[15:38:43] AaronSchulz: you are being paged in security [15:39:17] paged? [15:40:10] Your presence is requested [15:40:18] please pick up a white curtesy phone [16:55:38] AaronSchulz: fyi: [16:55:38] 16:41 <+thciprian> lots of this in beta: Fatal error: Object does not implement ArrayAccess in /srv/mediawiki/php-master/includes/filerepo/file/LocalFile.php on line 258 [16:56:12] AaronSchulz: do you want a task for that, or ....? [17:05:02] greg-g, I guess, not sure what that is [17:07:03] eww [17:07:13] "WP:BEANS" is a valid username on metawiki, but not enwiki [17:08:26] :) [17:09:34] getting that in vagrant just now and wasn't earlier today [17:10:00] legoktm, ... how? [17:10:14] Krenair: because WP is a namespace alias on enwiki, but not metawiki [17:10:26] it disallows namespace aliases in usernames? [17:10:41] what about full namespaces? [17:10:57] yes [17:11:10] but 'Wikipedia:BEANS' would be invalid on meta because it's an interwiki prefix [17:11:39] is there a good reason for it to disallow such names? [17:12:02] usernames should be valid on all wikis... [17:13:04] it defeats the point of global accounts if your account only works on some projects [17:13:45] so in other words no? [17:14:00] https://phabricator.wikimedia.org/T98757?workflow=create [17:14:11] we'll figure that out in like 5 years [17:14:21] greg-g, <<$cachedValues = $cache->get( $key );>> is giving the cache store object ... [17:15:28] oh, duh [17:15:28] wtf [17:19:34] AaronSchulz: they're talking about it in -operations as well (it's making it hard for people to test their patches) [17:19:57] greg-g, fix incoming [17:20:16] ty [17:22:24] greg-g, https://gerrit.wikimedia.org/r/#/c/210091/1/includes/filerepo/file/LocalFile.php [17:22:47] Thanks for the nod of confidence, but I can't review that ;) [18:16:19] bd808: [Engineering] Full MediaWiki log volume back on logstash [18:16:20] \o/ [18:16:22] awesome! [18:16:26] kudos for seeing that through. [18:57:05] anomie, hey. you around for a quick chat? [18:57:05] just wanted to clarify something from api code review [18:58:19] Krenair: Sure [18:58:34] so https://gerrit.wikimedia.org/r/#/c/209974/1/includes/api/ApiUndelete.php [18:59:01] dieUsageMsg only takes the message [18:59:04] not parameters [18:59:26] ori: thanks :) Now to get back to looking at that actual log events and figuring out which need more context data and which are just noise [19:00:11] Krenair: just use dieUsage directly... [19:00:27] dieUsageMsg takes the supposed message key, then looks it up in this stupid array ApiBase::$messageMap. When changing it to a dieUsage() call, use the code (and maybe text string) from ApiBase::$messageMap. [19:02:15] If the supposed message key isn't a constant (as is the case in ApiEditPage), you can pass it to $this->parseMsg() to look up the code and info-string. [19:02:25] Keegan|Away: all the invalid users should be taken care of now, except for ɱ which has something weird going on...I think someone did it manually and didn't update everything [19:03:13] legoktm, I am using dieUsage directly [19:03:41] anomie, you mean I should look up the message via parseMsg? [19:03:47] and just give it the code? [19:04:20] Krenair: For ApiEditPage, yes. For ApiUndelete, just use the code that you know you'd get back from parseMsg instead of calling the stupid function. [19:04:52] * anomie is going to kill all that mess one of these days [19:05:03] Apparently messageMap is Roan's fault [19:05:26] Anyway I am using the error code directly in ApiUndelete [19:05:42] and the expected message without calling parseMsg [19:05:54] Krenair: [19:05:56] 'blockedtext' => array( [19:05:56] 'code' => 'blocked', [19:05:56] 'info' => "You have been blocked from editing" [19:05:56] ), [19:06:04] The code is "blocked", not "blockedtext". [19:06:32] Oh, I see what you mean. [19:06:48] > 04:58, 11 January 2013 MBisanz (talk | contribs) renamed user ɱ to Ɱ (user request.) [19:07:56] legoktm: did they think MW was going to auto-capitalize their name? :) [19:08:34] greg-g: at some point ɱ was considered the uppercase version, and then we got a unicode update