| *** rdale has quit IRC | 00:21 | |
| *** nimish2711 has joined #buildstream | 02:26 | |
| *** tristan has quit IRC | 04:27 | |
| *** tristan has joined #buildstream | 04:30 | |
| *** ChanServ sets mode: +o tristan | 04:34 | |
| tristan | laurence, for when you get in: I don't know why you receive any moderator emails for buildstream-list@gnome.org, the admin UI tells me that jjardon and myself are the only moderators | 04:36 |
|---|---|---|
| tristan | laurence, Also, I am having a hard time tracking down the email automatically sent to me which informs me of the admin password, it appears that I can only use the UI because the password is cached | 04:36 |
| * tristan thinks he had jjardon send him the password at some time | 04:36 | |
| *** nimish2711 has quit IRC | 04:42 | |
| *** alatiera has quit IRC | 04:50 | |
| *** alatiera has joined #buildstream | 04:53 | |
| *** nimish2711 has joined #buildstream | 05:02 | |
| gitlab-br-bot | tristanvb opened issue #999 (Warnings and errors whenever overriding subproject junction configurations) on buildstream https://gitlab.com/BuildStream/buildstream/issues/999 | 05:28 |
| tristan | Ohhh nice number ! | 05:29 |
| tristan | appropriate, it is a very nice issue ! | 05:29 |
| tristan | cs-shadow, juergbi fwiw, elaborately summarized and detailed ^^^^ | 05:29 |
| *** tristan has quit IRC | 05:45 | |
| *** tristan has joined #buildstream | 06:22 | |
| laurence | tristan, i think you both are the oweners, and i'm just a mod? anyway, yeah i have that issue with passwords too | 06:32 |
| laurence | should really use a password manager | 06:32 |
| * laurence adds to todo list | 06:32 | |
| laurence | anyway i'll ask jjardon to send the password my way once he's awake (think he's west coast usa atm) - thanks ! | 06:33 |
| *** lchlan has quit IRC | 06:35 | |
| laurence | i wonder who will open issue #1000... | 06:37 |
| *** rdale has joined #buildstream | 06:44 | |
| *** lchlan has joined #buildstream | 06:59 | |
| *** nielsdg has joined #buildstream | 07:10 | |
| *** nimish2711 has quit IRC | 07:31 | |
| *** swick has left #buildstream | 07:43 | |
| *** toscalix has joined #buildstream | 07:57 | |
| *** admin has joined #buildstream | 08:18 | |
| *** alatiera has quit IRC | 08:40 | |
| *** alatiera has joined #buildstream | 08:41 | |
| *** jonathanmaw has joined #buildstream | 08:43 | |
| *** tristan has quit IRC | 08:59 | |
| *** phildawson_ has joined #buildstream | 09:03 | |
| *** phildawson has quit IRC | 09:04 | |
| *** tristan has joined #buildstream | 09:17 | |
| *** alatiera_ has joined #buildstream | 09:32 | |
| *** alatiera has quit IRC | 09:32 | |
| *** alatiera_ is now known as alatiera | 09:33 | |
| jennis | lchlan, tlater, I'm finding that with your benchmarking tool, the default behaviour is failing due to: https://gitlab.com/BuildStream/buildstream/issues/988 | 09:36 |
| jennis | Are you also seeing this? | 09:36 |
| *** lachlan has joined #buildstream | 09:37 | |
| *** alatiera_ has joined #buildstream | 09:39 | |
| *** alatiera has quit IRC | 09:39 | |
| *** alatiera_ is now known as alatiera | 09:40 | |
| *** phildawson_ is now known as phildawson | 10:04 | |
| phildawson | Could I get a quick review for !1295 please. It fixes a bug where the external plugin testing will fail if no plugins are provided because a .coverage file is not generated. | 10:04 |
| gitlab-br-bot | MR !1295: ix empty external tests https://gitlab.com/BuildStream/buildstream/merge_requests/1295 | 10:04 |
| *** ChanServ sets mode: +o tristan | 10:17 | |
| tristan | phildawson, Done | 10:17 |
| *** nimish2711 has joined #buildstream | 10:18 | |
| phildawson | cheers tristan. FYI keys specified in environments which inherit form [testenv] etc overriden, not extended. | 10:19 |
| *** lachlan has quit IRC | 10:19 | |
| phildawson | You can verify this by doing tox --show-config to inspect the configuration for each environment after tox has interpreted tox.ini | 10:19 |
| *** lachlan has joined #buildstream | 10:33 | |
| tristan | phildawson, they are only overridden if specified in other environments though, right ? | 10:35 |
| tristan | phildawson, That means that whenever you work on another environment, you *must* declare something if you want to ensure you are not inheriting something from [testenv] | 10:35 |
| tristan | phildawson, I setup tox.ini specifically so that this would not be the case | 10:36 |
| tristan | and that nothing is inherited by accident | 10:36 |
| tristan | phildawson, it would be nice if say, we could declare all the juicy test bits inside a [testenv:py{35,36,37}] block, and thus have clearer looking conditionals in there | 10:37 |
| tristan | but I don't think that's possible | 10:37 |
| *** lachlan has quit IRC | 10:42 | |
| phildawson | tristan, you're correct on both counts there. I see where you're coming from with how you initially wrote it but it was starting to make it hard to read (for me at least). I'll admit though I'm struggling to think of an environment where you wouldn't specify anny commands? | 10:42 |
