[04:14:55] what's up with all the NO_FUNCTION_GIVEN() in error stack traces? is that some HHVM quirk? [11:09:24] tgr|away: hmm if memory serves that's usually an OOM where the preserved padding didn't suffice [11:09:34] Or stack limit exceeds [11:09:57] Is there a spike? [14:04:10] Anyone knows how to validate my extension.json locally? [14:52:29] stephanebisson: There's a maintenance script. [14:53:21] James_F: Thanks. I found `tests/phpunit/structure/ExtensionJsonValidationTest.php` [14:54:07] I can't get ResourceLoaderImageModule to work with manifest_version: 2. I mean, it works, but it doesn't validate :( [14:55:11] stephanebisson: Oh, yes, there's that too, but ValidateRegistrationFile is the one of which I was thinking. [16:39:25] James_F RoanKattouw: Hey, for when you have a minute: https://gerrit.wikimedia.org/r/#/c/479266/ [16:57:05] Krinkle: don't think so, it was just my first time seeing it (and I searched Phabricator and there were lots of tasks with such backtraces from the last couple months) [17:00:20] stephanebisson: I don't think the manifest version affects anything other than config/attributes [17:25:09] Reedy wondering if you could +2 https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/ORES/+/480544/ please? :) [17:25:14] should fix PageTriage [17:25:22] see https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/PageTriage/+/479995/ [17:43:11] tgr: Indeed. I figured it out: I had an extra 'target' attribute but because of the way the schema is designed (with anyOf) it was complaining that all other attributes were unexpected. [17:58:44] Reedy: Found this: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/480554/1/maintenance/mssql/archives/patch-user_groups-ug_expiry.sql [17:58:46] :)))) [17:59:18] Ouch [18:00:21] Amir1: I'll CR+2 the patch... I wonder if we should file a bug for followup/cleanup for mssql [18:02:00] a general clean up you mean? to check if similar mistakes happened or not? or you mean the change_tag patch? [18:03:39] I mean to remove that column/constraint from that table on mssql if they exist [18:04:02] I've just backported and RELEASE-NOTE'd it in 1.30-1.32 [18:05:18] Though.. I'm guessing the patch would actually fail [18:05:24] ALTER TABLE /*_*/user_groups ADD CONSTRAINT pk_user_groups PRIMARY KEY(ug_user, ug_group); [18:05:32] uh [18:05:32] ALTER TABLE /*_*/tag_summary ADD CONSTRAINT pk_user_groups PRIMARY KEY(ug_user, ug_group); [18:05:44] Can't add a PK on fields that don't exist... [18:05:55] Not sure if the command after would still run [18:08:45] yeah, the fact that no one complained after three releases is a sign that we probably should drop support for mssql until we have proper abstraction to build such SQL commands [18:08:50] My two cents [18:08:58] https://phabricator.wikimedia.org/T212241 [18:09:02] heh [18:10:54] I'll poke a mssql box later/tomorrow and see how it behaves [18:11:16] Reedy: do you want me to +2 the backports? [18:11:24] I don't know if I need to anything beside that [18:11:27] Can do, yeah :) [18:11:36] Let you sanity check my release-notes additions [18:13:19] tgr: Thanks for digging into the OATHAuth logging issue <£ [18:13:21] <3 [18:25:16] @Reedy thanks for your comments on the T202110 patch. I'm working on the callback fix right now (T212175) and will circle back to the other task afterward. [18:25:16] T202110: Create GuzzleHttpRequest class as new default for HttpRequestFactory - https://phabricator.wikimedia.org/T202110 [18:25:17] T212175: Support callback functions in GuzzleHttpRequest and MultiHttpClient - https://phabricator.wikimedia.org/T212175 [18:25:45] bpirkle: :) Like I say, it's not wrong, but I don't think it's needed [18:26:43] It also means people can pick up bug fix sub versions without needing to change mw's composer.json/wait for a release [18:26:57] Deleting unnecessary code is one of the best parts of the job! [18:41:07] Reedy: for when you have a minute: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/479266 [18:41:27] one of the patches you mentioned actually has lots of things. I will fix that in a follow up [18:42:05] Cool. Will have a look :) [18:45:58] Thanks [18:49:29] Amir1: It's very easy to fix in this one acttually [18:59:02] Reedy: this very likely fail on tests: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/479266/3..4 [18:59:11] because the other indexes got dropped too [18:59:36] let's wait and see, if it works, I'm happy [19:00:52] heh [19:02:13] but it needs the indexes to run the migration script [19:02:15] :/ [19:02:53] https://github.com/wikimedia/mediawiki/blob/master/maintenance/archives/patch-change_tag-change_tag_rc_tag_id.sql [19:03:14] Let's fix it [19:04:47] Just swapped it to do a dropIndex and look for the index it's trying to drop first [19:05:50] that's what I was thinking too [19:09:52] Or we could use addIndexIfNoneExist [19:10:02] This probably works fine though [19:18:32] Reedy: \o/ [20:43:12] <_joe_> tgr: thanks for pointing me in the direction of the MainStash hornet's nest [20:44:23] <_joe_> (I'm sincerely thanking you, although I was happier not realizing we introduced an interface that promises arbitrary k-v multi-dc active/active document storage) [20:44:53] <_joe_> and that we fail completely with our current storage for it to fullfill any on such promises, even [20:49:08] yeah, that was a bit of a surprise [20:51:49] not sure if any of the callers actually need that promise... [21:08:07] <_joe_> I'm pretty sure they don't [21:08:12] <_joe_> else we would've noticed