| mborzecki | morning | 06:51 |
|---|---|---|
| mup | PR snapd#8035 closed: data/selinux: workaround incorrect fonts cache labeling on RHEL7 (2.43) <Created by bboozzoo> <Merged by bboozzoo> <https://github.com/snapcore/snapd/pull/8035> | 06:51 |
| mup | PR snapd#8014 closed: tests: run `uc20-snap-recovery-encrypt` test on 20.04-64 as well <Simple 😃> <Created by mvo5> <Merged by mvo5> <https://github.com/snapcore/snapd/pull/8014> | 07:02 |
| jamesh | looks like the nakedret static check is failing CI runs that were passing previously | 07:15 |
| jamesh | seems the tool fixed a bug 5 hours ago and showed up a bunch more errors: https://github.com/alexkohler/nakedret/pull/10 | 07:16 |
| mup | PR alexkohler/nakedret#10: Fix bug with not detected nested returns <Created by programmer04> <Merged by alexkohler> <https://github.com/alexkohler/nakedret/pull/10> | 07:16 |
| mup | Bug #5: Plone Placeless Translation Service metadata missing from po files <feature> <iso-testing> <lp-translations> <Launchpad itself:Fix Released by daf> <https://launchpad.net/bugs/5> | 07:16 |
| jamesh | it seems the tool previously only detected naked returns at the top level of the function, ignoring any embedded in blocks | 07:24 |
| jamesh | ah. Looks like the new nakedret version is throwing up false positives | 07:32 |
| zyga | o/ | 07:37 |
| jamesh | it doesn't understand nested functions | 07:37 |
| mborzecki | jamesh: hmmm ehh, looks like a bug in nakedret | 07:40 |
| mborzecki | maybe something i can fix | 07:40 |
| jamesh | mborzecki: yeah. I'm filing a bug report now | 07:41 |
| zyga | thank you guys | 07:41 |
| jamesh | mborzecki: here's the bug report with a trivial example reproducing it: https://github.com/alexkohler/nakedret/issues/11 | 07:44 |
| mborzecki | jamesh: thanks for the report! | 07:45 |
| jamesh | mborzecki: I've got some idea of how to fix the script. Are you already working on it, or should I give it a crack? | 07:52 |
| mborzecki | jamesh: i'm looking into nakedret, feel free to fix the script | 07:53 |
| jamesh | mborzecki: I meant nakedret :-) | 07:54 |
| jamesh | should have said tool rather than script | 07:54 |
| mborzecki | jamesh: ah, ok, then go ahead :) don't want to spoil the fun | 07:54 |
| jamesh | okay | 07:55 |
| mborzecki | jamesh: heh, never played with go ast, i guess since ast.Inspect is depth first, i gets to the return in the nested func() block first without noticing it's a separate nested func, is that your thinking as well? | 08:02 |
| jamesh | mborzecki: I was thinking of just doing the whole job in Visit, substituting a new Visitor whenever we encounter a new function | 08:02 |
| pstolowski | morning | 08:04 |
| mborzecki | pstolowski: hey | 08:06 |
| mborzecki | jamesh: hmm looking more, i have a feeling nakedret doesn't actually handle naked return in function literals | 08:16 |
| zyga | hmm, when rebooting my ubuntu system "a stop job is running for Service for snap application lxd.daemon" | 08:18 |
| zyga | and just sits there until 10 minutes elapse | 08:18 |
| jamesh | mborzecki: yeah. that'd be an ast.FuncType with no ast.FuncDecl, I think | 08:20 |
| jamesh | no. That's not quite right | 08:21 |
| mborzecki | jamesh: FuncLit | 08:22 |
| zyga | brb | 08:22 |
| jamesh | mborzecki: thanks | 08:22 |
| mborzecki | jamesh: so this bit skips function literals https://paste.ubuntu.com/p/ZTYjQkDHbt/ | 08:25 |
| jamesh | mborzecki: I was thinking of adding literals rather than skipping them | 08:25 |
| mborzecki | jamesh: yeah, i suppose returnVisitor could be smarter and just accept FuncDecl and FuncLit, while the nested ast.Inspect could skip FuncLit | 08:26 |