| phildawson | "<tristan> but I don't think that's possible" not that I'm aware of, but I may be missing something | 10:42 |
| *** alatiera has quit IRC | 10:46 | |
| *** alatiera has joined #buildstream | 10:49 | |
| phildawson | I spent some time trying to work out a way to do that when writing the external testing stuff. | 10:51 |
| tristan | Yeah the conclusion I keep coming to is that we need the conditionals to be done at the assignments of keys | 10:56 |
| tristan | tox has multiple ways | 10:56 |
| *** lachlan has joined #buildstream | 10:56 | |
| tristan | but that seems to be the only one that works for all the cases we use conditionals for | 10:56 |
| tristan | phildawson, also I was hoping to say... be able to *explicitly* include declarations from [testenv] in a separate env, but it seems there is a hard coded list of variables/keys and you cannot declare things in [testenv] and then refer to them in other sections :-/ | 10:57 |
| tristan | phildawson, for commands it is not very relevant, but for deps, setenv, passenv, etc... everything else I think it is important to not allow adjacent environments to inherit anything | 10:58 |
| phildawson | tristan, would you be happy if I reinstate the conditionals for the keys which aren't command? I personally like using the deps as a default, but I don't feel strongly enough to worry either way. | 11:01 |
| tristan | phildawson, If you want to change the approach entirely, then it might be better - in which case I would want all environments to explicitly override everything, consistently; even in cases where no override is needed and only the setting inherited from [testenv] is unneeded | 11:03 |
| tristan | i.e. then we have each env very explicit and *clear* about it being *required* to set everything | 11:03 |
| tristan | And it would be good to have comments to that effect near the top of the file explaining why | 11:04 |
| tristan | I.e. I don't think it's fair to say "You have to know all about tox" or read upstream docs just to fiddle with the tox.ini file, I prefer that it is self documenting | 11:04 |
| tristan | phildawson, The main reason I had set it up that way was to ensure that others who happen to fiddle with the tox.ini in other sections, dont find themselves wondering why settings they didnt declare are curiously set | 11:05 |
| tristan | So if we can make it 100% explicit across the board and reduce the conditionals in [testenv], I think that also solves the problem | 11:05 |
| tristan | It is however a complete switch in the way we use tox | 11:06 |
| tristan | phildawson, I would however prefer consistency; I.e. not having a dual standard is better | 11:06 |
| *** nimish2711 has quit IRC | 11:31 | |
| *** alatiera has quit IRC | 11:35 | |
| *** lachlan has quit IRC | 11:37 | |
| cs-shadow | question about import plugin - it does not do anything with its build dependencies, right? | 11:43 |
| tristan | cs-shadow, iirc juergbi has a branch that disallows build dependencies of imports, or was planning one | 11:45 |
| tristan | cs-shadow, only blocker is that a lot of test cases need tedious updates | 11:45 |
| cs-shadow | tristan: ah I see. Thanks for confirming my suspicions | 11:46 |
| *** lachlan has joined #buildstream | 11:47 | |
| *** alatiera has joined #buildstream | 11:54 | |
| *** lachlan has quit IRC | 11:58 | |
| juergbi | cs-shadow: yes, see juerg/import-no-build-deps branch. not planning to work on this right now, though | 12:00 |
| cs-shadow | thanks. I'll see if I have time to push it over the finish line | 12:01 |
| tristan | Probably take an hour or two to s/depends/runtime-depends/ in all the tests which use import elements | 12:01 |
| phildawson | tristan, I agree a proper rework would be nice, but I don't really have time to do that right now. I just attempted to throw up a quick patch tidying things up a little while I was already in tox.ini for something else. I know it's not the perfect final solution, but I've attempted to leave things slightly better than I found them. If you'd prefer, I can drop those commits and open an issue about refactoring tox.ini. | 12:02 |
| tristan | phildawson, Ok so then please omit the tidying until later | 12:03 |
| tristan | phildawson, I know it's verbose, but right now the strategy is to have [testenv] not export things which get inherited, so it should be pretty consistent in this regard, switching it to the other way around and making everything explicit in the adjacent envs would be equally consistent, but let's not have a situation where "some keys work this way, and some keys work the other way" | 12:05 |
| tristan | interesting... tox not working "right now" for me | 12:06 |
| tristan | pip is timing out | 12:06 |
| juergbi | tristan, cs-shadow: just switching everything to runtime-depends may not lead to identical tests. the issue is that (I think) we're using import elements also to test scheduler behavior with regards to build dependencies | 12:13 |
