| *** tristan has quit IRC | 01:16 | |
| *** narispo has quit IRC | 02:06 | |
| *** narispo has joined #buildstream | 02:08 | |
| *** narispo has quit IRC | 02:51 | |
| *** narispo has joined #buildstream | 02:51 | |
| *** tristan has joined #buildstream | 08:31 | |
| *** rdale has joined #buildstream | 09:30 | |
| *** benschubert has joined #buildstream | 09:40 | |
| *** phildawson has joined #buildstream | 09:41 | |
| *** tpollard has joined #buildstream | 09:53 | |
| *** jonathanmaw has joined #buildstream | 10:00 | |
| benschubert | tpollard: hey, do you have the exact ocommit I need to benchmark and the exact command? :) | 10:03 |
|---|---|---|
| tpollard | benschubert: https://gitlab.com/traveltissues/scraps/blob/buildsubprocess/results.md :) | 10:04 |
| *** bochecha has joined #buildstream | 10:04 | |
| benschubert | wait what? The code was profiled for your branch and not for master? Or was it not? | 10:05 |
| tpollard | that profile can only be ran on my branch | 10:05 |
| benschubert | yup, but the benchmarks were run without right? | 10:06 |
| tpollard | I believe traveltissues ran the benchmarks with that profile turned on. It will have added a small overhead to the time, but I don't think it's too relevant right now | 10:06 |
| tpollard | I think getting an exact matching profile of the branch is the main issue | 10:07 |
| tpollard | traveltissues is off today, I'm sure we can get another end to end time without the profiler on | 10:07 |
| benschubert | ok | 10:08 |
| juergbi | the times I reported were without profiling | 10:10 |
| juergbi | ftpmirror.gnu.org is down breaking CI due to the autotools example | 10:11 |
| tpollard | ouch | 10:13 |
| *** misterwhatever has joined #buildstream | 10:29 | |
| *** lachlan has joined #buildstream | 10:29 | |
| *** misterwhatever has quit IRC | 10:35 | |
| *** tristan has quit IRC | 10:50 | |
| *** lachlan has quit IRC | 10:50 | |
| *** lachlan has joined #buildstream | 10:53 | |
| *** akvilebirgelyte_ has quit IRC | 10:59 | |
| *** akvilebirgelyte_ has joined #buildstream | 10:59 | |
| *** lachlan has quit IRC | 11:11 | |
| *** lachlan has joined #buildstream | 11:20 | |
| *** mahmoud has joined #buildstream | 11:26 | |
| *** mahmoud has quit IRC | 11:27 | |
| gitlab-br-bot | tpollard approved MR !1723 (tlater/update-workspace-manpages->master: man/bst-workspace-reset.1: Update workspace re-open help message) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1723 | 11:29 |
| tlater[m] | Hrm, wonder if marge is smart enough to handle that | 11:29 |
| *** lachlan has quit IRC | 11:31 | |
| gitlab-br-bot | cs-shadow approved MR !1722 (valentindavid/compose-integration-commands->master: compose: Allow compose elements to run integration commands) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1722 | 11:34 |
| *** misterwhatever has joined #buildstream | 11:46 | |
| *** lachlan has joined #buildstream | 11:47 | |
| benschubert | tpollard: terminate is broken on your branch btw :) | 11:50 |
| tpollard | the signal handling on master for me on that build is broken for me too | 11:51 |
| benschubert | wait signal handling is broken on master? | 11:51 |
| tpollard | it behaves really intermittently for me on that build | 11:52 |
| tpollard | but not in more realistic builds | 11:52 |
| benschubert | so, if you ctrl + c then terminate on master it just continues? Oo | 11:53 |
| tpollard | ctrl+c quite often doesn't behave | 11:55 |
| benschubert | ah yeah that's another problem | 11:55 |
| benschubert | I meant once ctrl + c has stopped, can you 'terminate' ? | 11:55 |
| tpollard | works fine on a smaller build, or something like fdsdk | 11:56 |
| tpollard | the child_handler for the child task in job start() is giving me all sorts of async/signal handling fun | 12:00 |
| tpollard | on that build, specifically | 12:01 |
| benschubert | what do you mean? | 12:01 |
| tpollard | presumably because the pid life is so short | 12:01 |