| jamesh | mborzecki: here's what I've got so far: https://paste.ubuntu.com/p/RQBnqKtjDP/ | 08:29 |
| __chip__ | 👋 | 08:42 |
| mborzecki | jamesh: got this https://paste.ubuntu.com/p/ThN9FnbpyH/ | 08:45 |
| mborzecki | __chip__: hey | 08:47 |
| __chip__ | network seems laggy | 08:47 |
| __chip__ | mborzecki: hiya | 08:48 |
| jamesh | mborzecki: here's my PR: https://github.com/alexkohler/nakedret/pull/12 | 08:48 |
| mup | PR alexkohler/nakedret#12: Handle function literals and nested functions <Created by jhenstridge> <https://github.com/alexkohler/nakedret/pull/12> | 08:48 |
| __chip__ | new nakedret, need to push a pr to address new failures | 08:48 |
| mborzecki | jamesh: cool, looking | 08:48 |
| __chip__ | unless somebody else already has? | 08:48 |
| mborzecki | __chip__: false positives, jamesh is trying to fix it upstream | 08:48 |
| jamesh | mborzecki: I might steal your naming for literals | 08:49 |
| mborzecki | jamesh: go ahead, was about to add a comment there :) | 08:49 |
| __chip__ | oh! | 08:49 |
| jamesh | there's two real problems in testutil/containschecker.go, lines 89 and 129 | 08:49 |
| __chip__ | jamesh: glad you're on it | 08:50 |
| mborzecki | tbh, i like how it's easy to work with go ast, i mean, both of us could address the problem quite easily | 08:50 |
| __chip__ | i'll go to the snaps performance session then | 08:50 |
| * __chip__ agrees with mborzecki | 08:50 | |
| __chip__ | ISTR even the python ast wasn't as nice | 08:51 |
| __chip__ | but i've grown since i tried to use that so maybe not | 08:51 |
| jamesh | __chip__: the old nakedret only detected nakedret only detected problems at the top level of a function, ignoring returns in blocks | 08:51 |
| jamesh | the new nakedret is too aggressive and misattributes return statements in embedded function literals | 08:52 |
| __chip__ | it was a very simplistic first pass, yes | 08:52 |
| __chip__ | seems they overshot in the new year :) | 08:52 |
| * __chip__ out | 08:53 | |
| mborzecki | in the meantine, we need to unblock master :P | 08:54 |
| abeato | zyga, hey, how is snapd development these days? I bind mount my binary to /usr/lib/snapd/snapd but I get this error: | 09:00 |
| abeato | cannot run daemon: state startup errors: [cannot obtain snap-seccomp version information: invalid format of version-info: "051a668fa5c47aa16280298ed72cc38c8ac3dc40 2.4.1 8c73f36d3de1f71977107bf6687514f16787f639058b4db4c67b28dfdb2fd3af"] | 09:00 |
| mborzecki | jamesh: i'll fix snapd master | 09:01 |
| abeato | s/how is/how is it done/ | 09:01 |
| jamesh | mborzecki: okay. Disable the check, and fix testutil/containschecker.go ? | 09:01 |
| mborzecki | jamesh: yes | 09:01 |
| pstolowski | abeato: you also need snap-seccomp binary rebuilt from same tree; if running snapd directly from builddir (instead of bind-mounting) you can copy snap-seccomp into cmd/snapd dir | 09:03 |
| abeato | pstolowski, ok, so I can rebuild snap-seccomp and bind mount to /usr/lib/snapd/snap-confine? | 09:04 |
| abeato | -seccomp I meant | 09:05 |
| zyga | Snap-seccomp | 09:05 |
| zyga | Yeah | 09:05 |
| zyga | But | 09:05 |
| zyga | I rarely do that | 09:05 |
| zyga | Just build and run from tree | 09:05 |
| zyga | Make sure snap-seccomp and snapd are siblings (symlink is ok) | 09:06 |
| pstolowski | yeah that's what i always do too; i've never bind-mounted it | 09:06 |
| abeato | I was building in another machine - but yeah, I can build there too | 09:06 |
| abeato | thanks | 09:06 |