| tristan | I see | 12:13 |
| juergbi | i.e., while import elements don't actually use build dependencies, the scheduler / element logic might still behave differently | 12:13 |
| tristan | https://gitlab.com/BuildStream/buildstream/-/jobs/197506163 looks like a spurious error caused by pytest-xdist | 12:14 |
| juergbi | I've not seen this INTERNALERROR before, but seems likely | 12:16 |
| tristan | complains that we're using signals in a thread and not in the main process | 12:16 |
| tristan | seems fragile that it pytest-xdist is using threads :-/ | 12:16 |
| tristan | well, not an issue if it hardly ever causes a problem | 12:17 |
| phildawson | tristan, I've dropped those commits, are you happy for me to assign to marge? | 12:19 |
| tristan | phildawson, marge it ! | 12:20 |
| phildawson | Cheers :) | 12:20 |
| juergbi | tristan: xdist is using multiple python processes, not threads, afaik, for the parallel jobs. however, there must be a strange issue somewhere with regards to asyncio | 12:23 |
| tristan | I see | 12:23 |
| tristan | yeah the INTERNALERROR is about using _signal in a thread | 12:24 |
| tristan | we dont do any threads ourselves so, maybe the error is just misplaced | 12:24 |
| tristan | or maybe we have another bad actor library going a head and *spawning threads* without being asked to !!! | 12:24 |
| juergbi | https://bitbucket.org/pytest-dev/pytest-timeout/issues/8/pytest-xdist-interop-signal-only-works-in | 12:25 |
| juergbi | https://github.com/pytest-dev/pytest-xdist/issues/204 | 12:25 |
| * tristan still recalls a mailing list thread from 15 years ago when we decided to make glib initialize the threading subsystem unconditionally | 12:25 | |
| juergbi | haha | 12:26 |
| *** lachlan has joined #buildstream | 13:33 | |
| *** lachlan has quit IRC | 13:39 | |
| *** lachlan has joined #buildstream | 13:46 | |
| laurence | nb all - monthly irc team meeting starting in 13 mins | 13:47 |
| laurence | over on #buildstream-meetings | 13:47 |
| *** lachlan has quit IRC | 13:59 | |
| *** alatiera_ has joined #buildstream | 14:01 | |
| *** alatiera has quit IRC | 14:01 | |
| *** lachlan has joined #buildstream | 14:01 | |
| *** alatiera_ is now known as alatiera | 14:01 | |
| *** lachlan has quit IRC | 14:23 | |
| gitlab-br-bot | tristanvb opened MR !1294 (tristan/rename-errno-fix-1.2->bst-1.2: plugins/sources/git.py: Cope with rename returning error EEXIST) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1294 | 14:37 |
| *** alatiera has quit IRC | 14:38 | |
| gitlab-br-bot | jonathanmaw approved MR !1291 (phil/rename-plugintestutils->master: Rename 'plugintestutils' package to 'testing') on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1291 | 14:39 |
| tristan | Huh ? | 14:39 |
| tristan | Did I ? | 14:39 |
| gitlab-br-bot | marge-bot123 merged MR !1294 (tristan/rename-errno-fix-1.2->bst-1.2: plugins/sources/git.py: Cope with rename returning error EEXIST) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1294 | 14:40 |
| tristan | gitlag-br-bot, are you labbing ? | 14:40 |
| gitlab-br-bot | phildawson opened MR !1295 (phil/fix-empty-external-tests->master: ix empty external tests) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1295 | 14:40 |
| phildawson | Erm, the bot appears to be rather confused | 14:43 |
| gitlab-br-bot | cs-shadow opened issue #1001 (Provide better error messages when launching a shell is not possible) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1001 | 14:45 |
| gitlab-br-bot | marge-bot123 merged MR !1295 (phil/fix-empty-external-tests->master: Fix empty external tests) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1295 | 14:48 |
| tpollard | marge is broken on the inside again | 14:49 |
| gitlab-br-bot | phildawson closed issue #944 (Expose templated tests for source plugins) on buildstream https://gitlab.com/BuildStream/buildstream/issues/944 | 14:51 |
| *** lachlan has joined #buildstream | 14:51 | |
| *** tristan has quit IRC | 14:53 | |
| Kinnison | Poor marge | 14:54 |
| *** tristan has joined #buildstream | 15:10 | |
| tpollard | juergbi: I've replied to your comments on !1271, hopefully we can resolve them :) | 15:12 |
| gitlab-br-bot | MR !1271: Follow-up from "Artifact 'abstraction' class" https://gitlab.com/BuildStream/buildstream/merge_requests/1271 | 15:12 |
| juergbi | ok, ta | 15:12 |
| *** lachlan has quit IRC | 15:20 | |
| *** lachlan has joined #buildstream | 15:56 | |
| *** lachlan has quit IRC | 16:01 | |
| *** toscalix has quit IRC | 16:08 | |
| *** lachlan has joined #buildstream | 16:12 | |
| *** lachlan has quit IRC | 16:16 | |
| *** tpollard has quit IRC | 16:59 | |
| *** jonathanmaw has quit IRC | 17:03 | |
| *** lachlan has joined #buildstream | 17:05 | |
| *** lachlan has quit IRC | 17:30 | |
| *** nimish2711 has joined #buildstream | 18:02 | |
| *** nimish2711 has quit IRC | 19:45 | |
| *** tristan has quit IRC | 21:34 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!