| *** nimish2711 has joined #buildstream | 00:31 | |
| *** connorshea[m] has quit IRC | 01:51 | |
| *** nimish2711 has quit IRC | 03:15 | |
| *** alatiera has quit IRC | 03:17 | |
| *** tristan has quit IRC | 05:14 | |
| *** nimish2711 has joined #buildstream | 05:26 | |
| *** tristan has joined #buildstream | 05:27 | |
| *** mohan43u has joined #buildstream | 05:52 | |
| *** nimish2711 has quit IRC | 06:03 | |
| gitlab-br-bot | tristanvb opened issue #1013 (Better error message when staging a file into a not-directory) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1013 | 07:29 |
|---|---|---|
| *** bochecha has joined #buildstream | 07:50 | |
| *** alatiera has joined #buildstream | 07:57 | |
| *** tpollard has joined #buildstream | 08:41 | |
| *** raoul has joined #buildstream | 08:52 | |
| *** jonathanmaw has joined #buildstream | 09:34 | |
| *** lachlan has joined #buildstream | 09:38 | |
| *** lachlan has quit IRC | 09:42 | |
| gitlab-br-bot | marge-bot123 merged MR !1313 (fixed-bug-for-pip-test->master: tests/testutil/python_repo.py: fixed executable path when running pip) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1313 | 09:44 |
| *** lachlan has joined #buildstream | 10:02 | |
| *** ChanServ sets mode: +o tristan | 10:17 | |
| tristan | It is damn difficult to make a test for #1012 :-/ | 10:17 |
| tristan | Maybe I should just fix it without a test | 10:17 |
| tristan | It would require a clean state for every `cli` invocation, which I guess by itself is doable | 10:18 |
| tristan | I need to cleanup the plugins table and re-initialize the counter, but by doing that I've already fixed the issue | 10:19 |
| tristan | juergbi, Any opinion on that ? Maybe I should just fix the bug in this case without regression test ? | 10:20 |
| tristan | The regression test also requires that (A) We have a source that is configured with an alias, but not a mirror, so that we hit the `track` scenario with a cloned Source, and we also require that that track implementation triggers a frontend message | 10:22 |
| tristan | Somehow I'm having trouble reproducing in there | 10:22 |
| tristan | actually it appears that cloned sources in the `cli` fixtures dont produce any messages at all, instead of crashing :-S | 10:23 |
| juergbi | tristan: it might not be worth the effort to have a test for this, imo | 10:24 |
| juergbi | we should make sure to add a proper comment for the itertools count change, though | 10:24 |
| tristan | yeah I have a patch which does exactly that (notes why the counter must start at 1) | 10:25 |
| tristan | on the other hand, if I do find time, getting to the bottom of it would have benefits... maybe | 10:25 |
| juergbi | maybe we could add a check for ID 0 in the plugin ID error message | 10:25 |
| juergbi | i.e., in case it were to happen again, we would get clearer error message | 10:25 |
| tristan | Right | 10:25 |
| tristan | I found another bug in the process of trying to wrangle up a test, so I aught to file/fix that too | 10:26 |
| tristan | it appears that validation of mirror configuration is not being done properly (if you use a string instead of a list of URL mirrors, you hit an assert instead of a LoadError() :-S) | 10:27 |
| *** lachlan has quit IRC | 10:39 | |
| *** lachlan has joined #buildstream | 10:43 | |
| gitlab-br-bot | tristanvb opened MR !1315 (tristan/fix-cloned-plugin-ids-1.2->bst-1.2: Tristan/fix cloned plugin ids 1.2) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1315 | 10:43 |
| gitlab-br-bot | tristanvb opened MR !1316 (tristan/fix-cloned-plugin-ids->master: Fix cloned plugin ids) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1316 | 10:48 |
| tristan | juergbi, Also added that assertion, in both branches (master and bst-1.2 MRs) | 10:48 |
| * tristan should roll out a 1.2.6 with this fix soon | 10:49 | |
| tristan | I think it effects any invocation of `bst track foo.bst` (single source tracking) in any project which has mirrors declared where the source does not implement source fetchers | 10:49 |
| tristan | (or even git, fetchers are orthogonal to tracking) | 10:50 |
| benschubert | tristan: the addition of the assert in 'lookup' might have a significant cost given how many times we call it :/ | 10:50 |
| tristan | benschubert, I'm rather ambivalent about that assert myself | 10:50 |
| tristan | I mean, it would be better to remain an assert, and introduce a Bug() exception that doesnt inherit from BstError | 10:51 |
| tristan | (i.e. make it safe to disable asserts) | 10:51 |
| benschubert | tristan: and for the change, what about setting clone._unique_id = self._unique_id, we would skip some indices but that should not be a problem, and this would be faster for the "norma" use case? | 10:51 |
| tristan | benschubert, Well, that's what I had initially, and I thought it was a horribly brute force hack of an eye sore ;-) | 10:52 |
