[08:12:11] hashar: do you know offhand the maximum length for the message field in CommentInfo? I don't see this in https://gerrit.wikimedia.org/r/Documentation/rest-api-changes.html#comment-info [08:12:35] undocumented :) [08:12:38] potentially 2G? [08:13:07] one of the potential issue is the comment made is stored iin the git repository [08:13:22] so all the CI build reports will end up being stored in git and get the repo growing and growing [08:13:44] albeit they are stored under `refs/changes/*` so I guess for a regular `git clone` it is not going to cause any havoc [08:13:53] but that definitely will grow on the server side [08:14:25] as for the answer, I am guessing CommentInfo.message is unlimited [08:17:36] ok [08:17:54] i'll try to remove all the cruft from at least phpunit jobs to reduce the amount of text we are adding as comments [08:18:32] or well just point to the console :] [08:18:33] https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GrowthExperiments/+/896016 is stil using 1.5.2 [08:18:47] the jobs are being updated still [08:18:51] k [08:20:03] manually processing the output might be rather fragile [08:20:18] I mean, parsing the PHPUnit console output [08:21:06] maybe we can have the javascript frontend code to retrieve the output and display it [08:21:14] this way at least it will not be stored in the git repos [08:24:00] hmm. For me, I'd like to see in my email (which is the first place I get notified about a failure) what the actual error is. So I don't need to follow another link. [08:25:02] indeed :] [08:26:23] kostajh: all jobs are now using 1.5.3 [08:30:02] cool [08:30:47] side question: I thought one concept of our CI is that all tests should run, even if a single one fails. So e.g. in https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GrowthExperiments/+/896016, there is a phpcs failure, but shouldn't we still run phpunit tests? [08:31:13] perhaps composer-test should have its own job, and it should be removed as a stage from the phpunit/qunit job [08:35:30] when I have developed it I have serialized the operations with the shortest one first [08:36:14] with the reasonning that if a php file has an invalid syntax there is no point in running any test [08:36:34] so all the jobs there are short circuited early on [08:36:39] with the excetion of https://integration.wikimedia.org/ci/job/quibble-vendor-mysql-php74-selenium-docker/29736/console : SUCCESS in 15m 46s [08:36:53] which does not run any lint but does run the phpunit tests (which pass) [08:37:22] the issue is there is no good way to make a multi stage pipeline in Zuul [08:37:35] well there is, but I am afraid it might be too complicated to implement [08:37:50] we would need the repositories to have a `lint` Pipeline in Zuul which would vote Lint+1 [08:38:10] which then in Zuul triggers a `test` pipeline which will run the other long running tests and ultimately Verified+1 [08:38:28] while on CR+2 we would still have `gate-and-submit` which runs all jobs from both `lint` and `test` [08:39:14] ok [08:39:30] some of the logic is carried and split between the various Jenkins jobs (via `--run` or `--skip`) and Quibble itself (which runs `composer test` before `phpunit` which runs before `selenium`) [08:39:32] but shouldn't the phpunit tests fail with https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GrowthExperiments/+/896016/1/tests/phpunit/unit/Config/GrowthExperimentsMultiConfigTest.php ? [08:39:33] so it is rather messy [08:40:26] I guess the test is not run? [08:43:49] yeah that test is never run cause in the jobs we run phpcs first which fails. So the jobs are shortcircuited. [08:44:06] which should make Zuul to report earlier but the Selenium tests are NOT shortcircuited [08:44:21] so still have to wait for the selenium tests to complete before having a report that the various jobs that failed all failed due to phpcs (which we thus run several time) [08:48:12] ok [09:02:50] re message length, it turns out the answer is "Response: One or more comments were rejected in validation: Comment size exceeds limit (29070 > 16384)" [09:18:29] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity, 10MW-1.40-notes (1.40.0-wmf.26; 2023-03-06): Provide early feedback when a patch has job failures - https://phabricator.wikimedia.org/T323750 (10kostajh) [09:21:53] hashar: it works https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GrowthExperiments/+/896016/2#message-511daa77d230bafb1eda926279ed8407e7381593 :) ... could be improved but seems like a good start [09:43:53] ahaha well done :] [09:44:59] I'm fixing the formatting to use `> ` for code block style [09:52:39] and you can probably strip ansi coloring stuff :) [09:53:25] also [09:53:33] you can wrap the block of text with ``` [09:53:49] that would make it use monospace font AND prefix the block similar to `> ` [09:54:54] (of course if the output has a line ``` that would cause the block to be terminated too early [09:55:47] sorry, wrap with what? it's hard to see in IRC :) [09:56:07] I mean, I can see the effect... but irccloud just shows me the formatted result [09:56:10] ah sorry [09:56:54] oh [09:56:56] I see [09:56:57] ``` [09:57:08] that's all I needed? oof... TIL [09:57:40] https://phabricator.wikimedia.org/F36912421 [09:57:55] that is render on the client side using javascript [09:58:09] it finds the end of the quote using `line.trim() === '```'` [09:58:11] err [09:58:24] it finds the end of the quote using: line.trim() === '```' [09:58:35] yeah well it is broken when using backtick hence the screenshot [09:58:39] so no `> ` is needed [09:58:43] yeah [09:59:01] until Gerrit frontend folks breaks the rendering in a future update [10:00:07] heh [13:43:57] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity: Capture output from failed command and transmit to earlywarningbot - https://phabricator.wikimedia.org/T331061 (10kostajh) 05Open→03Resolved [13:44:01] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity, 10MW-1.40-notes (1.40.0-wmf.26; 2023-03-06): Provide early feedback when a patch has job failures - https://phabricator.wikimedia.org/T323750 (10kostajh) [13:44:13] 10Quibble, 10Developer Productivity: earlywarningbot: Provide instructions for "recheck" - https://phabricator.wikimedia.org/T330749 (10kostajh) 05Open→03Resolved [13:44:15] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity, 10MW-1.40-notes (1.40.0-wmf.26; 2023-03-06): Provide early feedback when a patch has job failures - https://phabricator.wikimedia.org/T323750 (10kostajh) [14:19:13] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity: Run Phan jobs via Quibble - https://phabricator.wikimedia.org/T332178 (10kostajh) [14:20:04] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity: Run MediaWiki core/skin/extension mwgate-node16 jobs via Quibble - https://phabricator.wikimedia.org/T332179 (10kostajh) [14:20:32] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity: Run MediaWiki core/skin/extension Phan jobs via Quibble - https://phabricator.wikimedia.org/T332178 (10kostajh) [14:25:42] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity: Run MediaWiki core/skin/extension Phan jobs via Quibble - https://phabricator.wikimedia.org/T332178 (10Jdforrester-WMF) Is this just {T280990}? [15:16:16] 10Quibble, 10Continuous-Integration-Infrastructure, 10Developer Productivity: Run MediaWiki core/skin/extension Phan jobs via Quibble - https://phabricator.wikimedia.org/T332178 (10kostajh) >>! In T332178#8698283, @Jdforrester-WMF wrote: > Is this just {T280990}? Maybe; that would be one way to solve it but...