[02:16:59] (03PS1) 10Aude: Add @group WikibaseRepo, cleanup repo test headers [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102374 [02:19:55] wow [02:20:36] massive change :D [02:22:36] (03CR) 10Hoo man: [C: 032] Add @group WikibaseRepo, cleanup repo test headers [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102374 (owner: 10Aude) [02:22:48] aude: Thanks for bringing these in line :) [02:23:10] not being able to use --group WikibaseRepo sucked [02:23:27] yeah [02:23:52] now i'm trying exclude repo and repo only [02:23:57] good night ;) [02:24:21] :) [02:26:19] (03Merged) 10jenkins-bot: Add @group WikibaseRepo, cleanup repo test headers [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102374 (owner: 10Aude) [02:34:56] (03PS1) 10Aude: Don't run OutputPageParserOutput client hook during tests [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102376 [02:42:49] [travis-ci] wikimedia/mediawiki-extensions-Wikibase#1443 (master - 58c1974 : aude): The build is still failing. [02:42:49] [travis-ci] Change view : https://github.com/wikimedia/mediawiki-extensions-Wikibase/compare/8daa3c6651af...58c1974e61d0 [02:42:49] [travis-ci] Build details : http://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/builds/15622719 [07:54:28] (03CR) 10Siebrand: [C: 031] "i18n/L10n reviewed." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/94939 (owner: 10Bene) [07:59:54] (03CR) 10Siebrand: [C: 031] "i18n/L10n reviewed." [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98281 (owner: 10Bene) [09:15:02] (03PS1) 10Aude: Add messages to EntityContentTest assertions [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102407 [10:28:44] (03CR) 10Tobias Gritschacher: [C: 04-1] "Author:: is missing in the documentation headers. see other files for examples." (034 comments) [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100591 (owner: 10Vldandrew) [11:03:40] aude: Scribunto has been added as a dependency to Wikibase jobs ( https://gerrit.wikimedia.org/r/#/c/102109/ ) [11:29:00] thanks hashar :) [11:29:47] (03PS2) 10Daniel Kinzler: Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 [11:30:32] hashar: any idea what happened here? https://integration.wikimedia.org/ci/job/mwext-Wikibase-client-tests/5261/console [11:30:34] DanielK_WMDE: Tobi_WMDE https://travis-ci.org/filbertkm/Wikibase/builds/15637680 passed with core wmf/1.23wmf5 + master of all our stuff [11:30:48] * aude wonders if there is a core regression related to saving content [11:31:16] DanielK_WMDE: i have seen that or similar so something is wrong [11:31:38] aude: what exactly is the problem you are investigating? [11:31:43] travis failures [11:31:49] which ones? [11:32:03] lots of various stuff but tends to be about saving, like in the api or property content [11:32:34] DanielK_WMDE: how about this one https://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/jobs/15622720 [11:32:46] (03CR) 10jenkins-bot: [V: 04-1] Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 (owner: 10Daniel Kinzler) [11:33:39] shall try with wmf6 now [11:34:24] wtf! [11:34:44] refresh the page and i get the property content failure on wmf5 [11:35:04] what, on your local install? [11:35:21] travis [11:35:31] local install never fails [11:36:46] ok, i made a mistake in .travis.yml so it wasn't using wmf5 [11:37:20] hashar: ok, so i know why that thing failed - but i don't know why it's not showing the actual error :) [11:38:33] (03PS3) 10Daniel Kinzler: Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 [11:38:35] moin andre__ [11:38:42] heja [11:41:17] (03CR) 10jenkins-bot: [V: 04-1] Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 (owner: 10Daniel Kinzler) [11:41:26] *sigh* [11:41:38] jenkins is fast [11:41:41] faster* [11:41:49] yea. but it doesn't tell me what's going wrong. [11:42:24] i also have a mystery failure locally. have had it for weeks now. not rep4roducible anywhere else. no output from phpunit. [11:42:31] bah [11:42:35] i think this one is new [11:42:36] this sucks, i'm blind. [11:42:53] i can't figure out exactly what is causing it and it doesn't always happen [11:43:10] and think it's unrelated to your patch [11:43:12] well, i did find an actual issue with the patch when running client tests alone. but i fixed that. now it's still failing... for *some* reason [11:43:19] oh [11:44:53] https://gerrit.wikimedia.org/r/#/c/102301/2..3/client/tests/phpunit/includes/WikibaseClientTest.php ? [11:44:58] suppose that is new [11:45:16] that's the fix i just pushed, yes [11:45:30] was it something i did in one of my patches? [11:45:45] hm? no. [11:45:49] ok [11:45:58] then i think there is another issue additionally [11:46:16] yes, it looks like several unrelated issues, and broken reporting [11:46:23] * aude nods [11:46:44] so. locally, phpunit stops when running RepoItemLinkGeneratorTest [11:46:51] could be it [11:46:53] if i run it alone, it works [11:47:12] if i run all the client tests, phpunit just stops there. no error, no nothing [11:48:18] oh! [11:48:38] with --filter 'WikibaseClientTest|RepoItemLinkGeneratorTest', i do get to see some errors. [11:48:42] lucky shot, that [11:48:43] that helpsoooh [11:48:48] gah [11:49:06] but again, my fault, an error on the patch [11:49:08] not in master [11:49:32] uh... why