[07:12:27] Harmonia_Boulot, hi, could you do https://www.wikidata.org/wiki/User_talk:Jitrixis/nameGuzzler.js too? Seems Jitrixis hasn't edited since February [07:13:33] Harmonia_Boulot, oh, and afterwards do this for your nameGuzzlerOptions.js: https://www.wikidata.org/w/index.php?title=User%3AJon_Harald_S%C3%B8by%2FnameGuzzlerOption.js&type=revision&diff=496742285&oldid=496742118 :) [07:14:20] I deleted some of these because they weren't handled correctly by the interface (is that also the name in English?) [07:15:15] and for nameguzzler, are you sure in these languages they keep the order "given name-surname" ? [07:15:53] I use nameguzzler to add names for people mostly, so I don't have all Latin-scripts languages checked [07:16:13] or I would make errors [07:23:54] Harmonia_Boulot, yup, very sure [07:25:02] Jhs: for Jitrixis, let a message on the discussion page [07:25:29] i pinged him in the request, so (s)he'll get the same notification anyways [10:36:57] any sparql whizzes here today? How do I build a query that searches for labels in Norwegian OR English that end in the string "Blakstad"? This is what I have so far, but it only works for one language. i want it to work for two (actually three, but who's counting?) [10:36:58] http://tinyurl.com/y763ag9c [10:44:29] abbe98[m], ^ :) [10:45:12] Lucas_WMDE: something for you? ---^ [10:45:25] Jhs: http://tinyurl.com/y7s2mhns [10:45:37] DanielK_WMDE_: I was just about to reply :) [10:46:09] Lucas_WMDE, DanielK_WMDE_ :D [10:46:13] thank you very much! [10:46:30] Jhs: to eliminate the duplicate results: http://tinyurl.com/yckfd43s [10:46:55] awesome! [10:47:39] now for the wishlist: a metatool with a GUI to build sparql queries ;) [10:49:00] Jhs: i have never seen a query builder that was actually useful. either they only allow trivial queries, so you don't need them, or they are so complicated, typing is easier. [10:49:41] DanielK_WMDE_, most of the time I just need trivial queries, but i forget the syntax so i have to look it up every time [10:50:38] (note: this one does not count as a trivial query in my eyes though) [10:51:07] Jhs: you should make you a subpage with examples [10:51:23] I have a dozen of maintenance queries I run every day [10:51:35] I don't bother with writing them up every day [10:51:44] I saved them in a subpage ;) [10:52:21] Harmonia_Boulot, good idea :) [10:52:26] did that too for really long or complicated ones [10:52:44] easier to amend the existing query than beginning from scratch [10:53:09] true that [11:19:25] Lucas_WMDE, final question for today, i promise. How do I find only the ones without a P734 statement? [11:19:48] I tried to add a line "rdf:type wdno:P734;", but it doesn't work [11:20:21] that selects for statements with “novalue” [11:20:33] missing statements is done with MINUS: http://tinyurl.com/y9abxmqh [11:20:34] ah [11:20:50] (or FILTER NOT EXISTS, but that’s slower for whatever reason) [11:21:01] danke schön :D [11:21:05] Lucas_WMDE: depend of the query [11:21:15] sometimes it's MINUS that is slower [11:21:39] pretty rarely, in my experience [11:21:51] though I don’t usually try both out, I just use MINUS [11:22:17] I try both for really heavy query whith timeout [11:22:28] and i have been surprised in both ways [11:22:49] urgh, my English is all mangled today [11:22:53] sorry [11:29:24] Lucas_WMDE, just a note: the query you gave me before had the first line as "SELECT DISTINCT ?item (STR(?itemLabel) AS ?itemLabel_ )" [11:29:36] however, PetScan for some reason strips away the underscore _ [11:29:44] huh [11:29:53] so for future reference, i think it's wiser to use ?itemLabelX or something [11:29:57] you can choose any other name, e. g. ?itemLabelWithoutLanguage [11:29:59] yeah [11:30:05] yeah, i was struggling to see what I did wrong there for a few minutes :P [11:30:29] it works the first time, but when you retry it the underscore is gone [11:30:54] Jhs: yup, no underscore in SPARQL queries for Petscan [11:30:58] it never works [11:32:09] hm, that sounds like a problem… underscores are also used in blank nodes [11:32:23] e. g. open http://tinyurl.com/yce4ad55, edit it with the query