| mup | PR snapd#8037 opened: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 09:07 |
| mup | Issue core18#145 opened: pollinate missing from core18 snap <Created by rcj4747> <https://github.com/snapcore/core18/issue/145> | 09:07 |
| mborzecki | zyga: pstolowski: ^^ trivial PR | 09:08 |
| mup | PR core18#146 opened: Add pollinate to the extra packages #144 <Created by rcj4747> <https://github.com/snapcore/core18/pull/146> | 09:11 |
| mborzecki | zyga: https://github.com/snapcore/snapd/pull/8037#pullrequestreview-347158408 imo the problem is that it's not immediately clear what could be the value returned by given piece of code | 09:36 |
| mup | PR #8037: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 09:36 |
| mborzecki | zyga: you may still need to read back and analyze, but it's even less clear when there just 'return' | 09:37 |
| mborzecki | zyga: tbh i find named return variables to have only documentation value, and maybe if you need to do some trickery in deferred code but that requires really good reason to do so | 09:38 |
| jamesh | zyga: there's also a proposal to remove them in Go 2: https://github.com/golang/go/issues/21291 | 09:39 |
| zyga | hmm | 09:41 |
| zyga | I guess it depends on context | 09:41 |
| jamesh | (that doesn't mean they will actually be removed though: just that some people think they should) | 09:41 |
| zyga | I'm feeling sick today, sorry for being slow | 09:43 |
| zyga | I'll warm up the office grab some meds and be back | 09:45 |
| sergiusens | pstolowski: is the configure hook triggered with a "snap start|stop" and is it possible to know the hook was triggered due to this event? | 09:58 |
| pstolowski | sergiusens: no, it's not triggered, start/stop simply wrap systemctl | 10:02 |
| sergiusens | pstolowski: ah, thanks | 10:03 |
| pstolowski | sergiusens: nb hooks are always run under a snapd task, so you would see a task under a change in snap changes / snap change <id> | 10:05 |
| zyga | back, feeling slightly better now | 10:14 |
| zyga | hmm | 10:36 |
| zyga | oddly enough I'm making progress | 10:42 |
| zyga | fresh ideas, even when I'd rather sleep all day | 10:42 |
| zyga | mborzecki: btw, I made the dprintf change | 10:43 |
| mborzecki | zyga: cool, did you push it already? | 10:43 |
| zyga | yesterday | 10:43 |
| mborzecki | hmm https://github.com/snapcore/snapd/pull/8037 SKIP_NAKEDRET isn't picked up from environment? | 10:44 |
| mup | PR #8037: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 10:44 |
| zyga | it's green :) | 10:44 |
| zyga | mborzecki: ... weird? | 10:45 |
| zyga | is the test correct? why exit 1 ? | 10:45 |
| mborzecki | heh, the unit test is run as a different user | 10:46 |
| zyga | ahhh | 10:46 |
| mborzecki | ok, force pushed a fix, should be good now | 10:56 |
| zyga | ... ... ... | 11:07 |
| zyga | iterating on core tests sucks | 11:07 |
| zyga | like badly | 11:07 |
| zyga | I wonder what we could do to make that faster | 11:07 |
| zyga | could we boot core, install lxd, pull the sources, build, install the snaps and carry on without rebooting and wiping the disk? | 11:08 |
| mborzecki | zyga: we need to build a custom core/snapd snaps and build a whole image with that | 11:14 |
| mborzecki | zyga: maybe we should try to take some time to improve that, but local testing sucks too | 11:14 |
| zyga | not sure if we need to build the whole image with that | 11:14 |
| zyga | it depends on the test | 11:14 |
| zyga | we could have a suite that does that | 11:14 |
| zyga | but I could iterate on pretty much all my core work without that property | 11:14 |