the hell does this work when running the test alone?.... [11:50:30] shall look in a minute [11:50:48] aude: naw, i'm on it, you keep poking at the travis issue [11:50:51] that's unrelated [11:50:55] it is a travis issue [11:50:59] i saw it on travis [11:51:25] the error i'm seeing now is caused by the patch. i renamed Wikibase\Lib\EntityIdParser. [11:51:39] forgot to update a few use-statements, it seems [11:51:54] phpunit is just good at hiding the errors >_< [11:52:00] ok, i see nothing in git log that would cause a problem in the test [11:52:02] that test [11:52:28] it means a typehint error probably [11:53:04] yes, it does. the error was trivial. just hard to find :) [11:53:23] (03PS4) 10Daniel Kinzler: Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 [11:53:25] there [11:55:24] k [11:56:08] (03CR) 10jenkins-bot: [V: 04-1] Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 (owner: 10Daniel Kinzler) [11:57:44] aude: hypothesis: something changed about how groups or permissions are cached in core. so save fails randomly, depending on what tests messed with the permissions. [11:57:59] the unit tests would bneed to be updated to reset the appropriate cache [11:59:26] could be [12:00:11] https://travis-ci.org/filbertkm/Wikibase/builds/15641236 passes... pretty sure it's wmf5 core here [12:01:15] https://travis-ci.org/filbertkm/Wikibase/jobs/15621444 baffles me. despite all the failures, i've never seen so many in one test run as here [12:01:32] the other "builds" in the same job all passed [12:04:27] aude: and they all seem related to permissions / saving pages [12:04:31] yep [12:04:50] i will try with wmf6 now [12:06:02] aude: i'd suggest to hackin some extra code into one of the test cases that fails most often. When it fails unexpectedly, grab all info about the user, groups, and permissions, and output them. [12:06:14] i'd hope that that would give us *some* hint about what is going wrong [12:06:28] but it's extremely random which test fails [12:06:53] i think pinpointing to a certain commit in core might be better [12:06:58] aude: PermissionsTest seems to fail quite often [12:07:05] not always [12:07:11] not that often [12:07:17] no, but often enough to get useful output [12:11:47] aude: Namespace::getRestrictionLevels is new [12:13:23] (03PS13) 10Vldandrew: Migrate simple RSpec browser tests to Cucumber. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100591 [12:14:59] where? [12:15:40] In Namespace.php [12:16:05] * aude looks [12:16:05] that's the only thing obviously relevant in the diff between wmf5 and wmf6 [12:16:18] aude: i don't see how that's causing a problem, though [12:16:38] https://travis-ci.org/filbertkm/Wikibase/builds/15641656 here has wmf5 and enabled our api tests [12:16:41] 2 failures [12:16:44] hm, i'm seeing this in a failure of PropertyContentTest: [12:16:53] Failed asserting that '

(permissionserrorstext-withaction: 1, (action-read)) [12:16:55] ... [12:17:06]

(badaccess-group0) [12:17:10] group0?? [12:17:18] what the hell is group0? [12:17:20] confused [12:18:51] oh, there's actually a message called badaccess-group0. whatever thast is supposed to mean. [12:18:59] huh [12:20:43] so, at some point, for some reason, the read permission (and also the write permission) is lost, and all access fails [12:20:53] odd [12:22:45] Unicode Normalization Test Suite ? [12:23:03] ok...... [12:23:22] we can probably exclude it [12:26:43] DanielK_WMDE: what i'll do is make a new non-voting test job with scribunto... otherwise i don't quite understand and they don't use @group [12:32:39] aude: how is scribunto related to the travis failures? [12:32:47] it's not [12:32:54] i'm talking about jenkins [12:33:47] ? [12:34:19] (03PS7) 10Zfilipin: Use snake case [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 (owner: 10Aaron1011) [12:35:36] oh, i see ... site link table [12:36:07] aude: yea. got a fix, will submit in a minute [12:36:13] k [12:36:26] i was looking at the normalization stuff but that gets skipped [12:36:42] aude: what i wonder is how that test got into master without failibng [12:36:55] i ran the tests! [12:37:11] (03CR) 10jenkins-bot: [V: 04-1] Use snake_case for methods and variables [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 (owner: 10Aaron1011) [12:37:12] didn't run into that [12:39:51] (03CR) 10Tobias Gritschacher: [C: 031] "this looks fine now. however I do not want to merge it right away, there is still an issue with the selenium test user on beta not having " [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100591 (owner: 10Vldandrew) [12:42:36] (03PS1) 10Daniel Kinzler: Skip LuaWikibaseLibraryTest if repo not present [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102430 [12:43:11] aude: well, chances are, it will just find your local repo database and run the test against it. [12:43:19] it will only fail if that db really isn't there [12:43:22] true [12:43:40] (03CR) 10Aude: [C: 032] Skip LuaWikibaseLibraryTest if repo not present [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102430 (owner: 10Daniel Kinzler) [12:44:04] it will run in allthestuffs build :) so ok [12:44:14] yes [12:44:25] we have quite a few tests like that [12:44:30] yep [12:48:29] (03Merged) 10jenkins-bot: Skip LuaWikibaseLibraryTest if repo