| tristan | I do see your concern though | 10:52 |
| benschubert | true, but if it's more performant :) | 10:52 |
| tristan | avoid a single if for every instantiation | 10:53 |
| benschubert | This part is called a lot and has a significant performance impact | 10:53 |
| tristan | yeah | 10:53 |
| benschubert | actually might not be that bad, plugin.__init__ takes 3s out of a 250s run for bst show | 10:54 |
| benschubert | so it's not _that_ bad | 10:54 |
| tristan | Heh | 10:56 |
| tristan | That depends I think | 10:57 |
| tristan | I.e. a 250s run of bst show may very well perform differently depending on what sources are employed (interrogation of source consistency is probably the most expensive) | 10:57 |
| tristan | benschubert, I'm going to step out and meed friends for dinner, I'll want to close this tomorrow though | 10:59 |
| tristan | benschubert, I would be more happy keeping the assert in the knowledge that the longer term plan is to replace important asserts with `raise Bug()` | 10:59 |
| tristan | And I am happy to change the __init__ approach if you deem it to be important | 11:00 |
| benschubert | tristan: sure, fine by me. We can always revert if that's a problem later :) | 11:00 |
| gitlab-br-bot | BenjaminSchubert approved MR !1316 (tristan/fix-cloned-plugin-ids->master: Fix cloned plugin ids) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1316 | 11:00 |
| tristan | Ok thanks :) | 11:00 |
| *** tristan has quit IRC | 11:05 | |
| *** lachlan has quit IRC | 11:06 | |
| laurence | would anyone object to making me a 'maintainer' at the group level ? https://gitlab.com/BuildStream | 11:14 |
| laurence | i would like to tidy up some of the benchmarks repo organisation, and i think i need higher permissions in order to create a new subgroup | 11:14 |
| laurence | i promise to not abuse my power | 11:14 |
| *** lachlan has joined #buildstream | 11:26 | |
| *** tristan has joined #buildstream | 11:31 | |
| *** lachlan has quit IRC | 11:37 | |
| valentind | tlater[m], let's talk about the cache server in tomorrow's meeting. I think we could just use the freedesktop sdk cache. But I want to know the opinions of others. | 11:50 |
| tlater[m] | valentind I'm not sure I can make tomorrow's meeting, unfortunately | 11:51 |
| tlater[m] | University has started again, and this particular week is the final week for a number of projects. | 11:52 |
| valentind | tlater[m], OK. But I will talk about that tomorrow before I start setting it up. | 11:52 |
| valentind | I will update you tomorrow on that. | 11:52 |
| tlater[m] | Thanks - I'll try to glance over irc as much as possible | 11:53 |
| *** lachlan has joined #buildstream | 12:08 | |
| *** lachlan has quit IRC | 12:12 | |
| *** raoul_ has joined #buildstream | 13:00 | |
| *** lachlan has joined #buildstream | 13:01 | |
| *** raoul has quit IRC | 13:02 | |
| *** CTtpollard has joined #buildstream | 13:02 | |
| *** tpollard has quit IRC | 13:03 | |
| gitlab-br-bot | phildawson opened MR !1317 (phil/separate-local-tests->master: testing._sourcetests: Don't special case 'local' in parameter list) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1317 | 13:07 |
| *** lachlan has quit IRC | 13:13 | |
| jennis | juergbi, regarding your review comment: https://gitlab.com/BuildStream/buildstream/merge_requests/1299#note_163509529, are you eluding to having such logic within the CacheKey base class or still within Element? | 13:31 |
| adds68 | 14:41 <@alexlarsson> ramcq: https://github.com/flatpak/flatpak/pull/2867 GET YA SOME DOCKER! | 13:47 |
| adds68 | Add flatpak in docker seccomp profile by alexlarsson · Pull Request #2867 · flatpak/flatpak | 13:47 |
| adds68 | This is a docker seccomp profile that allows you to run flatpak inside a docker container, given some special requirements: The host kernel must support unprivileged user namespaces (Supported by ... | 13:47 |
| adds68 | 14:42 <ramcq> d to the o the c to the k! | 13:47 |
| adds68 | 14:42 <ramcq> nice | 13:47 |
| adds68 | 14:44 <bochecha> can something similar help run buildstream (i.e bubblewrap) inside unpriviledged Docker? | 13:47 |
| adds68 | 14:44 <@alexlarsson> bochecha: yes | 13:47 |
| adds68 | 14:45 <ramcq> bochecha: surely yes | 13:47 |
| adds68 | 14:45 <bochecha> this is going to be very nice for a lot of CIs :) | 13:47 |
| adds68 | 14:45 <bochecha> (gnome, fdo sdk, …) | 13:47 |
| adds68 | benschubert ^ | 13:47 |
| adds68 | good catch bochecha :D | 13:48 |
| bochecha | adds68: well, this is the kind of something we (flatpak, buildstream, …) have all been wondering for a while now | 13:48 |
| bochecha | Alex just went and did it ^^ | 13:48 |
| bochecha | s/the kind of // (wth did I try to type?) | 13:48 |
| adds68 | bochecha haha, praise to Alex :) | 13:49 |
| *** nimish2711 has joined #buildstream | 13:52 | |
| juergbi | jennis: I expect Element class to be a better fit (see also tristan's comment) | 13:58 |