| zyga | but I see your point | 11:15 |
| mborzecki | although local testing is much fater, because my deskop is way faster than the gcp vms we use | 11:15 |
| mborzecki | s/fater/faster/ | 11:15 |
| mborzecki | that, unless you need to work with a device :/ then it sucks again | 11:15 |
| mborzecki | quick errand at the tax office, back in 1h or so | 11:16 |
| zyga | holly shit, fixed it :D | 11:16 |
| zyga | mborzecki: my local testing is slower | 11:16 |
| zyga | mborzecki: but it's always network bound | 11:16 |
| zyga | mborzecki: I hacked (over two years ago) totally offline spread test loop | 11:16 |
| zyga | and it was insanely fast | 11:16 |
| zyga | with many-core qemu | 11:16 |
| zyga | pre-cached snaps inside VM | 11:16 |
| zyga | apt-cacher-ng for all apt stuff | 11:16 |
| zyga | it was super nice | 11:17 |
| zyga | I even hacked it so that builds were incremental | 11:17 |
| zyga | (as long as you --reuse) | 11:17 |
| zyga | but we don't have that property | 11:17 |
| * zyga runs one clean loop to see | 11:19 | |
| zyga | mborzecki: that's one fixed and one or two to go (forgot exact number) | 11:19 |
| zyga | but progress :) | 11:19 |
| zyga | whee | 11:23 |
| zyga | so happy | 11:23 |
| zyga | this is real :) | 11:23 |
| zyga | PR coming up in 15-20 minutes | 11:23 |
| zyga | mborzecki, pstolowski: please review | 11:36 |
| zyga | https://github.com/snapcore/snapd/pull/8038 | 11:36 |
| mup | PR snapd#8038 opened: tests: fix gadget-update-pc test leaking snaps <Test Robustness> <Created by zyga> <https://github.com/snapcore/snapd/pull/8038> | 11:36 |
| mup | PR #8038: tests: fix gadget-update-pc test leaking snaps <Test Robustness> <Created by zyga> <https://github.com/snapcore/snapd/pull/8038> | 11:36 |
| * zyga high-fives himself, onto the next failure! | 11:36 | |
| pstolowski | zyga: looking | 11:37 |
| roadmr | zyga: I call that progress | 11:37 |
| zyga | roadmr: I stared on this early last year | 11:38 |
| zyga | roadmr: and just never managed to come up with something that would work | 11:38 |
| zyga | this is so simple now | 11:38 |
| zyga | eh | 11:45 |
| zyga | nacked returns haunt me | 11:45 |
| zyga | and the fix for that failed | 11:45 |
| zyga | did they fail because of hooks snap | 11:45 |
| zyga | are we in a catch 22? | 11:45 |
| zyga | + [[ -n 1 ]] | 11:46 |
| zyga | /bin/bash: line 89: skip: unbound variable | 11:46 |
| zyga | mborzecki: ^ | 11:46 |
| zyga | what? | 11:46 |
| zyga | oh | 11:47 |
| zyga | I see it | 11:47 |
| zyga | btw, why did you say if [[ -n ... ]] | 11:47 |
| zyga | rather than if [ -n ... ] ? | 11:47 |
| zyga | mborzecki: ^ | 11:47 |
| zyga | mborzecki: https://github.com/snapcore/snapd/pull/8037/files#r370072706 | 11:47 |
| zyga | please apply | 11:47 |
| mup | PR #8037: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 11:48 |
| jamesh | zyga: btw, I refactored https://github.com/snapcore/snapd/pull/7588 a bit to move the cgroup based pid->snap helper to sandbox/cgroup. I was hoping to have a green CI tick by now to ask for more review, but got stuck on the nakedret problem | 11:55 |
| mup | PR #7588: cmd/snap: add a "snap routine portal-info" command <Created by jhenstridge> <https://github.com/snapcore/snapd/pull/7588> | 11:55 |
| zyga | jamesh: I'm sorry, I'll have a look at that | 11:55 |
| zyga | jamesh: looks great, once either branch lands I'll adjust the other | 11:56 |
| zyga | pstolowski: like this? https://github.com/snapcore/snapd/pull/8038/files#r370078320 | 12:01 |
| mup | PR #8038: tests: fix gadget-update-pc test leaking snaps <Test Robustness> <Created by zyga> <https://github.com/snapcore/snapd/pull/8038> | 12:01 |
