[07:41:25] for a software version it makes more sense to me to use "publication date" as a qualifier instead of "point in time" ( I'm asking because of https://www.wikidata.org/w/index.php?title=Q541691&curid=510202&diff=378338826&oldid=378338573 ). What do people think ? [09:24:51] dachary: publication date seems more meaningful to me... point in time sounds weird (you could argue that at that point in time the software version is X, but then why not use "start date"?) [09:25:42] basically, point in time could be any point in time that the statement is valid, but that's not the intended meaning [09:26:10] He could still use "retrieved" in source, but then he should add some source in some way. [09:26:30] nikki: I can remember a discussion about "official website" and this, though. [09:26:58] dachary was asking about the date qualifier to use for software versions [09:27:21] nikki: sjoerddebruin thanks for the confirmation :-) [09:27:49] nikki: I know, he used "retrieved" as qualifier before that but that was hitting constraints. [09:28:08] ah [09:28:11] I didn't know about that [09:28:56] https://www.wikidata.org/w/index.php?diff=377886591&oldid=377673512&title=Q24448646&type=revision [09:31:14] sjoerddebruin: I documented this mistake in the request for permission I filed this morning at https://www.wikidata.org/wiki/Wikidata:Requests_for_permissions/Bot/FLOSSbot_3 . Thanks again for spotting it quickly. [10:03:57] Thiemo_WMDE: thanks <3 [10:04:59] How did this made it's way into master? There should be a test running on the CI that makes sure stuff is compatible with PHP 5.5. [11:36:40] DanielK_WMDE: Hey, one question: What do you mean exactly by "adapter"? https://phabricator.wikimedia.org/T135146 [11:38:49] Thiemo_WMDE: We have some CI tests in ORES extension, but I should write more tests [11:39:07] Amir1: Ah, I see. Ok. [11:55:44] Amir1: https://en.wikipedia.org/wiki/Adapter_pattern [11:57:06] Amir1: the article makes it sound complicated. it's really trivial. You implement InterwikiLookup, and your implementation gets a SiteLookup to work with. [11:57:33] So what you write is an adapter for InterwikiLookup on a SiteLookup. [12:04:31] DanielK_WMDE are you planning on implementing something like [[:entity:de:foo]], which would link to the equivalent on dewiki of [[:currentwiki:foo]] [12:04:32] 10[9] 04https://www.wikidata.org/wiki/:entity:de:foo13 => [12:04:35] 10[10] 04https://www.wikidata.org/wiki/:currentwiki:foo [12:04:59] (if so, thumbs up ^^ ) [12:07:19] Wikidata power http://wiki.remixthecommons.org/api.php?action=query&meta=siteinfo&siprop=interwikimap&sifilteriw=local [13:44:05] Wish https://www.wikidata.org/wiki/Property:P1534 was more used, very useful property. [14:17:03] sjoerddebruin: maybe you can add a suggestion to the "end time" property https://www.wikidata.org/wiki/Property:P582 that one might want to add P1534 as well [14:17:17] I guess by just dediting the description [14:17:37] I am not sure whether there is a mecanism to hint one about another prop when filling in via the web interface [14:17:58] Well, that would be great for qualifiers and sources. [15:09:26] sjoerddebruin: shall you create a warning? [15:10:11] matej_suchanek: I could write a text yeah [15:10:30] https://www.wikidata.org/wiki/MediaWiki:Abusefilter-warning-92 is for you [15:10:47] and you can also make a Dutch version on "/nl" subpage [15:10:54] :) [15:14:40] matej_suchanek: do you know if I can use templates like {{Q}}? [15:14:41] 10[11] 10https://www.wikidata.org/wiki/Template:Q [15:14:53] I suppose yes [15:15:38] btw you can already test the filter although I'm sure what it does with a not yet created warning [15:15:52] I think it takes