| *** lachlan has quit IRC | 12:02 | |
| tpollard | benschubert: basically, I'm trying to handle this failure (which only happens in CI for me) https://gitlab.com/BuildStream/buildstream/-/jobs/357133118 | 12:03 |
| tpollard | a watcher is getting triggered that should not have even been attached in this instance (one in Stream, which only gets added if running build) | 12:04 |
| benschubert | but that's only on your branch right? | 12:05 |
| tpollard | I got around this by explicitly setting the global event loop to None in streams cleanup, and explicitly add the childjob watchers to schedulers event loop (not the global one) | 12:06 |
| tpollard | however, on that debian build it seems to trip up adding the childjob handler explicitly to a given loop and not the global one because of pids being reaped | 12:06 |
| benschubert | Why do you need to change this? | 12:06 |
| benschubert | there should be _exactly_ one event loop on the whole program no? | 12:06 |
| *** lachlan has joined #buildstream | 12:09 | |
| benschubert | tpollard: how can https://gitlab.com/BuildStream/buildstream/merge_requests/1613/diffs#65c9089123965199008f5d09bc3e5d13b56777dd_189_209 ever be true? | 12:09 |
| tpollard | benschubert: gitlab's not loading that link to a specific change, which exactly? | 12:10 |
| benschubert | load the page twice :) | 12:10 |
| tpollard | still does nothing for me, just takes me to the change list | 12:11 |
| benschubert | scheduler.py line 209 | 12:13 |
| tpollard | potentially because it had chosen to auto collapse it for me | 12:15 |
| tpollard | benschubert: the notify_front_queue should only exist if running bst build | 12:16 |
| *** lachlan has quit IRC | 12:17 | |
| tpollard | s/running bst build/running a subproccessed stream method | 12:17 |
| tpollard | which in that branch for now is just build | 12:17 |
| tpollard | if there's no multiprocessing, it just notifies as a callback | 12:18 |
| benschubert | Ah no sorry, I meant: | 12:20 |
| benschubert | ``` | 12:20 |
| benschubert | if not subprocessed: | 12:20 |
| benschubert | self._casd_process = casd_process | 12:20 |
| benschubert | _watcher = asyncio.get_child_watcher() | 12:20 |
| benschubert | ``` | 12:20 |
| benschubert | Why is there a difference there? | 12:20 |
| tpollard | in the current implementation, if subprocessing I have stream in the 'main process' watching casd & the interrupt signals. Due it being the parent process where casd was launched | 12:23 |
| tpollard | if we're not subproccessing, then the scheduler is executing in the main process, and as such can watch the casd child | 12:26 |
| tpollard | this may have changed since the casdprocessmanager landed, I haven't rebased to support that yet | 12:28 |
| tpollard | side effect of having the frontend initialise casd before it's actually really needed | 12:31 |
| *** narispo has quit IRC | 12:41 | |
| *** narispo has joined #buildstream | 12:42 | |
| *** narispo has quit IRC | 12:53 | |
| *** tristan has joined #buildstream | 12:54 | |
| *** narispo has joined #buildstream | 12:56 | |
| gitlab-br-bot | marge-bot123 merged MR !1724 (juerg/umask->master: Respect umask for created file and directories) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1724 | 13:07 |
| *** narispo has quit IRC | 13:19 | |
| *** narispo has joined #buildstream | 13:19 | |
| *** misterwhatever has quit IRC | 13:29 | |
| *** narispo has quit IRC | 13:41 | |
| *** narispo has joined #buildstream | 13:42 | |
| *** akvilebirgelyte__ has joined #buildstream | 13:50 | |
| *** akvilebirgelyte_ has quit IRC | 13:50 | |
| *** akvilebirgelyte__ has quit IRC | 13:52 | |
| *** akvilebirgelyte__ has joined #buildstream | 13:52 | |
| *** akvilebirgelyte__ has quit IRC | 13:57 | |
| *** akvilebirgelyte__ has joined #buildstream | 13:57 | |
| *** akvilebirgelyte__ has quit IRC | 14:21 | |
| *** akvilebirgelyte__ has joined #buildstream | 14:24 | |