not present [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102430 (owner: 10Daniel Kinzler) [12:52:58] (03PS5) 10Daniel Kinzler: Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 [12:54:34] (03PS1) 10Zfilipin: Updated Ruby gems [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102431 [13:05:22] [travis-ci] wikimedia/mediawiki-extensions-Wikibase#1444 (master - 6749856 : jenkins-bot): The build is still failing. [13:05:22] [travis-ci] Change view : https://github.com/wikimedia/mediawiki-extensions-Wikibase/compare/58c1974e61d0...674985624206 [13:05:22] [travis-ci] Build details : http://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/builds/15643836 [13:06:41] gah [13:34:20] (03Draft6) 10Henning Snater: Implemented basic front-end value formatters [extensions/DataValues] - 10https://gerrit.wikimedia.org/r/101876 [13:34:25] (03Draft2) 10Henning Snater: Implemented front-end ValueFormatterFactory [extensions/DataValues] - 10https://gerrit.wikimedia.org/r/102089 [13:49:28] (03Draft2) 10Henning Snater: Implemented ApiBasedValueFormatter [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102404 [13:50:13] (03Draft2) 10Henning Snater: Implemented front-end QuantityFormatter [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102406 [13:50:46] (03Draft1) 10Henning Snater: Make use of valueFormatterFactory in $.valueview [extensions/DataValues] - 10https://gerrit.wikimedia.org/r/102421 [13:51:45] (03Draft2) 10Henning Snater: Use formatted value in QuantityType expert [extensions/DataValues] - 10https://gerrit.wikimedia.org/r/102422 [16:01:23] (03CR) 10Jeroen De Dauw: [C: 032] Use correct EntityIdParser interface and implementations. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102301 (owner: 10Daniel Kinzler) [16:15:57] why did we decide to have an item for Special:RecentChanges? [16:20:20] did we? :O [16:23:56] there are like a bajillion lang links when I visit it on enwiki [16:34:20] [travis-ci] wikimedia/mediawiki-extensions-Wikibase#1445 (master - 80edaca : daniel): The build is still failing. [16:34:20] [travis-ci] Change view : https://github.com/wikimedia/mediawiki-extensions-Wikibase/compare/674985624206...80edaca54dab [16:34:20] [travis-ci] Build details : http://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/builds/15654515 [17:00:18] legoktm: lol; what does WD:N say about special pages? [17:00:43] To be valid, a link must not be a talk page; MediaWiki page; special page; user page; file; translations page; or subpage of a template, portal, or module.[1][2][3] Note that a single Wikimedia page cannot have more than one sitelink in Wikidata and that a sitelink cannot point to a redirect.[4] [17:00:46] hmmmm [17:00:49] so I can just delete it? [17:00:56] someone will probably complain :< [17:01:08] I'll do then :p [17:01:36] wut https://www.wikidata.org/wiki/Talk:Q6293548 [17:02:02] wtf [17:09:46] legoktm: I requested deletion [17:09:50] ty [17:09:58] it's screwing up the cologneblue skin [17:10:08] it's screwing up everything :p [17:10:25] who needs so many interwiki links? :D [17:11:45] like whaaaat: http://cl.ly/image/3j153o0F1V1C [17:12:11] :DD [17:18:21] (03PS1) 10Daniel Kinzler: Make ResultBuilder's "indexed mode" work consistently. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102466 [17:21:05] (03CR) 10jenkins-bot: [V: 04-1] Make ResultBuilder's "indexed mode" work consistently. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102466 (owner: 10Daniel Kinzler) [17:21:07] (03PS2) 10Daniel Kinzler: Make ResultBuilder's "indexed mode" work consistently. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102466 [17:22:12] (03PS3) 10Daniel Kinzler: Make ResultBuilder's "indexed mode" work consistently. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102466 [17:22:37] (03CR) 10jenkins-bot: [V: 04-1] Make ResultBuilder's "indexed mode" work consistently. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102466 (owner: 10Daniel Kinzler) [17:23:24] wtf jenkins? i just rebased that!" [17:23:48] ohj, that'S for the old change set [17:24:32] (03PS8) 10Daniel Kinzler: (bug 40157) store entity status in page_props [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/99385 [17:24:44] (03PS5) 10Daniel Kinzler: (bug 47135) Make row IDs use BIGINT. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/99637 [17:24:58] (03PS7) 10Daniel Kinzler: (bug 45529) use composite indexes on wb_terms. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/99660 [17:25:17] (03PS8) 10Daniel Kinzler: Refactor SqlStore::doSchemaUpdates [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100361 [17:25:44] (03PS2) 10Daniel Kinzler: (bug 47070) No import of wikitext to entity NS [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100565 [17:26:43] (03PS6) 10Daniel Kinzler: (bug 58342) Inject TermBox when viewing [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101064 [17:26:52] (03PS2) 10Daniel Kinzler: (bug 58422) omit TOC with less than 3 sections [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101433 [17:30:34] (03PS2) 10Daniel Kinzler: Use EntityRevision for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101851 [17:30:58] (03CR) 10jenkins-bot: [V: 04-1] Use EntityRevision for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101851 (owner: 10Daniel Kinzler) [17:37:08] (03PS3) 10Daniel Kinzler: (bug 57231) Use ResultBuilder for EntityData. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101855 [17:39:19] (03PS5) 10Daniel Kinzler: Remove DB use from tests for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101453 [17:39:44] (03CR) 10jenkins-bot: [V: 04-1] Remove DB use from tests for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101453 (owner: 10Daniel Kinzler) [17:40:33] "the test would run for 285 thousand years" ... :D [17:41:48] (03CR) 10jenkins-bot: [V: 04-1] (bug 58342) Inject TermBox when viewing [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101064 (owner: 10Daniel Kinzler) [17:43:18] (03PS3) 10Daniel Kinzler: Use EntityRevision for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101851 [17:43:35] * DanielK_WMDE cires [17:43:40] cries even [17:44:16] (03CR) 10jenkins-bot: [V: 04-1] (bug 58422) omit TOC with less than 3 sections [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101433 (owner: 10Daniel Kinzler) [17:44:24] aaaahhhhgggg! [17:45:55] (03PS6) 10Daniel Kinzler: (bug 58344) use getUserLang for EntityView [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100829 [17:46:47] (03CR) 10jenkins-bot: [V: 04-1] (bug 57231) Use ResultBuilder for EntityData. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101855 (owner: 10Daniel Kinzler) [17:47:07] * aude cries! [17:48:14] baah [17:48:15] jshint ? [17:49:01] (03PS3) 10Daniel Kinzler: (bug 56711) use prefixed IDs in wb_terms table. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101197 [17:51:12] hashar: these failures with no output are extremely annoying. is there nothing we can do about that? [17:51:32] DanielK_WMDE: yup addshore tricked me in adding $wgDebugLogFile logging [17:51:37] they are attached to the jenkins build report [17:51:47] beside that, I have not investigated the failures / lack of logs more than that [17:52:04] (03CR) 10jenkins-bot: [V: 04-1] (bug 56711) use prefixed IDs in wb_terms table. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101197 (owner: 10Daniel Kinzler) [17:52:07] i have tried to fix any output buffering issues, which helped some; same with error reporting flags. but still, more often then not, errors in a full run make phpunit abort silently. [17:52:31] gah, why is *everything* failing no? [17:52:35] i just rebased a bunch of changes! [17:52:40] the only thing I got is the debug log such as https://integration.wikimedia.org/ci/job/mwext-Wikibase-repo-tests/6433/artifact/log/mw-debug-cli.log [17:52:55] though in that case the error is dumped to the console https://integration.wikimedia.org/ci/job/mwext-Wikibase-repo-tests/6433/console [17:54:09] o_O [17:54:12] wtf [17:54:25] got to leave sorry [17:56:29] (03PS4) 10Daniel Kinzler: (bug 57231) Use ResultBuilder for EntityData. [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101855 [17:56:56] (03CR) 10Daniel Kinzler: [C: 04-1] "incomplete" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101197 (owner: 10Daniel Kinzler) [17:57:39] (03PS6) 10Daniel Kinzler: Remove DB use from tests for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101453 [18:00:10] (03PS7) 10Daniel Kinzler: (bug 58342) Inject TermBox when viewing [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101064 [18:03:00] (03CR) 10jenkins-bot: [V: 04-1] Remove DB use from tests for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101453 (owner: 10Daniel Kinzler) [18:06:08] (03PS7) 10Daniel Kinzler: Remove DB use from tests for EntityData [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101453 [18:06:59] (03PS3) 10Daniel Kinzler: (bug 58422) omit TOC with less than 3 sections [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/101433 [18:36:01] * aude eagerly awaits travis results [18:36:01] (03PS1) 10Aude: Revert "Refactor Permissions tests into 2 seperate classes" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102478 [18:36:21] DanielK_WMDE: JeroenDeDauw ^ think this fixes the issues [18:36:32] awating travis though [19:03:17] [travis-ci] wmde/Wikibase#5 (fixtravis - c497278 : jenkins-bot): The build failed. [19:03:17] [travis-ci] Change view : https://github.com/wmde/Wikibase/compare/fixtravis [19:03:17] [travis-ci] Build details : http://travis-ci.org/wmde/Wikibase/builds/15663495 [19:12:10] hmmmm [19:13:17] maybe reverting isn't enough... maybe there's something wrong with the test before/after revert [19:14:54] PropertyContentTest::testSaveFlags P10 [19:15:25] get entities, UsageException: Could not create a new page [19:16:42] (03Abandoned) 10Aude: Revert "Refactor Permissions tests into 2 seperate classes" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102478 (owner: 10Aude) [19:21:46] EntityTestHelper::getId( 'Oslo' ) looks suspect [19:23:12] aude: Still hacking on the tests? [19:24:42] (03PS8) 10Tobias Gritschacher: Use snake_case for methods and variables [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 (owner: 10Aaron1011) [19:25:40] i think we're close to a solution [19:25:48] \o/ [19:25:57] permissions test is suspect [19:26:25] * aude wait for travis [19:29:01] aude: question [19:29:11] how can I create some test entities for testing special pages? [19:30:36] hmmm, good question [19:31:15] as much logic as possible