[01:07:01] mwselenium-quibble-docker is broken on wikibase again :( [05:49:55] JOIN [05:50:38] CLEAR [07:29:02] SMalyshev: hooray [07:29:20] SMalyshev: is it a random failure though? that job just generally seems pretty unreliable, silly old ruby tests [10:04:11] Hi, how shoulk I deal with P31 (instance of) with Saints? [10:05:38] Should I use P31 = human and set saint as occupation? [10:06:07] Or can I directly set P31 =saint? [10:06:17] Arch2all: saint isn't an accupation in my mind. [10:06:38] ...because saint is a subclass of human [10:06:47] Why not just got with instanceof human *and* instanceof saint? [10:07:30] Isn't this redundant, because saint is a subclass of human? [10:07:38] well, since it'S a subclass, just useing instanceof saint is sufficient in theory. but not necessarily. for instance, that person being a saint is according to (e.g.) the catholic churc, and may be disputed by others. [10:07:43] while them being humand is undisputed [10:08:05] that kind of thing is a good reaons to have redundant-looking statements like instanceof human and instanceof saint [10:08:44] the instanceof saint class could also have a qualifier that sais since when they have been considered a saint. that qualifier wouln' [10:08:51] ...wouldn't apply to them being human [10:08:53] etc [10:09:20] the point is: just because the main value is redundant doesn't mean the statement is redundant [10:10:16] OK, than I just add a P31 = saint Statement [10:11:54] thanks :) [10:13:32] Just noticed, that there is a property P411 (canonization status), which is used in some cases for saint attribution [10:16:10] It is better now to set ((P411 = saint) and (= human)) OR (P31 = human and saint) OR ((P31 = human and saint and (P411= human)) ? [10:17:36] typo: first option ...and (P31=human) ... [10:29:21] https://usercontent.irccloud-cdn.com/file/AopNtXZS/image.png [10:29:23] ^^ senses :D [10:55:13] makes sense to me [10:56:10] Arch2all: hmmm, we also have https://www.wikidata.org/wiki/Property:P411 [11:24:17] the gloss field for senses is too small [11:27:37] and I could really use a lexeme search that can filter by language [11:33:55] and I could save so much time if I didn't have to manually focus everything, it knows I want to edit it so why isn't the focus in the form fields >_< [11:41:35] luckily I know enough javascript to make it work for me :P [11:54:50] @sjoerddebruin Yes, I noticed that: It is better now to set ((P411 = saint) and (P31= human)) OR (P31 = human and saint) OR ((P31 = human and saint and (P411= human)) ? [11:56:25] @sjoerddebruin corrected: It is better now to set ((P411 = saint) and (P31= human)) OR (P31 = human and saint) OR ((P31 = human and saint and (P411= saint)) ? [12:20:33] Arch2all: ((P411 = saint) and (P31= human)) probably [12:54:44] @sjoerddebruin ok, will tag it this way [15:23:28] hm... I wonder how entering languages for senses could be improved [15:23:47] realistically there's only a couple of languages I'm going to be able to write glosses for senses in [15:24:33] maybe it should be like labels, with the languages you claim to speak always present [15:24:55] Auregann_WMDE: ping? [15:40:45] nikki: everyone is celebrating the deployment of senses and having cake :) [15:40:54] but yes, i agree [15:44:05] New error in wmf.26 with production impact: Argument 2 passed to Wikibase\Lib\Store\Sql\WikiPageEntityMetaDataLookup::__construct() must implement interface Wikibase\Lib\Store\Sql\PageTableEntityQuery, MediaWiki\Storage\NameTableStore given in /srv/mediawiki/php-1.32.0-wmf.26/extensions/WikibaseQualityConstraints/src/ServiceWiring.php on line 254 [15:44:23] This is a train blocker [15:50:31] please help ^ :) [15:56:23] I think the main developer is on holiday... [15:56:27] nikki: that's actually part of the plan, making the editing experience of glosses look more like labels :) [15:56:50] but since we're also currently rethinking the way termbox/labels/languages behave, we'll try to think about it globally [15:58:21] sjoerddebruin: who would that be? [15:58:42] addshore: could you have a look at the error message above? [15:59:27] Yeah maybe Adam knows, mostly Lucas works on the extension and I also see some stuff by Jonas_WMDE recently [16:01:26] *looks up* [16:01:53] twentyafterfour: greg-g is there a ticket? [16:02:07] https://phabricator.wikimedia.org/T207394 [16:02:10] addshore: ^ [16:02:10] *looks* [16:02:52] i wrote it, clearly my fault [16:04:14] addshore: Daniel asks why the QualityConstraints constructs its own instance [16:04:36] Auregann_WMDE: there is a ticket for that [16:05:23] Auregann_WMDE: duesen https://phabricator.wikimedia.org/T206164 it should be picked up by the campsite in the next story time [16:21:45] Auregann_WMDE: is there a ticket about not immediately showing the error message when typing a language code for