| *** lachlan has joined #buildstream | 14:42 | |
| *** lachlan has quit IRC | 14:53 | |
| *** bochecha_ has joined #buildstream | 15:09 | |
| gitlab-br-bot | tlater opened (was WIP) MR !1645 (tlater/artifactserver-casd->master: Refactor casserver.py: Stop relying on the buildstream-internal `CASCache` implementation) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1645 | 15:15 |
| tlater[m] | All CI is failing, I guess | 15:17 |
| tlater[m] | Are we skirting around marge until the server issues are resolved or do we just accept no movement for a few days? | 15:18 |
| benschubert | tlater[m]: is it a planned outage? what's happening? :) | 15:18 |
| tlater[m] | benschubert: I haven't checked, I tend to assume not-planned | 15:19 |
| benschubert | Because I seem to be able to access http://ftp.snt.utwente.nl/pub/software/gnu/ | 15:20 |
| tlater[m] | Oh, no, the version of automake we're using looks to just have dropeed D: | 15:21 |
| tlater[m] | s/dropeed/dropped/ | 15:21 |
| tlater[m] | Yeah, let me fix that | 15:21 |
| benschubert | ah :D | 15:21 |
| tlater[m] | benschubert: I think it's a patch release and they've pulled the original | 15:21 |
| tlater[m] | Probably a security flaw? | 15:21 |
| benschubert | possibly :) as long as we can fix easily, I don't mind :) | 15:22 |
| tlater[m] | benschubert: Eugh, I'm going to have to run the docs generation. But yeah, not too bad, 2 line fix outside of auto-generated stuff | 15:23 |
| benschubert | great, let me know when I should review the fix | 15:24 |
| *** narispo has quit IRC | 15:41 | |
| *** narispo has joined #buildstream | 15:41 | |
| benschubert | tpollard: where is best for me to add my times and profile? on the MR in a comment? I can't add to traveltissues' repo as I don't have access | 15:56 |
| tpollard | sounds perfectly fine to me | 15:57 |
| benschubert | https://gitlab.com/BuildStream/buildstream/merge_requests/1613#note_250083875 tpollard Kinnison juergbi :) | 16:00 |
| Kinnison | are those timings without profiling? | 16:00 |
| benschubert | correct | 16:01 |
| benschubert | and minimal usage of my computer | 16:01 |
| benschubert | no other wsl thing running | 16:01 |
| Kinnison | Urgh | 16:01 |
| tpollard | at least it's constant behaviour..... | 16:01 |
| benschubert | I should have access to wsl2 in some time (hopefully before thursday) and will add new benchmarks with this | 16:02 |
| * tlater[m] wonders why we have an in-tree dependency on fuse.py | 16:12 | |
| tlater[m] | It's a little... annoying. | 16:13 |
| juergbi | will go away with buildbox-run | 16:13 |
| * tlater[m] is looking forward to that then :D | 16:17 | |
| tlater[m] | It's currently impossible to run fuse.py without patching it on NixOS | 16:17 |
| juergbi | tlater[m]: ooi, what's the issue? | 16:24 |
| tlater[m] | juergbi: It's not smart enough to find libfuse in /nix/var/nix/... | 16:25 |
| *** lachlan has joined #buildstream | 16:25 | |
| * tlater[m] assumes it doesn't check some path variable that's supposed to be checked | 16:25 | |
| tlater[m] | The nix package for fusepy is patched to simply use a static file path for that | 16:25 |
| tlater[m] | But that obviously doesn't apply to the in-tree version in BuildStream, so it takes manual patching | 16:26 |
| *** lachlan has quit IRC | 16:28 | |
| *** bochecha_ has quit IRC | 16:46 | |
| *** bochecha has quit IRC | 16:48 | |
| *** dylan-m_ has joined #buildstream | 17:02 | |
| * tlater[m] wishes we didn't use ostree in the docs' examples | 17:21 | |
| benschubert | jjardon: you added flawfinder-sast right? https://gitlab.com/BuildStream/buildstream/-/jobs/360710103 will never be running it seems or am I missing something? | 17:25 |
| *** lachlan has joined #buildstream | 17:32 | |
| tlater[m] | Would someone mind reviewing https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/59? | 17:37 |
| tlater[m] | I sort of need that to fix our CI | 17:37 |
| benschubert | link to docs is a 404 | 17:38 |