should be factored out of the special page and then can just use data model objects to test the logic [19:31:48] aude: that however won't work [19:32:02] as the special page get's plain entity ids passed [19:33:33] aude: ah, got an idea :) [19:33:44] will ping you if I have further questions [19:33:46] :P [19:33:55] we also want to see the special page does correct output, handles permissions etc correctly [19:34:03] that's more of a system test [19:40:05] aude: i suggest to hack EntityContent::save(): at the end, check whether $status->isOK(). If now, print the user name, user groups, group permissions, and of course the status object. [19:40:22] of course, this will generate noise even if the failure was expected. [19:40:33] but it should allow you more insight into what is going wrong [19:40:34] * aude rage https://travis-ci.org/filbertkm/Wikibase/jobs/15665431 [19:40:49] ok [19:41:39] damn debug mode... that makes firefox unusable slow... [19:41:53] the huge amount of text, I mean [19:44:01] so many tests [19:46:25] Yeah... running them threaded would make that more bearable, but I guess our tests aren't ready for that [19:54:41] https://travis-ci.org/filbertkm/Wikibase/builds/15665428 did better (5/6 pass) but still something not right [19:57:26] mh... might also be random [19:57:58] back then I debugged that I think I found EntityTestHelper to be weird sometimes [19:58:04] yeah [19:58:56] DanielK_WMDE: aude: maybe someone of you can try to help me with https://gerrit.wikimedia.org/r/#/c/98281/ ? [20:01:40] benestar: any particular help, or just review? [20:01:54] DanielK_WMDE: read my latest comment down htere [20:02:25] (03PS1) 10Bene: Create more special pages for special pages [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102495 [20:03:24] benestar: eek, four levels of nesting in a 80 line method! break it up, please :) [20:03:48] (03CR) 10Bene: "A little more detailed explanaitions:" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/98281 (owner: 10Bene) [20:03:58] benestar: uh, you have two EditIntity instances? Why? [20:04:03] ok, that's another problem [20:04:13] DanielK_WMDE: it's merge ;) [20:04:27] mainly the same steps as in the api module [20:04:41] oh, because the "from" entity is also updated. right. [20:04:44] thus I wonder why it doesn't work in the special page [20:05:05] DanielK_WMDE: just added another comment; the edit is performed but the status fatal anyhow [20:06:18] seems like that fatal status is thrown wrongly [20:06:19] eewww, why is ChangeOpsMerge using ItemContent, not just Item?? [20:06:49] that's just WRONG. [20:06:51] and useless [20:06:52] (03CR) 10jenkins-bot: [V: 04-1] Create more special pages for special pages [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102495 (owner: 10Bene) [20:06:58] bah [20:07:20] DanielK_WMDE: not my fault :/ [20:07:26] i know :) [20:07:42] * DanielK_WMDE throws a large trout at addshore [20:08:31] aude: Disabling the permissions test isn't helping locally [20:08:37] :/ [20:08:52] getting failure in EditEntityTest (not the API one) still [20:09:28] DanielK_WMDE: but this cannot be the problem as the api module is working correctly [20:09:36] hoo: you can reproduce the issue locally? [20:09:40] reliably? [20:09:47] DanielK_WMDE: Test failures? Yes [20:09:57] I get 100% failures with mariadb [20:10:03] hoo: then you should be able to debug it and see why it fails, no? [20:10:20] I started doing that on the weekend, but I got nowhere and then I gave up... [20:10:32] can you do what i suggested to katie, regarding hacking EntityContent::save()? [20:10:42] Oh, I can try that [20:11:09] benestar: well, for which save do you get the "nothing changed" error? when changing the "from" or the "to" entity? [20:11:27] when changing the first, the from [20:11:35] tried it with some debug output ;) [20:11:37] hoo: my suspicion is that saving fails because something messed up the group permissions. [20:11:58] DanielK_WMDE: nope [20:11:59] 0 => 'edit-already-exists', [20:12:12] benestar: well, does the "from" entity change? after pplying the change op, it should be empty. have you checked that? [20:12:39] DanielK_WMDE: yes, it changed [20:12:51] that is what I mean saying it had a history entry [20:13:10] hoo: hm... in that case, just check what entity id fails, and grep the source for that ID. it's probably hardcoded somewhere, and shouldn't be. [20:13:18] mh [20:13:25] all hardcoded ids must be killed from the tests, otherwise we will have that issue again and again. [20:13:32] I did that on Saturday but we have a lot of (expected) failures there [20:13:46] so it's kinda hart to filer [20:13:48] * filter [20:13:53] benestar: now i'm completely confused. it fails, but it gets a history entry? [20:14:04] can you describe again what exactly happens`? [20:14:06] yes, right [20:14:11] ok [20:14:12] o_O [20:14:28] 1. I go to the special page, enter two entities and click merge [20:15:12] 2. I get the message "Your edit was ignored because no change was made to the text." [20:15:45] 3. I go to the "from" item which should be empty and see an entry in the history: Merged item in to "Q142" [20:15:52] benestar: giving myself 15 minutes with that, then i'll leave the office. mind is melting [20:16:18] the diff shows that label