and it became lowercase and since first character lowercase is invalid, they couldn't login anymore [19:09:44] hah [19:09:52] I mean, suck, but you know [19:09:58] greg-g, legoktm: Not so much "ɱ was considered the uppercase version" as that the old Unicode treated it as caseless, IIRC. [19:10:10] ah [19:17:54] Krenair: Going to add the block info to the you-are-blocked errors from ApiBlock and ApiUnblock too? [19:19:42] anomie, I should, I probably missed those because they don't reference the messageMap keys directly [19:20:41] I searched for calls to "->isBlocked()". [19:26:18] they call into SpecialBlock which returns ipbblocked or ipbnounblockself.. hm [19:34:03] bd808: So are we sure enough yet that logstash isn't going to melt that I can bug greg-g about T1272? [19:34:52] anomie: probably should give it a few days. I just turned on the full firehose this morning [19:34:57] ok [19:35:01] but soon! [19:36:40] Fatal error: Call to undefined method User::newFromSession() in /home/alex/Development/MediaWiki/includes/context/RequestContext.php on line 270 [19:36:44] what.. [19:40:45] Fatal error: Class 'Wikimedia\Assert\Assert' not found in /home/alex/Development/MediaWiki/includes/libs/ProcessCacheLRU.php on line 132 [19:40:53] from a newer version o.O [19:41:10] the Assert lib is brought in with composer [19:41:34] ah, I need to update composer again [19:41:34] ok [19:42:47] Krenair: this git hook is kind of handy for that sort of thing -- https://phabricator.wikimedia.org/P107 [19:45:49] or you clone the wikimedia-vendor repo, and update that when you update core etc [20:07:01] anomie, what about changing dieUsageMsg to accept extra data? [20:09:02] Krenair: I want to kill dieUsageMsg, not improve it :/ [20:09:15] heh [20:09:18] okay then [20:10:34] * anomie 's todo list: Finish AuthManager. Finish the ApiSandbox rewrite. Internationalize API warnings and errors, which will include killing dieUsageMsg. [20:12:57] <3 [20:34:25] * anomie disappears slightly early today to run an errand [21:25:05] manybubbles: can search queries that exceed the maximum length not trigger an exception? there are quite a lot of these in the logs, and they aren't actionable: UsageException from line 1517 of /srv/mediawiki/php-1.26wmf4/extensions/CirrusSearch/includes/Searcher.php: Prefix search request was longer than the maximum allowed length. (1210 > 255) [21:25:35] ori: yeah - I'll open a bug. It was a silly thing to do. [21:25:50] thanks [21:25:56] i can file a task too if you prefer [21:26:09] I got it [21:27:41] ori: https://phabricator.wikimedia.org/T98786 [21:30:02] thanks [22:42:30] sup, looks like mw-config is borked in the REL1_25 branch. is it safe to install with REL1_24 then checkout? [22:42:51] PJosepherum: how is it borked? [22:43:29] installing and then switching branches won't work since you won't have the schema updates for 1.25 [22:43:47] right [22:45:31] http://hastebin.com/ixojaleyin.pas [22:45:40] apache2 error log [22:46:02] are you installing from git? [22:46:37] PJosepherum: https://www.mediawiki.org/wiki/Download_from_Git#Fetch_external_libraries [22:46:39] yes [22:46:45] ah thanks [22:48:31] PJosepherum: Other people are likely to run into similar issues, is there any docs or places you looked at first that weren't updated for the new external libraries requirement? [22:49:08] it jsut wasn't in the install guide for 124 [22:49:46] i do recall seeing this yesterday though [22:49:47] :p [23:05:11] when 1.25 is released, how simple will upgrading from 1.24 be? [23:05:20] wrt database compatibility [23:06:50] PJosepherum: looks like only 2 schema changes: https://github.com/wikimedia/mediawiki/blob/REL1_25/includes/installer/MysqlUpdater.php#L269 [23:10:34] of course, if i dump and import via xml i can avoid this? [23:14:11] that would probably be even slower? [23:14:31] Keegan|Away: were you going to leave ~~-~nlwiki a custom message or should I give them the stock one? [23:49:43] yeah i guess it would be [23:50:15] basically i've done a fresh install of mw1.25rc0 with the intention of migrating our 1.24.2 database once it's released