| benschubert | that's a change in the API, I'm not sure this would pass on a non-fedora syste | 17:38 |
| tlater[m] | benschubert: But then it doesn't pass on fedora systems | 17:39 |
| benschubert | would you mind checking if one of the debian works with it? | 17:39 |
| tlater[m] | I'll give it a try | 17:39 |
| *** lachlan has quit IRC | 17:40 | |
| * tlater[m] couldn't see any change in the API going through the gobject-introspection history, though | 17:40 | |
| tlater[m] | That interface has been the same for as long as it existed | 17:40 |
| benschubert | that's interesting x) | 17:46 |
| benschubert | Ok then | 17:46 |
| *** jonathanmaw has quit IRC | 17:47 | |
| tlater[m] | benschubert: My exact thoughts, hehe | 17:49 |
| benschubert | tlater[m]: workspace_reset: | 17:49 |
| benschubert | @click.option("--track", "track_", is_flag=True, help="Track and fetch the latest source before resetting") | 17:49 |
| benschubert | Is this an oversite? | 17:49 |
| benschubert | (I can remove it) | 17:49 |
| tlater[m] | Huh, yes, very much so | 17:50 |
| benschubert | ok, I'll add it to my PR :D | 17:51 |
| tlater[m] | benschubert: If you don't mind, there are a couple other references to --track | 17:51 |
| tlater[m] | i.e., one on the deprecated fetch | 17:51 |
| tlater[m] | And one in the completion code | 17:51 |
| * tlater[m] must have made a mistake grepping for those | 17:52 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1733 (bschubert/track-cleanup->master: _stream.py: Remove dead code and comments about tracking) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1733 | 17:52 |
| benschubert | tlater[m]: ^ or am I missing more? | 17:52 |
| benschubert | ah | 17:52 |
| benschubert | ok, let me remove those too | 17:52 |
| tlater[m] | Hrm, sorry about that | 17:53 |
| tlater[m] | I figure some of that may have snuck back in after the black reformatting | 17:53 |
| benschubert | no worries | 17:53 |
| benschubert | PR should be ready for review now if you want :) | 17:53 |
| tlater[m] | benschubert: 4 more references from a grep | 17:57 |
| benschubert | tlater[m]: ah, can you give me the files or the grep and I'll gix them :) | 17:57 |
| benschubert | *fix | 17:57 |
| benschubert | ah nice | 17:58 |
| tlater[m] | benschubert: https://gitlab.com/BuildStream/buildstream/merge_requests/1733#note_250146839 | 17:58 |
| tlater[m] | Also, the gobject stuff works on debian-9: https://gitlab.com/BuildStream/buildstream/-/jobs/360727781 | 17:59 |
| benschubert | omg there's actually so much more of the track :) | 17:59 |
| tlater[m] | Gah! | 18:00 |
| tlater[m] | benschubert: I assume you're backtracking from `workspace_open`? | 18:00 |
| benschubert | yep | 18:01 |
| tlater[m] | I figure that's just the tip of the iceberg, because that one happens to have something searchable | 18:01 |
| * tlater[m] should dig through the files manually | 18:02 | |
| benschubert | we should just ensure we cleanup when we see something | 18:03 |
| * tlater[m] is very annoyed that he can't get `bst2html` to do what it should :| | 18:04 | |
| tlater[m] | `junctions-build.html` just doesn't want to be updated | 18:04 |
| benschubert | tlater[m]: updated my PR :) | 18:05 |
| tlater[m] | benschubert: Does `bst track` implicitly track junctions? | 18:06 |
| tlater[m] | Doesn't look like it | 18:07 |
| benschubert | I mean, it's like any other element, you need to run 'bst track my_element' | 18:07 |
| benschubert | no? | 18:07 |
| tlater[m] | benschubert: Regarding the comment there, you need to run `bst track --track-cross-junction` | 18:08 |
| tlater[m] | To be able to track elements inside a junction | 18:08 |
| benschubert | Ah right, the last part of the comment is still true | 18:09 |
| tlater[m] | Actually, `--cross-junctions` | 18:09 |
| benschubert | but where you added your comment I don't think it is | 18:09 |
| tlater[m] | Ah, yes, that's right | 18:09 |
