opened: Add mmx64.efi (MokManager) to support mokutil === joedborg_ is now known as joedborg === jamespage_ is now known as jamespage === stoopkid_ is now known as stoopkid [21:37] PR snapd#6831 opened: tests: retry govendor sync to minimize the number of connection errors on prepare [22:45] PR snapcraft#2556 opened: cli: snapcraft promote [23:00] PR snapcraft#2557 opened: ci: remove dependency on LXD from travis tests === msalvatore_ is now known as msalvatore