| pstolowski | zyga: yep, it's even better than my suggestion, ty | 12:02 |
| zyga | cool, I'll integrate that back when the master blockers are fixed | 12:02 |
| mborzecki | re | 12:06 |
| zyga | welcome back, please check the comment above | 12:06 |
| zyga | mborzecki: https://github.com/snapcore/snapd/pull/8037 | 12:06 |
| zyga | this should unbreak your PR | 12:07 |
| mup | PR #8037: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 12:07 |
| zyga | which will unbreak master and many other PRs | 12:07 |
| mborzecki | huh, it failed? | 12:07 |
| zyga | mborzecki: shellcheck | 12:08 |
| zyga | mborzecki: $skip is unset | 12:08 |
| mborzecki | zyga: shellcheck does not complain | 12:08 |
| zyga | hmm? didn't it in the log | 12:08 |
| zyga | ah | 12:09 |
| zyga | mborzecki: /bin/bash: line 89: skip: unbound variable | 12:09 |
| zyga | well, anyway | 12:10 |
| zyga | not shellcheck | 12:10 |
| zyga | (which is weird) | 12:10 |
| zyga | but still | 12:10 |
| mborzecki | weird, don't remember spread tasks runnig with set -u | 12:10 |
| zyga | fixing next test, ubuntu-core-refresh | 12:18 |
| zyga | should be simple | 12:18 |
| zyga | wow, I feel like this week will end without leaks | 12:18 |
| zyga | and re-enabled mount-ns tests | 12:18 |
| mborzecki | haha nakedret fix landed, before we managed to land 8037 | 12:23 |
| zyga | shall we skip it then | 12:24 |
| zyga | and just re-trigger? | 12:24 |
| mup | PR snapcraft#2888 closed: elf: read ELF type when extracting attributes <Created by cjp256> <Merged by sergiusens> <https://github.com/snapcore/snapcraft/pull/2888> | 12:32 |
| * roadmr keeps reading that as nakedregret | 12:32 | |
| zyga | to me it just reads as nekker | 12:32 |
| mborzecki | zyga: too much witcher :P | 12:35 |
| zyga | "fuck" ;-) | 12:36 |
| mborzecki | hahah | 12:36 |
| mborzecki | zyga: i think i'll push an update to task.yaml, but will leave the knobs in run-checks | 12:36 |
| zyga | sounds reasonable, | 12:37 |
| jamesh | there were real nakedret failures in master, so 8037 needs to be landed either way | 12:44 |
| roadmr | nakedrat!! https://en.wikipedia.org/wiki/Naked_mole-rat | 12:53 |
| zyga | ok, one more test fixed | 12:54 |
| zyga | PR a soon as I confirm | 12:54 |
| mborzecki | this is nice https://rakyll.org/inlined-defers/ | 12:59 |
| === ricab is now known as ricab|lunch | ||
| zyga | brb | 13:17 |
| zyga | coffee | 13:17 |
| pstolowski | @mborzecki: oh, interesting, "This overhead is why Go developers started to avoid defers in certain cases to improve performance" | 13:20 |
| pstolowski | is nakedret SFW? ;) | 13:20 |
| mborzecki | pstolowski: yeah, must have skipepd the memo on that or sth | 13:20 |
| mvo | pstolowski: hey, quick question - what's the state of the "snap refresh" of a disabled service will re-enable this service | 13:31 |
| pstolowski | mvo: hi! it was fixed by ijohnson some time ago | 13:32 |
| mvo | pstolowski: do you remember if the fix is in 2.43 or even earlier? | 13:32 |
| pstolowski | mvo: no, let me check | 13:33 |
| zyga | re | 13:33 |
| zyga | hey mvo | 13:33 |
| zyga | another test fixed :) | 13:33 |
| mvo | pstolowski: it's fine, got asked this question here during the sprint | 13:33 |
| mvo | zyga: hey! nice to see you | 13:34 |
| pstolowski | mvo: appears to be fixed in 2.43, commit 98c1da1c2b75883585e606d935dfff694e257753 on 2019-11-19 | 13:34 |
| mvo | pstolowski: thank you! | 13:35 |
| mup | PR snapd#8039 opened: tests: remove revision leaking from ubuntu-core-refresh <Test Robustness> <Created by zyga> <https://github.com/snapcore/snapd/pull/8039> | 13:37 |