| benschubert | Anyways, I don't believe the trackin cross junction should be there, it should be on the cli no? | 18:10 |
| tlater[m] | Hm, that's debatable | 18:12 |
| tlater[m] | It's in both places currently | 18:12 |
| tlater[m] | Here it just explains it in a bit more detail | 18:12 |
| * tlater[m] thinks it's ok here because it explains how to work with junctions | 18:12 | |
| cs-shadow | it's not in both places currently | 18:12 |
| benschubert | ok, I'll add it back :) | 18:12 |
| cs-shadow | the CLI doesn't talk about this pecularity at all | 18:12 |
| tlater[m] | cs-shadow: `bst source track` has docs for it | 18:13 |
| cs-shadow | tlater[m]: the docs document this option, but not in this prose form | 18:13 |
| tlater[m] | Yeah, that's fair | 18:13 |
| cs-shadow | i meant to say that this text is useful, and we should ensure that we don't throw it away | 18:13 |
| * tlater[m] thinks it makes sense for it to be explained in detail in both locations | 18:13 | |
| tlater[m] | Since I'd be in junctions.html when I first figure out how to use junctions | 18:14 |
| tlater[m] | But then read `bst source track --help` to figure out why my build doesn't work again | 18:14 |
| benschubert | Pr updated with the text back | 18:14 |
| tlater[m] | benschubert: You did get me curious; how did you manage to duplicate that line in an entirely unrelated file? :D | 18:19 |
| benschubert | I didn't know which junction file you were talking about | 18:19 |
| benschubert | so was looking at them | 18:19 |
| benschubert | and must have ctrl + y instead of z | 18:20 |
| benschubert | still getting that quite often, due to CH keyboard <-> UK keyboard | 18:20 |
| tlater[m] | Ahhh | 18:22 |
| tlater[m] | Yeah, qwertz took me a while to forget as well :D | 18:22 |
| * tlater[m] thinks he'll give up on fixing CI today | 18:24 | |
| tlater[m] | Early tomorrow morning... | 18:24 |
| tlater[m] | It's just really hard to run that doc regeneration script and actually get something out of it. | 18:25 |
| benschubert | We seem to have security scans on CI now? Is it expected that they'd be failing all the time? | 18:25 |
| benschubert | we probably should rework it to have it work correctly all the time :) | 18:25 |
| tlater[m] | benschubert: The flawfinder stuff? | 18:26 |
| * tlater[m] thinks it's a broken filepath, yes | 18:26 | |
| benschubert | ah x') | 18:26 |
| benschubert | How did it go through review while failing all along? | 18:27 |
| tlater[m] | Oh, yeah, typo'd copy-paste | 18:27 |
| tlater[m] | Should be something like `flawfinder-sast.json`, I assume, just reusing another jobs' | 18:27 |
| tlater[m] | Huh, strange, we don't configure that | 18:29 |
| tlater[m] | Must be a gitlab bug | 18:29 |
| benschubert | are we missing a file that we should have created? | 18:29 |
| * tlater[m] just wonders where those templates are | 18:30 | |
| tlater[m] | Not much we can do without seeing the actual config | 18:30 |
| *** lachlan has joined #buildstream | 18:43 | |
| *** lachlan has quit IRC | 18:53 | |
| *** lachlan has joined #buildstream | 19:08 | |
| *** bochecha has joined #buildstream | 19:35 | |
| *** bochecha has quit IRC | 19:40 | |
| *** bochecha has joined #buildstream | 19:40 | |
| *** bochecha has quit IRC | 19:44 | |
| *** lachlan has quit IRC | 19:48 | |
| *** lachlan has joined #buildstream | 19:50 | |
| *** bochecha has joined #buildstream | 20:06 | |
| *** misterwhatever has joined #buildstream | 20:08 | |
| *** lachlan has quit IRC | 20:13 | |
| *** lachlan has joined #buildstream | 20:16 | |
| *** kapip has quit IRC | 20:29 | |
| *** lachlan has quit IRC | 20:58 | |
| *** rdale has quit IRC | 22:13 | |
| *** cs-shadow has quit IRC | 22:32 | |
| *** bochecha has quit IRC | 22:48 | |
| *** dylan-m_ has quit IRC | 22:59 | |
| *** dylan-m_ has joined #buildstream | 23:25 | |
| *** slaf has quit IRC | 23:45 | |
| *** slaf has joined #buildstream | 23:58 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!