helper, and see what happens [14:08:13] Lucas_WMDE: I could be completely wrong but I always assumed that minus tends to be faster because it's often simpler (do this simple query, then remove the ids which are returned from the original data), whereas filter not exists goes the other way (take this existing data, apply these extra triples and remove the original data if the extra triples are true) [14:10:34] (although there's plenty of times where minus has been slower, *because* it's not dependent on the existing query... like a regex on a subset of 10 results is almost certainly going to be faster than applying that regex to everything and removing the matches from the original subset :P) [14:12:37] nikki: I always assumed that BlazeGraph doesn’t actually run the query that way, because MINUS { ?item wdt:P31 wd:Q5. } seems to be fast, but finding and subtracting a few million items feels like it should take longer [14:13:01] yeah, blazegraph does strange things [14:13:24] like I still don't understand how combining two things which run really fast can result in something that times out [14:14:18] it does seem like it likes simple things though, even if it involves a lot of data [14:16:51] which feels rather odd and counterintuitive to me... "return a million results with this single triple" seems crazy to me, but it seems perfectly happy to, but then return a couple of thousand results with a few triples? what sort of absurd request is that? [14:17:44] some day I’ll have to sit down, stare at EXPLAIN output for a few hours, and try to understand what the hell it means [14:18:01] that would probably be super useful for understanding performance [15:22:22] why don't we have a property uuid ? [15:25:42] Probably because it is a universal identifier. [17:19:21] Creative, but not correct. https://www.wikidata.org/w/index.php?title=Q3185132&type=revision&diff=497020054&oldid=470351001 [19:23:52] WikidataFacts: maybe you can help me :P I have http://tinyurl.com/y7vtbl5n which works fine and want to change it to de-at, but that times out. I've tried various things like turning the optimiser off, filtering more, filtering less, but I can't get it to work [19:26:43] nikki: both de to de-at? or just one of them? [19:28:49] both [19:29:05] basically looking for any descriptions other than the "proper" one [19:29:49] if you remove the description, it works, with >350k results: http://tinyurl.com/y89qwmo6 [19:30:05] so that’s probably too many items to get the de-at description for [19:37:44] it works for de though :/ [19:38:01] probably because the minus removes enough items, I’d assume [19:38:15] but most items will just have no de-at description at all, right? [19:38:32] probably [19:38:47] hard to find out when all the queries I try time out :P [19:40:03] I wish I could just search for items with a label or description or whatever in a particular language, without having to select it and filter by language [19:40:27] I bet if I could do ?item something:hasdescriptioninlang "de-at" it'd be perfectly able to find what I want :P [19:41:22] yeah, that would be sweet [19:41:41] or ?item schema:description-de-at ?description [19:52:08] wow, just finding 25 items takes almost 60 seconds: http://tinyurl.com/y9zkrymu [19:52:21] (nikki: perhaps your version is better, I haven’t tried that one with a LIMIT) [19:52:56] is that url right? [19:53:54] *sigh* noscript… [19:55:17] nikki: try this one https://tinyurl.com/yc96lmz7 [19:55:38] yeah, that looks more like a query :D [21:30:21] Hi, I could use help with this query: http://tinyurl.com/yb7fg3jo How can I get the Software Version "1.5" from this Item Q19060005 instead "Android Wear 1.5"? [21:52:20] WikidataFacts: I just made https://phabricator.wikimedia.org/T167361 if you have any more examples to add [21:54:35] nikki: I’ll try to remember, thanks for creating the task [21:59:22] How can I show a qualifier on a query? [22:24:27] puchrojo: use pq prefix; ?item p:Pxxx ?statement . ?statement pq:Pyyy ?qualifierValue . [22:24:59] uhm ... I hope you don't see the smilies I see^^ [22:43:19] MisterSynergy: Thanks, I search more than a hour. Now I can go sleep.