| ijohnson | hi mvo! Yes the fix is in 2.43 | 13:41 |
| mup | PR snapd#8040 opened: gadget: skip update when raw structure content is unchanged <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8040> | 13:44 |
| pstolowski | mvo, ijohnson it was https://bugs.launchpad.net/snapd/+bug/1827237 | 13:45 |
| mup | Bug #1827237: disabled services are re-enabled after revert <snapd:Fix Committed by anonymouse67> <https://launchpad.net/bugs/1827237> | 13:45 |
| zyga | mborzecki: +1 | 13:45 |
| zyga | mborzecki: question about this feature | 13:45 |
| mborzecki | zyga: hm? | 13:46 |
| zyga | mborzecki: is there anything one could do to fix a device that has broken raw partition content | 13:46 |
| zyga | mborzecki: e.g. interactively working on a device | 13:46 |
| zyga | dd gone souht | 13:46 |
| zyga | south8 | 13:46 |
| zyga | south* | 13:46 |
| zyga | device still booted so you can fix it | 13:46 |
| zyga | is there any command that would "ensure" that partitions are as expected | 13:46 |
| zyga | and that the content is as expected? | 13:46 |
| mborzecki | zyga: we have no such things yet | 13:46 |
| zyga | ok | 13:47 |
| mborzecki | zyga: and there's probably scenarios where gadget.yaml would be insufficiently expressive | 13:47 |
| mborzecki | zyga: idk if there's a plan to have a like a factory reset perhaps | 13:48 |
| zyga | I think it would have to be a debug command | 13:48 |
| zyga | "debug ensure-partitions" or something | 13:48 |
| mborzecki | zyga: i had a debug command that would dump the expected layout, but nothing that actually compares and lists differences if any | 13:53 |
| mborzecki | hhm perhaps a dry-run gadget update would be useful | 13:54 |
| zyga | hmm | 13:55 |
| zyga | indeed | 13:55 |
| zyga | even as a "here's what would happen so that you can make a more informed decision" | 13:55 |
| roadmr | does anyone know of any snaps that bundle a mysql server as required by the snapped software? is there something I can reuse for that? | 14:14 |
| jamesh | nextcloud maybe? | 14:15 |
| zyga | yes, nextcould for sure | 14:19 |
| * zyga needs to check up on kids, fix for the last test seems very easy, just needs about 2-3 tries to ensure | 14:20 | |
| zyga | brb | 14:20 |
| roadmr | thanks jamesh zyga I'll look at nextcloud! | 14:21 |
| zyga | huh | 14:21 |
| zyga | my wife just returned home | 14:21 |
| zyga | and bought me two spools of duct tape | 14:21 |
| roadmr | 💪 | 14:21 |
| zyga | like... what? | 14:21 |
| roadmr | zyga: what's she trying to tell you 🤔 | 14:21 |
| zyga | :D | 14:21 |
| === ricab|lunch is now known as ricab | ||
| mborzecki | lp timeouts again | 14:37 |
| ijohnson | mborzecki: #8037 is green | 14:37 |
| mup | PR #8037: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 14:37 |
| mborzecki | and now we'll fill up the travis build queue :) | 14:38 |
| ijohnson | :-) | 14:38 |
| mup | PR snapd#8037 closed: travis, tests, run-checks: skip nakedret <Simple 😃> <Created by bboozzoo> <Merged by bboozzoo> <https://github.com/snapcore/snapd/pull/8037> | 14:38 |
| ijohnson | mborzecki: have you had a chance to review #8001 again? is there anything else you think I should add to that PR before we can merge it? | 14:39 |
| mup | PR #8001: boot: enable UC20 kernel extraction and bootState20 handling <UC20> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/8001> | 14:39 |
| mborzecki | ijohnson: let me see, i think it shoudl be +1 for me | 14:40 |
| mborzecki | ijohnson: we're still blocked on samuele/mvo, aren't we? | 14:40 |