| juergbi | with the weak_cached simplification, these methods might actually be completely cache key independent | 13:58 |
| *** lachlan has joined #buildstream | 14:00 | |
| *** raoul_ has quit IRC | 14:03 | |
| *** raoul_ has joined #buildstream | 14:04 | |
| *** lachlan has quit IRC | 14:06 | |
| *** nimish2711_ has joined #buildstream | 14:11 | |
| *** nimish2711 has quit IRC | 14:11 | |
| *** nimish2711_ is now known as nimish2711 | 14:12 | |
| *** nimish2711_ has joined #buildstream | 14:23 | |
| *** nimish2711 has quit IRC | 14:23 | |
| *** nimish2711_ is now known as nimish2711 | 14:23 | |
| *** lachlan has joined #buildstream | 14:46 | |
| *** lachlan has quit IRC | 14:49 | |
| *** CTtpollard is now known as tpollard | 14:55 | |
| *** bochecha has quit IRC | 15:26 | |
| gitlab-br-bot | marge-bot123 merged MR !1312 (jonathan/reduce-update-state-calls->master: Reduce the amount of times we call Element._update_state()) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1312 | 15:35 |
| tpollard | I'm trying to update artifactcache/push.py test to support artifact as a proto | 15:51 |
| tpollard | the element being pushed is a stack element, and internally requires a method associated to to it's __artifact member in element.py to be called. | 15:52 |
| tpollard | my issue is that it seems to not be initialising an artifact member for it (this use to be done in the constructor for element, but now it's done in _update_state) | 15:53 |
| tpollard | I've tried to force a _preflight() and update_state() on it (to match pipelines resolve_elements) but it still lacks an __artifact | 15:54 |
| tpollard | the integration tests are working fine, so I'm pretty sure this is to do with poking around internally on the specific test | 15:56 |
| tpollard | so I'm wondering if a stack element purposely doesn't get initialised with keys or not | 15:58 |
| juergbi | tpollard: I don't think so, stack elements don't have special treatment in that regard | 16:07 |
| juergbi | at what point in the test do you get an error and what's the backtrace? | 16:09 |
| *** lachlan has joined #buildstream | 16:12 | |
| tpollard | juergbi: if you take https://gitlab.com/BuildStream/buildstream/commits/tpollard/prototemp and run tests/artifactcache/push.py | 16:13 |
| tpollard | _push_artifact in _artifact_cache it attempts to call the simple accessor to the artifact method for _get_proto | 16:14 |
| tpollard | which worked fine before moving __artifact creation to update_state | 16:14 |
| tpollard | I have the MR working without going through artifact, but we'd prefer to just have one entry point | 16:15 |
| *** lachlan has quit IRC | 16:15 | |
| tpollard | juergbi: it has me pretty stumped, but that's not too unusual especially on a Monday... | 16:34 |
| laurence | Hi all, I am on holiday next time there's a Buildstream monthly irc team meeting | 16:35 |
| laurence | Tuesday 14 May | 16:35 |
| laurence | Can someone volunteer to run it? | 16:35 |
| laurence | including a call for agenda the week before it | 16:35 |
| laurence | the 'shake the tree' email, as i like to call it | 16:35 |
| *** phil has joined #buildstream | 16:37 | |
| juergbi | tpollard: the issue could be that the elements are not marked as required when fiddling with the internals | 16:37 |
| juergbi | although, update_state should create Artifact in any case, afaict | 16:38 |
| *** phildawson_ has quit IRC | 16:38 | |
| tpollard | indeed :S | 16:41 |
| juergbi | tpollard: have you also tried calling update_state on the dependencies? | 16:44 |
| tpollard | if it can be sussed then _artifact_cache.py having its own implementation of getting the proto can be removed | 16:44 |
| juergbi | tpollard: https://pastebin.com/i6PSU1KY | 16:48 |
| gitlab-br-bot | BenjaminSchubert opened MR !1318 (bschubert/remove-duplicated-code->master: _yaml.py: remove duplicated check) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1318 | 16:52 |
| tpollard | juergbi: that might do the trick, I'll check in the morning :) | 16:54 |
| laurence | i'll ask tomorrow morning, if i remember | 17:00 |
| *** jonathanmaw has quit IRC | 17:02 | |
| *** lachlan has joined #buildstream | 17:18 | |
| *** lachlan has quit IRC | 17:22 | |
| *** raoul_ has quit IRC | 17:23 | |
| *** lachlan has joined #buildstream | 17:25 | |
| *** lachlan has quit IRC | 17:29 | |
| *** lachlan has joined #buildstream | 17:39 | |
| *** lachlan has quit IRC | 17:44 | |
| *** phil has quit IRC | 17:52 | |
| *** phil has joined #buildstream | 17:52 | |
| *** nimish2711 has quit IRC | 18:42 | |
| *** lachlan has joined #buildstream | 19:17 | |
| *** lachlan has quit IRC | 19:29 | |
| *** lachlan has joined #buildstream | 19:40 | |
| *** lachlan has quit IRC | 19:49 | |
| *** gokcennurlu has quit IRC | 20:04 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!