[00:16:56] is 12.339999999999999857891452847979962825775146484375 our new version numbering scheme? [00:17:19] sorry, php-12.339999999999999857891452847979962825775146484375 [01:33:44] anomie: would you mind if I camelcased AuthenticationRequest field names? HTMLForm uses that convention and it would be nice if they matched up [13:17:56] tgr|away: It'll have the drawback that API login will have camel-cased parameters in contradiction to the usual convention for the API, but go ahead. Rewrite the whole AuthenticationRequest thing if you have a better way to do it. [15:23:04] Why does my dev wiki not have any instant common thumbs? :( [15:23:46] !debug | Reedy [15:23:59] Amusingly, the file pages work fine [15:24:06] And any embedded thumbs in those [15:26:31] ori: easy mw-vagrant review I think -- https://gerrit.wikimedia.org/r/#/c/246254/ [15:40:10] wtf mw [15:40:14] the thumb exists on disk [15:40:18] http://10.0.0.190/w/images/thumb/e/e6/Mantelaffe-mit-Baby-Zoo-Muenster.jpg/793px-Mantelaffe-mit-Baby-Zoo-Muenster.jpg [15:40:22] Says src doesn't exist [15:41:02] ah, thumb.php rewrites [15:45:01] This feels like a bug [15:50:06] Reedy: unpossible! there are no bugs in the thumbs pipeline. Most solid code we have :/ [15:50:41] https://phabricator.wikimedia.org/T115619 [15:50:42] Filed [15:52:12] .... that error message doesn't make any sense [15:52:22] for a completely different image? [15:53:10] Oh [15:53:12] Ignore that [15:53:15] I was changing stuff around [15:55:24] I was playing with a few different files, different copies etc [15:55:35] *nod* [15:57:11] Works fine for local images etc [15:58:15] there were changes in instantcommons because of https, but I'm not too sure what they were really [15:58:32] tgr spent some time making things less sucky there as I recall [15:58:36] heh [15:58:41] It seems to be pulling the file locally etc [15:59:22] Sizes look sane [15:59:35] The other file name was a new file I tried to check pulling was working [16:00:03] The original and the thumbs are in the thumb dir [16:00:07] This seems a bit strange [16:00:16] https://commons.wikimedia.org/wiki/Mantelaffe-mit-Baby-Zoo-Muenster.jpg is a redlink [16:00:27] https://commons.wikimedia.org/wiki/File:Mantelaffe-mit-Baby-Zoo-Muenster.jpg [16:00:37] herp derp. thanks [16:00:56] hdh [16:00:58] heh [16:01:39] Hmm, is it thumb vs thumb_handler? [16:01:57] I'm getting a 404 with vagrant too [16:11:43] * Reedy updates some comments [16:16:00] * bd808 decides this is a great way to test https://gerrit.wikimedia.org/r/#/c/234811/ [16:16:32] Sounds like a good idea to me [16:18:01] the broken looking bit in the logs is "MediaHandler::getHandler: no handler found for ." [16:25:17] Interestingly, mine doesn't work where the thumb image switches [16:25:21] * Reedy updates his apache config [16:27:55] That fixes the images where they exist already [16:35:03] * bd808 futzes with his logging settings [16:53:29] * bd808 is starting to wonder if this is a cache problem [16:58:30] Reedy: I'm lost in a tangle maze of passages since I don't really know how this is supposed to work. What I can see is that thumb.php decides this is a local file but there is no data in the image table for it. Not sure which part of that is wrong. [16:59:43] Reedy: Add a release note to https://gerrit.wikimedia.org/r/#/c/246706/ and I'll +2 it. [17:02:26] anomie: done :) [17:02:52] bd808: heh. Thanks for chasing it... I suspect AaronSchulz might have a better idea [17:03:31] I was looking at moving it to EP, and was like, it's not used anywhere... [17:03:32] DELETE [17:04:42] class ApiListStudents extends ApiBase { [17:04:44] * Reedy blinks [17:06:55] * Reedy ignores [17:07:41] ? [17:08:19] I was looking at the EducationProgram code [17:08:27] DELETE [17:11:16] Gonna try and move the ORM code out of core this evening [17:11:45] \o/ [17:12:02] I was trying to decide if doing via an IDE might be necessary [17:12:24] Probably gonna be easier, NS it at the same time [17:12:45] Reedy, do it via combustion [17:13:18] If we break EP fully, people will get sad [17:13:49] like, 3 of them [17:14:09] More wikis are asking for it to be installed [17:14:45] If you move it out of core, please keep in mind that Wikibase only branches every second week (although we *might* branch on Tuesday) [17:14:59] The ORM stuff is gone on master, but not yet on the wmf branch [17:15:21] Yeah, merging both commits was going to be blocked on knowing you guys don't need it in branch [17:16:00] Gonna take some effort to get it done anyway... :P [17:16:27] * hoo wonders how hard it would be to migrate EP as well [17:16:39] If it's build all around the ORM stuff... well fuck [17:16:52] I don't think everything is, but almost everything [17:17:01] Stop Jereon from doing anything else till he fixes EP [17:17:02] xD [17:25:01] Reedy: if the wiki is trying to access an InstantCommons file via thumb.php you have already lost [17:25:22] the thumb should be there and so the 404 handler should never be invoked [17:25:42] lol [17:27:04] the relevant bug is https://phabricator.wikimedia.org/T27958 but normally you should not run into it because the thumbnails are pregenerated [17:27:35] the bug is with asking for a new thumb in a different size at this point [17:27:59] and the really relevant bug is