| ijohnson | mborzecki: great, still waiting for pedronis and/or mvo on that but just wanted to get it as close as possible before they look | 14:40 |
| ijohnson | mborzecki: yes still blocked waiting for them | 14:40 |
| zyga | trying a fix for the last one | 14:45 |
| zyga | fingers crossed | 14:45 |
| mup | PR snapd#8041 opened: tests, run-checks, many: fix nakedret issues (2.43) <Simple 😃> <Created by bboozzoo> <https://github.com/snapcore/snapd/pull/8041> | 14:51 |
| zyga | 3rd reboot | 15:00 |
| zyga | fingers crossed | 15:00 |
| zyga | maybe it will be green now | 15:00 |
| zyga | success | 15:04 |
| zyga | wow | 15:04 |
| zyga | all three fixed in one day :) | 15:05 |
| zyga | mborzecki: is master green now? | 15:05 |
| mborzecki | zyga: should be | 15:05 |
| zyga | super | 15:05 |
| zyga | mborzecki: can you review the two test robustness PRs | 15:06 |
| zyga | I'm opening the last one | 15:06 |
| mup | PR snapd#8042 opened: tests: remove revision leaking from remodel-kernel <Test Robustness> <Created by zyga> <https://github.com/snapcore/snapd/pull/8042> | 15:08 |
| zyga | I rebased all three fixes | 15:09 |
| zyga | and will grab lunch now | 15:09 |
| zyga | afk | 15:10 |
| mborzecki | i'm going for a late lunch with the kids somewhere in the city, i'll check the PR state in the evening | 15:12 |
| zyga | pstolowski: could you review the 2nd of the fixes | 15:33 |
| zyga | https://github.com/snapcore/snapd/pull/8039/files | 15:33 |
| mup | PR #8039: tests: remove revision leaking from ubuntu-core-refresh <Test Robustness> <Created by zyga> <https://github.com/snapcore/snapd/pull/8039> | 15:33 |
| zyga | I will have half of reviews this way | 15:33 |
| zyga | I'll do a pass to add --purge to them as well | 15:34 |
| pstolowski | sure | 15:34 |
| zyga | I still have three red PRs but one will turn green soon | 15:34 |
| zyga | and that's only two remaining for tomorrow | 15:34 |
| zyga | I may get one more today, need to check how much work it is | 15:34 |
| pstolowski | zyga: nb purge only makes sense when the last revision gets dropped | 15:35 |
| zyga | ah | 15:35 |
| zyga | didn't think of that | 15:35 |
| pstolowski | zyga: i mean, you can pass it, just has no effect | 15:36 |
| pstolowski | *always pass it* | 15:36 |
| * zyga runs larger batch of tests and takes a break | 15:45 | |
| zyga | I'll come back to explore | 15:45 |
| zyga | jdstrand: hey | 15:45 |
| zyga | jdstrand: how does your queue look like? | 15:46 |
| zyga | jdstrand: do you think you could review the setgid PR this week? | 15:46 |
| zyga | (7980) | 15:46 |
| jdstrand | zyga: it is the queue for 2.44. I doubt it will be this week, but hope to have a big PR push next week | 15:49 |
| zyga | jdstrand: understood thanks! | 15:49 |
| zyga | I'm working on the backlog so no emergency but I'd love to get some progress on those eventually | 15:49 |
| zyga | jdstrand: I have enough things to work on this week | 15:49 |
| zyga | ijohnson: hey | 19:25 |
| zyga | ijohnson: just use syscalls package directly | 19:25 |
| ijohnson | hey zyga | 19:25 |
| zyga | we use loads of syscalls that are not in the golang stack | 19:25 |
| ijohnson | zyga: you mean like Syscall6 etc. ? | 19:25 |
| zyga | yeah | 19:25 |
| mup | PR snapd#8041 closed: tests, run-checks, many: fix nakedret issues (2.43) <Simple 😃> <Created by bboozzoo> <Merged by bboozzoo> <https://github.com/snapcore/snapd/pull/8041> | 19:25 |
| ijohnson | zyga: ah ok cool | 19:25 |
| ijohnson | zyga: thanks! | 19:25 |
| zyga | ijohnson: just make sure about the requirements of renameat2 | 19:26 |
