[10:12:14] Amir1: Hey! About T277362... I'm trying to better understand how much trouble this caused for Wikidata/Wikibase teams. Did this break CI for any repo? Could it have been caught in CI? [10:12:14] Cleaning up core means deprecating many things, and I'd like to avoid this kind of situation in the future... [10:12:15] T277362: Deprecation warning client-repo wikitext link - https://phabricator.wikimedia.org/T277362 [11:16:01] duesen: hey, sorry I'm drowning in meetings today but I think deprecation warnings supposed to fail tests. Maybe got disabled in the flood of warnings? [11:16:43] if not, I can't think of any easy way to catch this, unless with some info level logging, statsd metrics? etc. [11:21:17] you can send it to another channel beside mediawiki in logstash [14:53:53] Amir1: I think the critical case doesn't have tests... [14:53:53] And yes, putting deprecation warnings into a separate channel would be good. But that needs some alignment, I think [20:05:10] Amir1, hoo: can you verify that this no longer causes issues in Wikibase? https://gerrit.wikimedia.org/r/c/mediawiki/core/+/672819/3 [20:05:18] A comment on the patch would be helpful. [20:07:58] Yes, will do :) [20:14:19] Thank you! [20:37:38] duesen: +2ed [20:37:43] I'll call it a day [20:37:45] See you [20:38:08] hoo: thank you! [21:53:48] TimStarling: did you bump composer for the timeout lib? [21:56:16] there's no tagged release beyond 1.0.0, so presumably not [22:26:57] should be 1.1.0 I guess [22:35:32] no comments on my @noDebugInfo patch? I guess nobody else uses var_dump()? [22:35:52] https://gerrit.wikimedia.org/r/c/mediawiki/core/+/672845/ [22:36:31] I'm open to suggestions on the names of things [22:42:51] I wonder if this might be useful more widely, and as such, would be better put in a library from the beginning [22:46:39] yeah, probably. ObjectFactory should probably use it but it's in a library already [22:47:55] I read it yesterday, but was more concerned how User (theoretically a value object) is linked to the entire MediaWikiServices but didn't finish looking into it [22:48:14] the reference chain is in the task [22:51:38] it's due to UserAuthority, before UserAuthority it was possible to dump User objects [22:52:35] I didn't realize Authority was actually in use already, I thought that was going through the new-RfC/decision process still [22:53:49] Title doesn't have this problem because it doesn't store references to the services it uses [22:54:02] UserAuthority also caused problems with serialization of User objects [22:56:05] there's UserAuthority but not Authority, here is the Authority WIP: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/620112 [23:06:19] left a review [23:12:47] thanks [23:16:06] var_dump() does have some advantages over a userspace implementation, for example you get cheap recursion protection via a bit in the zval, a feature you can't access any other way [23:19:03] I assume there's some bootstrapping issue at play, but I think the LanguageConverterFactory chain can be broken by having the closure hold onto LanguageFactory and MainConfig directly instead of all of MediaWikiServices