the default one [15:16:10] even if I set a different one... [15:16:15] No, just "ā§¼abusefilter-warning-92ā§½" [15:16:40] ah, yes [15:16:56] which is "correct" [15:18:43] by the way, the discussion "system" on [[Wikidata talk:Abuse filter]] is really weird as it's mixed [15:19:09] Yeah, could look at that too. The page is very hidden too. [15:19:49] also we have many filters which I don't belive could be used by anyone [15:20:03] having them just slows editing [15:20:07] :O [15:20:14] Abuse filter needs better support for Wikibase imo [15:20:45] or contra :) [15:23:03] good candidates for turning of are #1, #19, #20 and mybe others [15:23:07] Is https://www.wikidata.org/wiki/MediaWiki:Abusefilter-warning-92 clear enough? [15:24:31] I usually add a '''Warning:''' in front of it but that's not that important [15:24:59] btw do you expect the changes to have a tag? [15:25:00] Warning sounds weird in this context imo [15:25:33] Well there is already a bot for fixing them for people after that date [15:25:42] Don't need a change tag imo [15:25:47] k [15:26:05] Is it possible to fetch the value of birth date though? [15:26:40] well, I can check that it exists but parsing the fake wikitext is almost impossible :/ [15:26:45] * matej_suchanek won't be at home during the next office hour :/ [15:27:06] but if you told me more, I could try [15:28:24] Nvm [17:30:31] SMalyshev: Did someone already mention that the example link at https://query.wikidata.org/ still points to https://www.mediawiki.org/wiki/Wikibase/Indexing/SPARQL_Query_Examples? [17:30:51] multichill: hmm no which link? [17:31:19] ahh I see in the menu [17:31:26] thanks, forgot about it, will fix [18:30:16] SMalyshev: List is getting quite impressive [18:30:54] Should probably add some more weird edge cases to help people figure those out [18:31:11] sure :) [18:32:05] http://tinyurl.com/zfn9gm4 http://tinyurl.com/hw7gulg what are the odds that wikimedia pages entities have no sitelinks whatsoever ? ;) [18:33:42] Alphos: that doesn't look normal, probably result of deletions [18:34:28] some false positives, some true positives [18:35:00] https://www.wikidata.org/wiki/Q26945333 is a true positive for instance [18:35:54] yeah I don't know what's going on there [18:36:16] looks like somebody just created an item for it, which is probably wrong [18:36:27] this is not how it's supposed to work [18:36:58] agree [18:38:05] true positive that shows potentially malicious intent, or, if anything, sandboxing https://www.wikidata.org/w/index.php?title=Q1043459&action=history [18:39:54] required the creation of a new item too https://www.wikidata.org/wiki/Q15960093 [18:40:09] in general, if a page is marked as wikimedia and no links, it's in 99% cases a ticket to RFD [18:40:36] in that case i'd like to think it calls for a merge [18:41:19] yeah Q1043459 and Q15960093 definitely needs to be merged [18:41:46] or actually Q1043459 just deleted, it doesn't have any content [18:43:07] https://www.wikidata.org/w/index.php?title=Q1854039&diff=371126363&oldid=334143602 [18:43:26] SMalyshev it's the old item, i think it deserves some respect :p [18:43:50] this one looks like vandalism [18:44:07] yup [18:44:40] although i should point out that i'm unsure as to what two do with multi-part lists [18:45:07] this one pointed to [[some list (A-I)]] [18:45:07] 10[12] 04https://www.wikidata.org/wiki/some_list_%28A%2DI%29 [18:45:32] well, I'm not 100% sure either but if they exist on wiki then I think wikidata should reflect that [18:45:54] if people on wiki decide to change it or we have better model we can amend it but right now it's page like any other [18:46:18] indeed [18:47:20] admittedly, pages 1 to n-1 in a set of