senses? [16:22:09] nikki: absolutely [16:22:19] do you have a link? :) [16:22:58] nikki: is that what you're talking about? https://phabricator.wikimedia.org/T205528 [16:23:48] yes, thanks [16:24:48] hm, is that not released then? [16:26:35] nikki: not yet, will be in the next few days [16:26:40] ok [16:27:09] is there a way to tell from a ticket whether something is actually on wikidata.org yet? [16:27:34] nikki: Auregann_WMDE its just about to be deployed actually [16:27:42] but the train just got rolled back again :) [16:28:23] nikki: that's a very good question that we're trying to solve these days ;) but what can help you already is the tag with the mediawiki version [16:28:43] nikki: https://tools.wmflabs.org/versions/ can help you there, that tells you what branch is deployed where (wikidata is in group1) and then on the task there should be a project tag that correlates to that [16:29:32] bah, can't load tools.wmflabs.org... will have to try looking later [16:35:08] hehe [16:37:03] I wish I knew what was going on, I spend enough time on wikidata as it is without it being the only thing on the internet that I can access [16:42:47] Auregann_WMDE: what about the glosses not lining up when you have multiple senses, is that known? [16:43:15] the language field is really wide sometimes, looks weird [16:44:57] Auregann_WMDE: not lining up? [16:45:03] meh, wrong ping, nikki ^^ [16:46:40] I would expect the gloss to start at the same horizontal position, but it doesn't [16:47:16] nikki https://usercontent.irccloud-cdn.com/file/V7kSIPFf/image.png [16:47:23] is it different for you? [16:47:31] multiple senses, not multiple glosses [16:47:40] gotcha [16:48:38] like on https://test.wikidata.org/wiki/Lexeme:L121, the first one's glosses are closer to the edit button than the language codes (for me at least) [16:49:23] hmm, could you send a screenshot? [16:50:56] dunno where to send it to... [16:51:24] seems ok in firefox, but not in chrome-based browsers [16:51:28] addshore: seems to be pretty consistent [16:51:31] nikki: you could send it to https://phabricator.wikimedia.org/file/upload/ :) [16:51:38] addshore: the selenium failure I mean [16:52:12] nikki: it seems to be in line for me on chrome afaik https://usercontent.irccloud-cdn.com/file/9WusUfNr/image.png [16:52:19] addshore: example: https://integration.wikimedia.org/ci/job/mwselenium-quibble-docker/2192/console [16:52:31] SMalyshev: yup, we found the issue just now actually! as it isn't just a browser test failure, stuff is actually broken [16:52:37] SMalyshev: https://phabricator.wikimedia.org/T207397 [16:53:21] https://phabricator.wikimedia.org/F26623471 [16:56:46] nikki: interesting, no that is not right, that deserves a phabricator ticket [16:57:23] browser details will definitely be helpful in the ticket,and even the exact item your seeing it on, as I can't reproduce it here with my default stuff [16:58:45] it also seems to happen in edge and IE11 [17:00:03] addshore: tests find stuff that actually broken? surprising! ;) ok, I'll then just wait for the fix [17:07:17] Auregann_WMDE: are there follow up tickets for the Entity suggestions from constraint definitions beta feature? I don't see anything for long term [17:09:59] sjoerddebruin: not at the moment, we're collecting feedback in the existing ticket [17:18:37] Auregann_WMDE: link? [17:20:37] sjoerddebruin: https://phabricator.wikimedia.org/T202712 [17:21:47] Ah, the order question is already asked. I've created https://phabricator.wikimedia.org/T207363 earlier today, but it feels weird to make it a subtask to a closed ticket. [17:23:04] yeah I know that's not the best, I'll try to reorder this a bit in the next days... unless someone elses does it before that [17:40:42] addshore: https://phabricator.wikimedia.org/T207401 [17:53:06] what about showing all senses for a lexeme when entering a lexeme id when searching for senses? is that a known feature request? [17:53:38] (i.e. if I enter L123 when looking for a sense, it should show L123-S1, L123-S2, etc) [18:22:48] hi [18:23:16] i'd like to know if there is a database or list where i could check which wikipedia language editions have the ArticlePlaceholder installed [18:24:21] mmecor: yes [18:25:39] let me find it [18:26:34] mmecor: https://gerrit.wikimedia.org/r/plugins/gitiles/operations/mediawiki-config/+/master/wmf-config/InitialiseSettings.php#18723 [18:26:42] thanks :) [18:26:42] nikki: I havn't heard that one yet [18:27:04] ok [18:27:10] ticket will have to wait, I need food [18:27:46] i'm surprised there are very few (just 13) [18:32:41] mmecor: im sure more can request it if they want it :) [18:32:51] probs :) [18:50:35] Is it just me or is the "Load more..." button on property suggestions gone [19:55:47] sjoerddebruin: https://nl.wikimedia.org/wiki/Wikidataverjaardag [19:56:09] multichill: ja, ik weet het. Ben