and description are removed as I expected [20:17:03] benestar: have you tried to use an actual debugger to see what is happening? [20:17:31] no, only output. [20:17:37] you know, debuggers are awesome, they let you see what the program does :) [20:17:39] some echo debugging [20:17:51] DanielK_WMDE: I don't have one for php :/ [20:17:53] are you using an IDE? [20:18:01] atm not [20:18:19] DanielK_WMDE: is there any you would prefer? [20:18:23] get xdebug and phpstorm. it'S free for FOSS, we have a key [20:18:26] maybe eclipse? [20:18:35] no, eclipse suchs for php [20:18:38] *sucks [20:18:51] heh [20:18:54] It's terribly slow [20:18:55] ^^ [20:18:56] phpstorm is good, but not open source. netbeans is an alternative [20:18:59] and I mean, terribly [20:19:09] but everyboday has been moving to phpstorm over the last year here. [20:19:19] (except katie, i guess) [20:19:21] but how should this work, I mean you call teh special page from the browser, don't you? [20:19:40] I'm quite happy with netbeans for PHP, only downside is that it soaks up a lot of RAM if you import a huge project (MediaWiki + Wikibase) [20:19:41] benestar: yes. the debugger attaches itself to the server process. it's magic. [20:20:04] works perfectly - i call the page in the browser, and the IDE hits the brakpoint. the browser waits. [20:20:22] :O [20:20:28] hoo: have you set it up for xdebug? [20:20:28] DanielK_WMDE: but only with localhost, right? [20:20:38] no [20:20:50] benestar: can be configured for remote hosts too. localhost just works. [20:21:23] Do you have some links / tutorials for setub? [20:21:24] benestar: i *strongly* recommend to spend a day setting up an ide and debugger [20:21:33] DanielK_WMDE: Nope, but that's possible [20:21:37] *setup [20:21:46] improves coding productivity a lot, and debugging productivity A HUGE LOT. [20:21:58] hoo: invest half a day :) [20:22:23] DanielK_WMDE: Then I could just move to phpStorm anyway... the netbeans Lua plugin just sucks so bad [20:22:30] benestar: http://confluence.jetbrains.com/display/PhpStorm/Zero-configuration+Web+Application+Debugging+with+Xdebug+and+PhpStorm [20:22:45] benestar: i recall i had to manually change something in the xdebug configuration... ask me if you run into issues [20:22:50] only saw that again yesterday... it's eg- crying about syntax errors in valid files [20:22:59] (03PS2) 10Bene: Create more special pages for special pages [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102495 [20:23:08] DanielK_WMDE: thanks :) [20:24:13] "debug_backtrace( DEBUG )" that's totally what I inteded... meh [20:24:49] hoo: what are the language settings for jenkings? [20:24:54] * jenkings [20:24:58] * jenkins [20:25:03] gnah, I cannot spell him correctly :P [20:25:13] jenkins :D [20:25:14] It's running with content language English AFAIR [20:25:26] well, my tests fail anyhow :/ [20:25:34] asserting the language is en [20:25:49] benestar: if your tests rely on something that depends on the user language, your tests must set the user language [20:26:18] I just tried to and waiting for jenkins [20:26:20] DanielK_WMDE: If you find time for that someday, could you mail me a product key for a version which supports PHP + JS + CSS + Lua? That's essentially what I use for WMDE [20:27:16] DanielK_WMDE: great, it worked :) [20:27:53] seems like you need a plugin for Lua support http://plugins.jetbrains.com/plugin/5055?pr=&offset=10&max=10 mh [20:28:38] benestar: ok, confirmed [20:28:45] now for finding out why. [20:29:22] DanielK_WMDE: is this supposed to be one sentence? [20:29:51] hm? what? [20:30:58] benestar: sorry, can't get this finished now. let me try tomorrow, ok? [20:31:01] ping me if i forget [20:31:20] DanielK_WMDE: of course :) [20:31:58] ewk... already 7 log entries for with the return Status::newFatal( 'edit-already-exists' ); :/ [20:32:30] * aude waiting for travis [20:32:55] aude: Keen to examine that data? :P Not sure I'll come to that today [20:35:19] * aude put in slow queue on travis [20:36:05] aude: :P Why you no use an own VM? [20:36:29] doens't mean the tests will fail for me [20:36:41] I see constant failures in mine [20:36:49] hmmm [20:37:09] not on my external hosting either... no failures [20:37:29] weird [20:37:40] I get 100% failures on mine with mariadb [20:38:13] using a slightly modified version of the travis scripts to set up, so that my set up should be rather compatible [20:38:32] $user = null in a bunch of places [20:42:01] if null, then WikiPage::doEditContent uses $wgUser [20:50:48] [travis-ci] wikimedia/mediawiki-extensions-Wikibase#1446 (master - 75db5ef : Translation updater bot): The build is still failing. [20:50:48] [travis-ci] Change view : https://github.com/wikimedia/mediawiki-extensions-Wikibase/compare/80edaca54dab...75db5ef827f2 [20:50:48] [travis-ci] Build details : http://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/builds/15669303 [20:54:44] aude: found something [20:56:58] ok, not the cause, but still bad [20:59:26] (03PS9) 10Aaron1011: Use snake case [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 [21:00:39] (03PS10) 10Aaron1011: Use snake case [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 [21:02:04] (03PS1) 10Hoo man: Don't use hard coded Item ids in SetClaimTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102548 [21:02:11] (03PS11) 10Aaron1011: Use snake case [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 [21:03:57] ok [21:09:14] * hoo waits for jenkins... [21:09:16] slowness... [21:12:35] awww, jenkins is faster now [21:12:41] oh really? [21:12:52] Looks like my change got lost [21:12:53] been much faster today [21:12:56] :/ [21:14:08] aude: Any idea why it's not picking up on https://gerrit.wikimedia.org/r/102548 ? [21:14:19] shall I rebase? [21:14:34] no idea [21:15:03] looks like jenkins is lost [21:15:10] fun ... [21:15:11] not just wikibase [21:16:58] running the suite with both MariaDB and SQLite locally on that change now (again) [21:17:20] I doubt we see any fixes, but it's still a (minor) improvement [21:19:07] it picked up the change by Aaron1011 [21:19:08] Failed asserting that '/* wbeditentity-override:0| */' matches PCRE pattern "!wbmergeitems!". [21:19:09] mh [21:19:19] for mergeitems [21:19:31] on master? [21:21:23] I'd love a phpunit mod that automatically enables and disables groups to isolate failure causes... that would of course take ages to run, but could deliver great results [21:22:42] well, travis but yes master [21:22:45] aude: https://gerrit.wikimedia.org/r/#/c/102548/ it took 20 minutes, but then jenkins picked it up... wanna merge? [21:23:10] (03CR) 10Aude: [C: 032] Don't use hard coded Item ids in SetClaimTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102548 (owner: 10Hoo man) [21:23:16] there [21:23:19] :) [21:25:09] interesting... now I no longer see the old one, but new ones in "EditEntityTest" [21:25:17] w/ both SQLite and Maria [21:25:30] * old failures [21:29:16] aude: "$sysopUser = User::newFromId(1);" that looks a little flawed :P [21:29:39] in EditEntityTest ... but probably also not the cause [21:32:28] going to smth. [21:33:48] huh [21:34:03] https://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/jobs/15669309 wtf [21:37:22] grrr [21:39:09] aude: do you mean me? [21:40:13] * benestar becomes scared [21:46:06] benestar: Nah... she's angry at our phpunit tests ;) [21:46:12] they suck [21:46:23] benestar: all the tests [21:46:29] :/ [21:46:43] and I am creating even more... :( [21:46:52] * aude won't merge special pages or other new stuff until the situation is resolved [21:47:06] more stuff only makes it harder to debug [21:47:11] yeah [21:47:47] so shall try to isolate the problem... excluding api tests now [21:48:04] we have to resolve that situation in the next couple of days... this is getting embarrassing... almost 2 weeks now [21:48:33] aude: Can you manually submit https://gerrit.wikimedia.org/r/102548 [21:48:37] I want to go on... [21:48:51] * aude nods [21:49:05] wtf jenkins? [21:49:33] (03CR) 10Aude: [V: 032] "seems jenkins is absent" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102548 (owner: 10Hoo man) [21:49:49] thanks :) [21:56:23] (03PS3) 10Bene: Create more tests for special pages [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102495 [21:56:32] aude: :P [21:57:06] :) [22:00:31] (03PS4) 10Bene: Create more tests for special pages [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102495 [22:01:27] gnah, I should reduce the amount of patchsets I am using [22:03:42] (03CR) 10jenkins-bot: [V: 04-1] Create more tests for special pages [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102495 (owner: 10Bene) [22:14:21] (03PS1) 10Hoo man: Some sanity changes in EditEntityTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102571 [22:15:09] that's also not going to fix anything [22:15:27] k [22:15:40] * aude may try it with travis [22:17:34] excluding api tests, we get "Wikibase\Test\PropertyContentTest::testSaveFlags" [22:17:40] i think that's fixable [22:18:03] aude: is that the one with P10? [22:20:38] yes [22:20:42] EntityPerPageTableTest [22:20:44] * aude makign patch [22:20:48] and DatabaseRowEntityIdIteratorTest [22:20:51] that one looks evil [22:21:06] DatabaseRowEntityIdIteratorTest isn't using the DB AFAIS [22:21:26] (not creating pages, I mean) [22:21:28] [travis-ci] wikimedia/mediawiki-extensions-Wikibase#1447 (master - 7eec1e1 : Aude): The build is still failing. [22:21:28] [travis-ci] Change view : https://github.com/wikimedia/mediawiki-extensions-Wikibase/compare/75db5ef827f2...7eec1e15d422 [22:21:28] [travis-ci] Build details : http://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/builds/15674037 [22:22:42] changing both [22:22:47] :) [22:23:47] wait for travis again [22:27:46] aude: how can tests run on my local device but fail on jenkins? [22:28:13] WMF should replace jenkins with a self hosted travis instance taking data from gerrit :P [22:28:56] benestar: There's a huge amount of things that can cause weird test failures (that's what aude and I are currently hacking on actually) [22:29:19] hoo: Have you seen what Timo's done? It's effectively the same. :-) [22:29:39] James_F: oh... what? Sounds exciting [22:30:41] benestar: might be stuff in the tests that depend on settings [22:30:48] hoo: Cf. https://integration.wikimedia.org/ci/job/mwext-VisualEditor-npm/20/console [22:31:04] hoo: Running grunt test through npm. [22:31:06] tests need to ideally not depend on settings at all, and if so then control for them [22:31:15] James_F: :) [22:31:29] aude: I only save a label in de [22:31:38] jenkins is set to english [22:31:46] try changing your wiki langauge to something else [22:31:50] then run tests again [22:32:07] aude: it is, but I call the speical page using Special:SetLabel/Q123/de [22:32:11] e.g. change $wgLanguageCode [22:32:19] in local settings [22:32:23] so it cannot depend on settings [22:32:23] benestar: Better do "$this->setMWGlobals( 'wgContentLanguage', \Language::factory( 'de' ) );" in your tests [22:32:36] benestar: right or has to control, per hoo [22:33:02] hoo: no, I do $item->setLabel( 'de', 'foo' ); [22:33:16] no settings dependence at all [22:33:37] benestar: no [22:33:54] if you test special page output html you're implicitly testing against your language [22:34:21] you language = the wiki's language [22:34:24] hoo: I set my language to 'en' in the test [22:34:25] Failed to match html output with tag 'value' passing two subpage values [22:34:53] yes, I know, everything else works fine [22:34:54] i think your test expects different content lang than jenkins [22:35:26] special page depends on settings [22:35:28] but I pass the language to the page --> does not depend on user language at all [22:35:52] benestar: the special page (as context)? [22:36:00] I need to see the code, to judge properly [22:36:19] hoo: list( $output, ) = $this->executeSpecialPage( 'Q42/de' ); [22:36:48] but that's not setting the user's interface language (at least it shouldn't) [22:38:03] hoo: of course, but I am jsut testing the value [22:38:41] not the interface language [22:55:35] aude: https://travis-ci.org/filbertkm/Wikibase/jobs/15675697 ... ??? wtf [22:57:14] yeah [22:58:00] https://travis-ci.org/filbertkm/Wikibase/builds/15675606 is good :) [22:58:07] or lucky [22:58:47] https://travis-ci.org/filbertkm/Wikibase/builds/15672049 (no api tests, also lucky) [22:59:27] seems they execute twice, under "build history" and "pull requests" [23:01:11] aude: the commits you submit for review in a PR run, and the merge commit into master also runs [23:02:04] it's one for my branch, one for pull request into master [23:06:45] * aude wonder why i get no mails from travis [23:20:24] (03PS1) 10Hoo man: Remove hard coded entity ids from EntityPerPageTableTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 [23:21:13] aude: ^ [23:21:30] that's more sane than just changing the hard coded ids [23:21:30] (03PS1) 10Aude: change hardcoded item and property ids [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102590 [23:21:33] ah [23:21:40] conflict! [23:22:09] pity [23:23:04] (03PS12) 10Aaron1011: Use snake case [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/100209 [23:23:40] (03Abandoned) 10Aude: change hardcoded item and property ids [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102590 (owner: 10Aude) [23:24:52] (03CR) 10Aude: "much better" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 (owner: 10Hoo man) [23:25:02] :) [23:27:24] (03PS2) 10Hoo man: Remove hard coded entity ids from EntityPerPageTableTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 [23:27:48] (03CR) 10Hoo man: "Removed pointless comment" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 (owner: 10Hoo man) [23:35:05] (03CR) 10Aude: [C: 032] Remove hard coded entity ids from EntityPerPageTableTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 (owner: 10Hoo man) [23:38:48] aude: But I wonder what's causing the phpunit abortions we started seeing today [23:38:59] not sure [23:39:33] Jenkins is probably still not doing gate submits :/ [23:40:08] i think it is [23:41:25] (03CR) 10Krinkle: [C: 032] "(per Aude, zuul was not responding due to a bug, fixed now)" [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 (owner: 10Hoo man) [23:42:26] aude: Starting test 'Wikibase\Test\PropertyInfoTableBuilderTest::testRebuildPropertyInfo'. [23:42:32] that's where master fails for me (SQLite) [23:43:33] my change wont fix this issue [23:43:52] really.... still failing? [23:44:18] I think so [23:44:41] I no longer have the chance around, as I did it on master and reseted to HEAD~1 to get a clean master again [23:44:44] * change [23:47:04] (03Merged) 10jenkins-bot: Remove hard coded entity ids from EntityPerPageTableTest [extensions/Wikibase] - 10https://gerrit.wikimedia.org/r/102589 (owner: 10Hoo man) [23:47:08] lolwat [23:47:26] loading of pages is so slow I'm sort of embarassed to link anyone to Wikidata now >_> [23:48:19] JeroenDeDauw: better keep linking them to https://travis-ci.org/wikimedia/mediawiki-extensions-Wikibase/builds :P [23:48:55] one of these even passed! [23:49:22] aude: I guess I have to reenable xdebug to get more info from phpunit? [23:49:32] Do you also see these things? [23:50:42] failures are still there on my version :( [23:51:01] trying to run the WikibaseStore tests isolated now [23:52:59] i'm updating my github stuff [23:53:22] i haven't seen property info fail in more than a day [23:53:39] mh [23:53:44] maybe I should reset my .sqlite [23:53:50] good idea, actually [23:53:57] wait [23:53:58] hoo: people that like red will be so extatic :) [23:54:03] 1) Wikibase\Test\PropertyInfoTableBuilderTest::testRebuildPropertyInfo [23:54:03] assert(): Assertion failed [23:59:51] ok, these tests pass fine on a clean SQLite DB