| zyga | kernel requirements | 19:26 |
| ijohnson | zyga: it's from 2.6.16 | 19:26 |
| ijohnson | zyga: is that too new? | 19:26 |
| zyga | ah, that's ancient | 19:27 |
| zyga | I thought it's something 4+ | 19:27 |
| zyga | ijohnson: I'd love a review of the three fix PRs for the tests leaking mounted snaps | 19:29 |
| zyga | ijohnson: one is green, the other two are still in testing | 19:29 |
| ijohnson | zyga: yes it's in my queue, I'll try to get to them this afternoon | 19:29 |
| zyga | super, thanks | 19:30 |
| zyga | I'll rebase the mount leak detector on that tomorrow | 19:30 |
| zyga | and focus on the last two red PRs | 19:30 |
| zyga | :) | 19:30 |
| ijohnson | zyga: one last question for you since you're around | 19:32 |
| zyga | yeah? | 19:32 |
| ijohnson | zyga: it appears that the syscall number for renameat2 is different on different architectures | 19:32 |
| zyga | yes, that's normal | 19:32 |
| ijohnson | the usual go thing is to have a sys_linux_arm64.go ,etc. and define the constant in those files | 19:32 |
| zyga | each architecture has a separate table | 19:32 |
| ijohnson | is there a place we do that already? | 19:32 |
| ijohnson | I don't see anywhere we do that currently | 19:33 |
| zyga | ijohnson: ish, not for go | 19:33 |
| zyga | ijohnson: btw, I think you may be lucky | 19:34 |
| zyga | last time I looked at it | 19:34 |
| zyga | go uses renameat2 internally | 19:34 |
| zyga | just not exposes it | 19:34 |
| zyga | so _maybe_ there's a syscall const for SYS_RENAMEAT2 | 19:34 |
| zyga | but if there is none, yeah, you have to define those | 19:34 |
| zyga | that sucks | 19:34 |
| ijohnson | I just looked and it uses renameat to implement Rename | 19:34 |
| zyga | ah | 19:34 |
| zyga | boo | 19:34 |
| zyga | well :/ | 19:34 |
| zyga | that's that | 19:34 |
| ijohnson | I'll double check, but yeah boo | 19:35 |
| zyga | start with x86 | 19:35 |
| zyga | and x86_64 | 19:35 |
| ijohnson | yeah I was just gonna do all the architectures that the core snap is available on currently, so arm64, armhf, i386, amd64, s390x, and ppcel64 | 19:36 |
| ijohnson | though I don't know if we support powerpc at all | 19:36 |
| zyga | ijohnson: tip: seccomp knows | 19:36 |
| ijohnson | ah yes that's a good idea | 19:37 |
| zyga | the magic command is scmp_sys_resolver (obviously :) | 19:37 |
| zyga | (it's a terrible name) | 19:37 |
| ijohnson | yes I've used that a few times before | 19:37 |
| zyga | cool | 19:37 |
| * zyga pulled fixed fixes and restarted local tests of all of core | 19:40 | |
| zyga | o/ | 19:40 |
| mup | PR snapd#8043 opened: osutil: add Renameat2() <UC20> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/8043> | 20:32 |
| mup | PR snapd#8044 opened: grub: support atomically renaming kernel symlinks <UC20> <Created by anonymouse64> <https://github.com/snapcore/snapd/pull/8044> | 20:43 |
| ijohnson | hey jdstrand, since you last touched the spread pulseaudio tests, any quick takes on what the problem might be here: https://pastebin.ubuntu.com/p/4Kqpj2CnPb/ ? | 20:49 |
| jdstrand | ijohnson: not otoh. based on the top of the past, seems a pulseaudio is already running. perhaps something from a previous test? | 20:59 |
| ijohnson | jdstrand: hmm could be | 20:59 |
| zyga | ijohnson: https://github.com/snapcore/snapd/pull/8033/commits/c1621b9f3250edc9c41f5ee5acd8163d6d584614 | 23:59 |
| mup | PR #8033: tests: switch mount-ns test to differential data set <Created by zyga> <https://github.com/snapcore/snapd/pull/8033> | 23:59 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!