n pages representing a list should point to the next page in the set [18:48:04] but that should be implemented as a policy, not a wikidata policy, shouldn't it ? [18:49:05] wikidata can have next/previous, there are properties for that - or can be crated if existing ones don't fit [19:02:09] SMalyshev yes, but if we have "list" that points to "list (A-I)", we need a way to have the rest of the list ;-) [19:02:35] followed by? [19:10:13] No comments after 9 days https://www.wikidata.org/wiki/Wikidata:Property_proposal/Nationalencyklopedin_Online_ID SHould I mark it as ready anyways, or wait for someone to support/endorese? [19:10:59] Quite unusual that nobody responds. [20:05:25] DanielK_WMDE: Thanks. I work on it in 23 minutes [20:05:33] (watching a movie atm) [20:05:46] Also the other patch is ready I guess [20:18:37] DanielK_WMDE: what you think is the proper process for deploying units config on wikidata> [20:18:39] ? [20:18:55] the problem is I can't really test it unless it's deployed on wikidata... [20:55:58] How would I look up all statements that have P248:Q26822184 as references? [21:00:26] hare: https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries/examples#number_of_statements_backed_by_a_reference_with_a_DOI might help [21:02:37] looks good to me, thank you [21:04:03] I got as far as http://tinyurl.com/z8sognm ... if that's right, that's a lot of statements :) [21:06:17] hare http://tinyurl.com/zlw3ktg mine's still running :/ [21:07:04] Would it be satisfactory to just say that Q26822184 is used as a source in hundreds of thousands of statements? :) [21:07:27] i think it would [21:07:34] do i smell a new property incoming ? :D [21:07:42] the property is P2880 [21:07:46] doh [21:10:08] hare ran another query, it would be satisfactory indeed http://tinyurl.com/zgou62b [21:10:25] 263609 Results in 8664 ms [21:11:38] oddly enough, with a DISTINCT, you get *more* results. http://tinyurl.com/j8hpwqc 263670 Results in 8842 ms [21:12:05] but regardless, there are still hundreds of thousands of statements using Q26822184 as P248 [21:13:06] Now, this might be more challenging, but I want all the statements that have a reference URL that begins with "http://www.cdc.gov/niosh/" [21:16:02] http://tinyurl.com/jfc4pp7 [21:16:17] Amazing! Thank you! [21:16:18] 24900 Results in 16430 ms [21:16:34] (string functions are notably slow, which isn't unexpected) [21:16:38] heh, I had http://tinyurl.com/gr6t62l [21:17:07] I have no idea why the short url link is creating links to the embed thing [21:17:20] nikki you don't need ?entry, so you don't need the first line of your WHERE ;) [21:17:34] I just started with the thing I linked above [21:17:38] ^^ [21:17:46] i did too, but don't tell anyone :D [21:19:27] "prov:wasDerivedFrom/pr:P854 ?ref" seems slightly slower than "prov:wasDerivedFrom [ pr:P854 ?ref ]" [21:19:37] 24900 Results in 18529 ms (instead of 16430 ms) [21:19:46] It's too bad we can't get anything more useful out of the Statement thing [21:19:59] hare what would you like to get out of the statement thing ? [21:20:39] maybe you could add/modify some example queries to use your method :) [21:20:52] I'm not actually sure what the difference is [21:21:19] nikki not even sure it's statistically significative [21:22:08] anyways, I did a union between a query I came up with (statements that cite an item that have a P2880) and your query, and in total, NIOSH is used as a source for 24,943 statements, not including the NIOSHTIC-2 bibliographic database which is used 200,000+ times as a reference for statements (for items about academic works) [21:22:15] My superiors will be pleased to hear this. Thank you for your assistance [21:23:38] hare