[07:58:29] L [07:59:16] Also good morning. [08:03:42] JJSJS [08:05:04] * nikki wonders if there are any legitimate users using that kiwiirc thing [08:05:57] nikki: It's very easy to use, thus a good way for non-IRC people to join us [08:06:05] no idea how well that's being used, though [08:39:46] :) [09:02:27] do we have stats about the number of sitelinks? [09:02:47] for example, how many items have only one sidelink, two, etc. [09:05:07] Bottom section of https://grafana.wikimedia.org/dashboard/db/wikidata-datamodel [09:06:30] thanks :) [09:08:35] Interesting. https://grafana.wikimedia.org/dashboard/db/wikidata-datamodel-statements?panelId=8&fullscreen [09:08:52] who deprecated 25k statements? [09:09:36] Oo [09:09:58] If you find out what happened, I'm curious [09:11:58] Not an expert in that. [09:14:12] Can you take a look at https://www.wikidata.org/wiki/Talk:Q191781 though, Ash_Crow? [09:24:01] hmm [09:24:15] no need to keep aliases with typos [09:25:45] Sure? [09:44:28] if it's a common mistake people make (i.e. people do often write it that way) I'd probably keep it [09:46:19] searching does seems to find plenty of pages spelling it that way [09:46:40] it's not a common typo [09:47:05] and you can have infinite aliases if you try to have all possible typos [09:48:23] sure, which is why I wouldn't keep all typos [09:59:24] plus, it can be counter productive ("if it matches then it's good" reasoning) [10:03:40] by that logic we should only ever find matches if you can spell perfectly with all the right letters and all the right accents in right places, but that wouldn't be very useful :P [10:17:32] sjoerddebruin: looks like it was proteinboxbot marking all statements on a bunch of items as deprecated o_O e.g. https://www.wikidata.org/w/index.php?title=Q20780100&diff=374523860&oldid=342861785 [10:17:49] Hm, right... [10:20:40] HI there. I have set up a SPARQL endpoint on my server following the wiki: https://www.mediawiki.org/wiki/Wikidata_query_service/User_Manual#Standalone_service [10:20:49] When I issue a complex query which requires a lot of time for its execution Blazegraph issues an exception "Query deadline is expired." [10:20:54] How can I fix it? [10:21:01] Thank you in advance! [10:22:47] WikidataFacts: ^ [10:23:22] hm, I don’t have much general advice on query optimization… [10:23:32] avoid subqueries like the plague [10:23:41] try to express as much as possible in triple patterns instead of FILTER [10:24:08] FILTER with any form of order or sorting (e. g. <=) can still be a lot more efficient than other kinds of expressions [10:24:30] (e. g. instead of YEAR(..) = x, check that the date is between jan 1st and dec 31st of that year) [10:24:52] jewels: can you show use the query, perhaps? [10:26:51] WikidataFacts: https://bpaste.net/show/b5989362992f [10:29:14] hm… you don’t use ?movie_label, so you could throw that away… [10:29:33] and that ?p looks very expensive too [10:30:48] you could at least assert that it’s from wdt: with a triple like [] wikibase:directClaim ?p [10:31:04] but still, the query runs in half a second on WDQS: http://tinyurl.com/jtt5ar4 [10:31:10] so I’m not sure why it times out for you [10:36:47] WikidataFacts: The problem seems to be with the OFFSET [10:37:10] WikidataFacts: If I increase its value I reach the timeout easily [10:39:11] I’m not sure if using OFFSET without ORDER BY is a good idea anyways [10:39:43] I don’t think you’re guaranteed to get the right results if you run the same query with different offsets without sorting [10:44:21] WikidataFacts: ORDER BY makes the query incredibly slower [10:45:13] WikidataFacts: Can i rewrite the query in some way in order to make it faster? [10:53:30] jewels: if you’re willing to restrict ?p to a certain set of properties, and you can live without labels, this one doesn’t time out even for large offsets: http://tinyurl.com/jhwurty [10:53:46] doesn’t have ORDER BY, but it’s the best I can do… [11:32:49] Who runs hte wikidata twitter account? :P [11:33:12] Lydia, Lea... [11:44:18] I'm just amused they liked my tweet :D [11:44:58] :3 [11:45:50] It's a amusing tweet, to be honest. [11:46:47] Nothing special on Q69 though. :) [11:49:32] xD [11:56:46] Very much a missed opportunity with Q69 and Q420. [11:56:51] The funny numbers [11:57:56] Or Q58008 :) [11:58:45] thanks addshore, awesome! :) [11:58:54] [= [12:05:47] hoo|away: how do I easilly add records to the sites table? for local wikis & testiung? [12:07:09] addshore: DBSiteStore, I guess [12:12:57] addshore: https://gerrit.wikimedia.org/r/313976 Could you review that as well? Only move stuff around [12:13:02] * moves [12:13:11] but I want to build on that for further changes [12:23:03] addshore: https://gist.github.com/filbertkm/0254b6fc417ae12bc1b1 [12:23:24] modify how you like and then use importSites.php script in core [12:23:56] * aude would like this to be easier, of course [12:30:52] eval.php aude [12:30:54] https://www.irccloud.com/pastebin/7a0M5BqI/ [12:33:35] heading out to lucnh! [12:34:04] addshore: heh [13:00:42] Anyone who might be able to answer: I understand that Wikibase does not expose raw JSON, but does it nonetheless have a JSON Schema validation system? [13:01:33] hare: Well, kind of [13:02:02] we convert everything into our own (PHP) object structure and create json from there [13:02:02] I ask because there's a MediaWiki RFC on JSON validation which might be relevant: https://phabricator.wikimedia.org/T147137 [13:02:13] on user input json, we convert that into objects and then to JSON [13:02:21] that way we can be sure it's ok [14:46:35] hoo: aude do you know of a way to make sure an mediawiki hook is the last hook of a type to be fired? [14:47:10] addshore: Nothing that is not batshit insane [15:13:23] :/ [15:13:28] hoo: whats your batshit insane idea? [15:13:46] im looking at the BeforeInitialize hook right now to maybe add it / ensure its at the end...) [15:17:17] addshore: (In your hook) $hookFunction = 'Foo::bar'; if ( end( $wgHooks['hook'] !== $hookFunction = { wgHooks['hook'][] = $hookFunction; … } [15:17:45] also need to remove the other occurrence of your hook [15:26:48] hmm, would the list of hooks be re read when adjusted though? :/ [15:43:16] hoo: https://gerrit.wikimedia.org/r/#/c/313979/ if your interested in the review of what we just discussed [15:45:59] So you're factoring that out of Wikibase? [15:48:08] yup [15:48:28] as this Cognate thing will need to do it too, and there were TODOS everywhere about this should probably be its own extension! [16:02:41] DanielK_WMDE: did you have a mechanism in mind for parsing the Wiktionary pages from Cognate? [18:59:29] hi guys, is there a way I can see a content of deleted item? [18:59:38] sparql query gave me an item: Q11814034 [18:59:49] but I am unable to find it through regular search [19:00:05] it was deleted, yes [19:00:07] Just use the URL: https://www.wikidata.org/wiki/Q11814034 [19:00:11] there we no sitelinks but some statements [19:00:20] [19:00:21] Paweł Dziubiński [19:00:21] Polish chess player [19:00:33] thanks [19:00:38] np [19:00:44] For the search, see https://phabricator.wikimedia.org/T129779 [19:01:16] (It's not possible to navigate to a deleted item via Special:Search) [19:03:43] ty, I will subscribe to that phab task [19:07:02] sjoerddebruin: Any idea why the channel is +r? Spammer again? [19:07:29] multichill: no idea [20:37:31] Hi there. How can I obtain, given the Wikipedia ID, the Wikidata URI? In DBpedia there is the "dbo:wikiPageID" property which let me link the two resources. [20:38:52] jewels: There are various ways… do you want a user interface or rather an API to use? [20:40:31] hoo: I want to obtain it using a SPARQL query because is an operation that I want to do in the backend [20:41:08] jewels: It can be done via SPARQL, but if you want to do just that [20:41:16] the API is the better solution [20:41:26] it's faster and more reliable for simple accesses like these [20:42:40] hoo: Yes SPARQL it's preferable for me because I use SPARQL queries for other purposes. Can you suggest to me how to obtain it? [20:57:45] SELECT ?item WHERE [20:57:45] { [20:57:45] ?sitelink schema:about ?item . [20:57:45] ?sitelink schema:isPartOf . [20:57:45] FILTER ( str(?sitelink) = "https://pfl.wikipedia.org/wiki/Berlin" ) [20:57:45] } LIMIT 1 [20:57:50] jewels: ^ [20:57:59] That's not particularly fast, though [20:59:22] (Wont work for en.wikipedia.org or other large sites) [20:59:43] hoo: Ok thank you! [21:00:10] hoo: At the moment is there a SPARQL endpoint which let me access information in DBpedia endpoint and Wikidata endpoint? [21:01:10] Not that I'm aware of [21:01:23] this is much simpler: [21:01:23] SELECT ?item WHERE { schema:about ?item. } [21:01:46] no need to stringify the sitelink, it’s already an URI :) [21:04:31] WikidataFacts: Thank you for that. Much cleaner solution. What about my previous question: At the moment is there a SPARQL endpoint which let me access information in DBpedia endpoint and Wikidata endpoint? [21:05:58] not that I know of [21:11:28] WikidataFacts: Really strange. I think that can be really useful to combine the two source of information. [21:11:57] I’m not saying it doesn’t exist, I just don’t know anything about DBpedia [21:12:36] they seem to have a endpoint [21:13:30] WDQS doesn’t allow federation, but perhaps it works the other way around? [21:14:57] Exists a DBpedia SPARQL endpoint but I don't know if it is possible to query Wikidata too