https://phabricator.wikimedia.org/T56202 [17:28:47] at least it used to be, I'm not sure if that still exists [17:31:06] hm, I think we fixed it in https://gerrit.wikimedia.org/r/#/c/223477/ and just forgot to close the bug? [17:31:10] I'll retest [17:31:21] Reedy, what settings were you using? [17:32:23] for which part? [18:11:17] does http://fpaste.org/279665/93264714/raw/ seem useful as a library? [18:11:30] (for all the other preg_* functions too) [18:32:19] legoktm: we certainly have a lot of preg_* calls that don't check for that sort of failure [18:33:09] yeah, I found code that does (bool)preg_match [18:33:22] and the preg_ functions are inconsistent, some return null as an error, others return false [18:33:35] * bd808 acts shocked [18:35:03] Something in PHP being inconsistent? I'm outright shocked! [18:35:34] hehe [18:35:52] I'll take that as a yes, that this library will be useful :) [18:46:26] I filed https://phabricator.wikimedia.org/T115639#1728864 for creating it [18:47:18] legoktm, add failed regexes to exceptions? [18:54:08] yeah, that's probably a good idea [18:54:22] we'll also have the warnings generated by the preg_ functions themselves [19:13:18] anomie: is it a bad idea for me to use AuthPlugin::getUserInstance() / AuthPluginUser::getId() in new code? I want to store global user ids for Echo without hardcoding CA stuff... [19:14:09] legoktm: You [19:14:23] legoktm: You'll have to fix it then when AuthManager is ready. [19:15:53] ok, I don't mind doing that [19:17:00] legoktm: It might be better to figure out if T111302 really belongs in AuthManager, or if it would be better done by moving the three functions OAuth uses for that purpose to somewhere else in core. [19:19:35] I don't know what there's a "$providerId" parameter in the task. In AuthManager it would go through each PrimaryAuthenticationProvider until one returns an ID, which would make sense in AuthManager, except the central ID could change if an earlier provider suddenly gains knowledge of the user. [19:20:42] So it might actually be better to have one specific central ID source configured. In which case it doesn't really have anything to do with AuthManager anymore. [19:28:19] 19:23:38 1) AutoLoaderTest::testAutoLoadConfig [19:28:20] 19:23:38 Failed asserting that two arrays are equal. [19:28:26] How do I fix that in an extension? [19:28:53] https://integration.wikimedia.org/ci/job/mwext-testextension-zend/12316/console [19:30:02] * anomie copies all that into T111302 [19:30:22] $wgHooks['UnitTestsList'][] = 'EducationProgram\Hooks::registerUnitTests'; [19:30:25] * Reedy squints [19:30:27] Reedy: you need to add those classes into $wgAutoloadClasses [19:30:31] 19:23:38 + 'EducationProgram\Tests\TestORMRowTest' => '/mnt/jenkins-workspace/worksp...st.php' [19:30:31] 19:23:38 + 'EducationProgram\Tests\TestORMTable' => '/mnt/jenkins-workspace/worksp...st.php' [19:30:45] Reedy: Somehow or other it's deciding that those classes should be in $wgAutoloadClasses. [19:30:59] I wonder how they worked in core [19:31:05] It didn't have them all [19:31:07] * Reedy adds [19:31:08] thanks [19:31:15] maybe the namespacing is throwing off the test? [19:33:49] Reedy: Looking at the unit test, it looks like maybe if those classes are in the same file as any class that is in $wgAutoloadClasses then it'll do that. [19:35:32] Woo, test passes [19:43:12] hoo/aude do you have a task for your branching process? [19:44:28] You mean regarding when we branch? [19:45:20] Well, do you have a task for blockers for the deployment etc [19:45:47] Doesn't matter if you don't... [19:45:48] Oh... no :/ [19:46:14] Oh well :) [19:46:40] I'm sure someone will merge it soon after you merge :P [20:22:58] ... That was unfun. A unit test started failing, complaining about ext.uls.something even though I don't have ULS installed locally. Turns out that the composer.local.json I had to add so extensions wouldn't break due to missing composer crap made composer decide that it had to install and load ULS from the autoloading. Ugh. Then I remove ULS entirely from my extensions dir, and composer reinstalls it thanks to Translate's composer.json. WTF? [20:25:38] why would you want Translate in the first place? :P [20:28:01] I had it in my extensions directory to look at its API modules at one point. If there's composer.local.json magic to only load installed extensions instead of extensions/*/composer.json, I don't know it. [20:32:55] you can manually list the extensions to install [20:33:06] instead of a wildcard [20:33:19] "include": [ [20:33:19] "extensions/SyntaxHighlight_GeSHi/composer.json", [20:33:19] "extensions/Elastica/composer.json" [20:33:19] ] [20:33:24] Then I need to remember to update it when I turn extensions on and off [20:33:41] true [20:34:31] * anomie has a complicated LocalSettings.php that actually detects if core unit tests are running and skips loading extensions [20:39:55] * anomie wishes he could figure out why the 'Magic Word: {{NUMBEROFFILES}}' parser test doesn't pass locally. [20:42:12] theoretically having extra libraries installed shouldn't interfere with anything [20:45:39] Hmm, using mw-collapsible and the [Expand] link is nowhere near the associated text box [20:45:48] * Reedy considers filing a bug for that [20:47:20] Probably my images