make sure you didn't select items with both kinds of statements, if that matters to you [21:23:56] should I do select distinct then? [21:24:12] i think there are very few that have both kinds [21:24:57] hare do you want items that have both kinds of statements, of items with single statements that fit both ? [21:25:15] individual statements that have a, b, or both [21:25:24] right then [21:25:25] here is the current query: [21:25:27] https://query.wikidata.org/#SELECT%20%3Fentry%20%3FentryLabel%20%3Fstatement%20WHERE%20%7B%0A%20%20%3Fentry%20%3Fp%20%3Fstatement%20.%0A%20%20%7B%0A%20%20%20%20%3Fstatement%20prov%3AwasDerivedFrom%2F%0A%20%20%20%20%20%20%20%20%20%3Chttp%3A%2F%2Fwww.wikidata.org%2Fprop%2Freference%2FP248%3E%2F%0A%20%20%20%20%20%20%20%20%20wdt%3AP2880%20%3Fn%20.%0A%20%20%7D%20 [21:25:27] UNION%20%7B%0A%20%20%09%3Fstatement%20prov%3AwasDerivedFrom%20%5B%20pr%3AP854%20%3Fref%20%5D%20.%0A%20%20%09FILTER%20%28STRSTARTS%28str%28%3Fref%29%2C%27http%3A%2F%2Fwww.cdc.gov%2Fniosh%2F%27%29%29%20.%20%0A%20%20%7D%0A%20%20SERVICE%20wikibase%3Alabel%20%7B%0A%20%20%20%20%20bd%3AserviceParam%20wikibase%3Alanguage%20%22en%22%20.%0A%20%7D%0A%7D%0AORDER%20BY%20 [21:25:27] %3FentryLabel [21:25:29] bah [21:25:39] http://tinyurl.com/he6op6f [21:27:17] oh, I just meant that including a variety of approaches just seems useful so that people can find other ways to do things, which might turn out to be helpful at some point [21:28:14] nikki indeed. [] isn't well documented or represented in /queries, and that's a darn shame [21:30:14] hare i'd go with http://tinyurl.com/jjx72ry 24813 Results in 18997 ms [21:30:28] what does this do different? [21:31:48] speed, if anything [21:31:55] SERVICE is notably slow ;) [21:33:18] its main advantage is that if you want fallback, it slows down less than multiple calls to rdfs:label and a COALESCE() [21:35:23] 24813 Results in 17593 ms with DISTINCT and OPTIONAL { rdfs:label } http://tinyurl.com/gnenxkb ; compare with 24813 Results in 20087 ms with DISTINCT and SERVICE [21:35:46] (the small difference is because lots of items are identical. for more items, it grows quickly) [21:36:12] SMalyshev: not sure... maybe ask aude :) [21:36:41] ok :) [21:39:04] hare as a very general rule, for n languages, with n up to 3, if you hit a time constraint, try multiple OPTIONAL{ ?item rdfs ?itemLabelLg1 . FILTER( LANG( ?itemLabelLg1 ) = "language1" . }, followed with BIND(COALESCE(?itemLabelLg1,?itemLabelLg2, ...) AS ?itemLabel ) instead of SERVICE [22:28:57] Can't someone create a tool/script to more easially submit new property sugestions. like a window with drop down menus and regex tester etc.. [22:29:02] :/ [22:29:27] http://www.ubio.org/browser/details.php?namebankID=6218218 ... [22:29:47] The whole property process should be more streamlined yeah [22:55:37] I don't think this is the right thing to add.... https://www.wikidata.org/w/index.php?title=Q15268585&diff=378820439&oldid=329672801 [23:05:44] how should I use "no value" for a property ? It could mean "I've checked and there is no value for this property". For instance, if a given software that is free software is not listed in the Free Software Directory ( P2537 ). [23:08:09] which is different from not having a P2537 at all, meaning we do not know if it exists or not [23:12:28] dachary: I think it is for when there 'should exist'. Like countries should have "official langauge" and ountries should have a "currency". But if theyre were not to have one, then no value should be used...I don't think it should be used for when "Optional things" does not exists. Like otherwise I could add "currency:no value" to an item about a person... [23:12:39] countries [23:12:41] *