[06:18:17] <_joe_> (I don't think being able to run mediawiki without the curl extension is something we should encourage, but well) [14:13:18] Krinkle: Go for it. Why would I mind? [14:40:12] Krinkle: connection timeout is supported only by the curl handler: http://docs.guzzlephp.org/en/stable/request-options.html?highlight=connect_timeout#connect-timeout [14:52:51] Who knows how the page at https://doc.wikimedia.org/ gets generated? There's no link from there to https://doc.wikimedia.org/Parsoid-PHP/master/. [14:54:00] <_joe_> anomie: hashar would know [14:54:02] anomie: https://github.com/wikimedia/integration-docroot [14:54:09] <_joe_> or Reedy heh [14:54:10] Not sure if it's generated, or just committed to that repo [14:54:24] https://github.com/wikimedia/integration-docroot/blob/master/org/wikimedia/doc/default.html [14:54:48] I will admit I have nfi how to deploy it without looking :P [14:58:24] Reedy: Thanks [15:20:50] Reedy: anomie: Deployment for that repo is mostly automatic by post-merge (for integration.wm.o pages). For doc.wikimedia.org pages it's ssh + git pull. The post-merge hook leaves a comment saying how to do it. (Access is limited to contint-admins, which should collerate with the ones that have +2 there) [15:55:13] bpirkle: Good morning. I am going through the unit testing docs and am trying to run the databaseless unit tests. But am getting "LoadBalancer.php: Cannot access the database: Unknown error (127.0.0.1)". Is the test group broken? [16:07:48] > /srv/mediawiki/php-1.33.0-wmf.20/includes/libs/HttpStatus.php: PHP Warning: Unknown HTTP status code process-wikitext [16:08:26] kostajh: Seeing this in Logstash with refers relating to you doing tests, I think. [16:08:53] ApiMain->handleException(Flow\Exception\WikitextException) [16:09:01] Krinkle: yes [16:09:18] for T217766 [16:09:19] T217766: Flow\Exception\WikitextException: ParseEntityRef: no name - https://phabricator.wikimedia.org/T217766 [16:09:40] k, so this errror is a side effect? [16:10:13] Krinkle: can you share a link to the log message you're looking at? [16:10:32] kostajh: https://logstash.wikimedia.org/app/kibana#/doc/logstash-*/logstash-2019.03.06/mediawiki?id=AWlTtUFVZ9NltGfJtOLH&_g=h@b74aee6 [16:12:29] Krinkle: thanks. I haven't dug into a lot of Flow's exception handling, but I'm surprised there haven't been other invalid HTTP status codes in the past, after a quick look at extensions/Flow/includes/Exception/ExceptionHandling.php [16:14:50] kostajh: interesting, yeah, http status codes and getErrorCode() there seem at odds, those look like i18n message keys, not http status code numbers. [16:14:52] anyway :) [16:14:58] just fyi, I'll ignore for now [16:41:25] hknust: hmmm, I haven't seen that, and I frequently run unit tests locally. I'm using the mediawiki vagrant box, so my setup is probably much different from yours. How are you invoking the tests? [16:42:10] make databaseless as described on https://www.mediawiki.org/wiki/Manual:PHP_unit_testing/Running_the_unit_tests [16:44:49] subbu: Shopping around CR for https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/294774/ - would this fit with the Parsing team and have time for it in coming week or two? [16:45:36] subbu: Main tricky bit is estimating and understanding impact based on current usage, we were hoping you have some prior experience with that to share, w.r.t. parsoid and Linter undertakings. [16:48:06] Krinkle, in the middle of something .. will get back to you in a little bit. [16:55:24] hknust: I get the same error as you when I execute "make databaseless" from my base OS, but it works okay when I execute it from within the vagrant box. I'm not sure how that corresponds to the way your local dev is set up, but if you're executing from your base OS, is there a way you can execute the command from within the box/container/whatever you're using? [16:56:55] Krinkle, do you know current usage of that magic word at all? in that case, a source search should reveal all usages instead of requiring linting to expose them, correct? and also is miser mode enabled in production? [16:59:48] subbu: Yes, miser mode is basically alias for WMF prod medium+big wikis. [17:00:00] i see. ok. [17:00:09] subbu: as far as we could see, it's only used in templates to detect preview, e.g. comparing to empty string. [17:00:15] not as actual integer. [17:02:17] i see .. that seems like missing functionality that editors are hacking around .. but, that apart .. seems reasonable at first glance. [17:03:51] I am going to +1 for right now and can followup in the next week or two. [17:04:05] but talking of lua modules ... we ran into another scenario where lua modules are hacking with nowiki tag. [17:04:40] T203293 [17:04:41] T203293: {{Row numbers}} completely fails on the Android app / VisualEditor - https://phabricator.wikimedia.org/T203293 [17:44:55] Krinkle: Just out of curiousity, did you merge https://gerrit.wikimedia.org/r/c/mediawiki/core/+/491539 because you thought it was a good idea on its own, or just because you thought it was still needed to resolve T216086? [17:44:55] T216086: ErrorException from includes/HeaderCallback.php: PHP Notice: Undefined offset: 1 - https://phabricator.wikimedia.org/T216086 [17:45:56] anomie: both. The code cannot know what exists, given PHP headers_list doesn't guruantee it. We fixed one instance, but could come back afaik. [17:46:52] OTOH, if it comes back that would point out another bug to fix. Now we won't see it. [17:51:28] Extracting cookies from headers_list() != validation. [17:51:39] Validation sounds good. [19:13:33] Reedy: Urbanecm, zelkof and myself tried to deploy https://gerrit.wikimedia.org/r/#/c/operations/mediawiki-config/+/494806/ today but when testing it didn't worked. Should we touch InitialiseSettings and or CommonSettings next time? [19:41:21] anomie: bd808: hi, thank you both for the patches on LdapAuthentication :) [19:41:43] going to deploy it [20:16:21] random though of the day: do we have people acting on the api-feature-usage deprecations? [20:16:36] eg filling tasks, reaching out to bot authors/owners etc [21:11:39] hashar: Not that I know of. It'd be nice if someone did. [21:13:27] anomie: maybe we can handle that as log spam and via https://phabricator.wikimedia.org/tag/wikimedia-production-error/ [21:13:43] or well. Something similar [21:14:04] really just get a phab project, fill tasks and cc folks ;) [21:14:23] I felt that whoever added the tracking was also acting on getting them fixed [21:18:01] I added the tracking, back in 2014, but mainly so I could better see how much people were still using deprecated stuff when I was considering removing it. [22:12:15] Is $wgUser deprecated yet? Is there a better way to get a user object? Maybe via request context? [22:15:45] https://www.mediawiki.org/wiki/Manual:User.php#See_also currently says it is but not in the code [23:01:11] anomie: should I backport activeusers? [23:02:14] AaronSchulz: Probably wouldn't hurt. [23:07:31] xSavitar: the request context is slightly better than the global variable. there is currently no "good" way, really.