druk bezig met vervanging regelen op werk. [19:57:13] En het ziet ernaar uit dat https://www.wikidata.org/wiki/Wikidata:Property_proposal/Wikipedia_suggested_article_name er gaat komen, maar eerst alleen in het Nederlands [19:58:39] Nice! [19:58:59] Ik heb artikelen zonder gender nu weer op nul voor nlwiki, zit nu in de geboortedata [19:59:18] Valt opzich nog mee voor dit jaar, zit met 250 items al in vorig jaar. [19:59:30] Is er nu een goed overzicht van de nieuwe artikelen of ze wel of niet gekoppeld zijn en aan wat? [19:59:39] Ik realiseerde me dat ik dat vrij gemakkelijk met SPARQL kan maken [19:59:54] Er is https://tools.wmflabs.org/wikidata-todo/duplicity.php?wiki=nlwiki&mode=list weet ik [20:02:54] Goed bezig. Ik moet zo te zien https://www.wikidata.org/wiki/User:Multichill/Humans_no_gender ook weer eens uitmesten [20:03:03] addshore: looks like the fix for selenium worked [20:07:25] multichill: veel zijn trouwens standaardopmaak qua geboorte- en sterfdata, wellicht je botje weer draaien? https://nl.wikipedia.org/wiki/Jan_van_Wijhe [20:07:45] Zie trouwens dat er weer 20 "nieuwe" mensen zijn :P [20:14:07] Heb ik die gedaan? Of was Edo dat? [20:16:30] Edo [20:16:49] 20 is leuk voor een avond, maar de laatste tijd was het 200/300 [20:24:27] SMalyshev: amazing :) [21:08:43] SMalyshev: so, im thinking of writing my own query service updater to read from kafka, can I quickly check I sort of understand the process? [21:18:05] hmm, maybe I need to finish figuring out event bus actually [21:19:00] https://www.mediawiki.org/wiki/Extension:EventBus -> https://github.com/wikimedia/eventlogging -> kafka -> updater -> query service [21:19:09] * addshore thought eventbuss wrote directly to kafka [21:30:07] addshore: https://phabricator.wikimedia.org/T200931 [21:30:40] addshore: on a meeting now, but after I'm done I'll look [21:30:41] ooooh [21:41:30] #wikidata [22:01:54] addshore: so, I'm back [22:02:17] addshore: what is the use case you are trying to do - is it for integrating with wikidata or your own wikibase? [22:02:44] nope, essentially I want to have multiple wikibases, but only 1 updater process [22:03:16] * addshore has to figure out how to setup the eventbus, eventlogging kafka pipeline first though [22:04:04] i like the sound of using eventstreams instead of kafka directly, sounds like it would make the updater bit a little easier [22:04:11] addshore: hmm so you can use recentchanges or kafka driver, I think for kafka you can run them in parallel without much problem [22:04:35] but that requires direct access to kafka streams (which e.g. are not public on wikidata) [22:05:08] so eventstreams *are* public for wikidata, but I have no idea how easy/hard it would be to get it going for your own wikibase [22:05:54] wiki rc changes stream, the only problem I think might be timestamp - I don't think it is per wikibase, so if you have several updaters they may step on each other's toes with timestamp [22:06:05] but that would be not too hard to fix... [22:07:12] I am not sure though how events get to kafka... that part is mostly magic to me :) [22:12:50] yeh [22:13:02] it turns out its quite a path [22:13:24] the mediawiki event bus extensions talks to a http endpoint for some event logging service, which then writes to kafka [22:13:42] so, whatever happens, I probably need to get all of that setup :D [22:14:39] addshore: or you can use rc driver... [22:15:06] rc driver being? [22:15:16] RecentChanges poller [22:15:28] yeh, my issue is that doesn't really scale that well :/ [22:15:34] ah, yes [22:15:38] well, in terms of multiple wikibases [22:16:13] well, depends on how many wikibases... it will be http req per wikibase but if it's 2-3 then shouldn't be too big of a deal? [22:16:16] i think in terms of scaling for multiple wikibases the kafka poller would work best, as afaik all edit events go into the same topic, no matter what wiki they come from [22:16:29] im thinking more, 1000 ;) [22:16:51] addshore: hmm that may not work that well since if you're going to put them into the same topic, how you know where to bring RDF from? [22:17:03] right now single poller only has single URL to fetch RDFs [22:17:23] well, the topic has the wikiname, and from the wikiname you can get the location of where to get the rdf [22:18:02] yeah I guess that'd require some work on the poller though... it will have to namespace IDs and do other stuff it currently doesn't do [22:18:14] so probably would require some work to do it [22:19:15] yup, i need to find some time :) [22:19:54] yes :) I don't think it's impossible, I think the difference is not that much (if wikibases have distinct base concept URIs etc.) but needs some work [22:32:27] PROBLEM - High lag on wdqs1010 is CRITICAL: 6852 ge 3600 https://grafana.wikimedia.org/dashboard/db/wikidata-query-service?orgId=1&panelId=8&fullscreen