[04:56:57] !admin [05:02:24] Danmichaelo ? er du her ? [05:03:47] noen admins på NO:WP her ? [05:25:41] any admins on Norwegian wikipedia ? [05:45:18] seems like the 'problem has gone away... and found something else to do instead of vandalising... [12:40:42] Lydia_WMDE: are items like https://www.wikidata.org/wiki/Q21686005 (instead of using novalue) an acceptable way to model things like that? [12:42:49] nikki: it seems fine to me. novalue would probably also work. i'm wondering if there is another reason a movie might not have a language besides it being a silent movie [12:43:35] purely visual movies? some documentaries without voiceover? [12:43:48] *nod* [12:44:05] videos of firesides? ;-) [12:44:09] well, the statement simply says what the language is, whether it's a silent film or not should be a separate statement [12:44:38] that's also a fair point [12:44:54] well the item seems to specify that it’s a silent film [12:45:01] i guess in the end it is similar to religion -> atheism [12:45:45] and it seems wrong to me to use somevalue to mean novalue :/ but the person who created that keeps creating/using items like that because they don't like novalue because you can't add statements to it [12:46:48] ? [12:46:56] ah [12:47:01] meh [12:47:40] like they also use "statelessness" as a country of citizenship, when by definition there is no country of citizenship if someone is stateless [12:48:10] right [12:48:22] that is a case for novalue [12:48:42] wait, I don’t understand what the concern is [12:48:47] “no value” can have qualifiers: http://tinyurl.com/z3fnecz [12:49:02] WikidataFacts: but you can't add statements to "no value" because it is not an item [12:49:09] i think that is a bit... [12:49:13] but oh well [12:50:10] well yes, but it’s not like adding “facet of: silent film” to no value would be correct either [12:50:16] we only have one novalue [12:50:33] yes [12:51:05] I think their point is that there's no connection between country of citizenship = novalue and the statelessness item, there's no link between language: novalue and silent film [12:51:24] where would it be needed? [12:51:39] not sure, I'm not the one who wants the link [12:51:49] yeah [12:52:06] but I could maybe see an infobox wanting to link to the page about statelessness if country of citizenship is novalue [12:52:20] but that can easily be done in the infobox [12:52:30] via lua and templates [12:53:47] yeah... although it would have to be done on a case by case basis since you can't extract it automatically [12:54:11] yes but i think that is needed anyway [12:56:02] yeah I guess for things like language there isn't a single meaning for novalue, films might be silent film but a book would probably be picture book or something like that [12:56:12] not a silent film in any case [12:58:27] somewhere I suggested maybe adding a property for properties to link to the item which describes the meaning of novalue [12:59:04] but I don't know if they saw it or responded [13:03:14] nikki: Do you happen to have figured out if it's possible to use regex in sparql to change a field? [13:03:34] what do you mean by "to change a field"? [13:04:19] Search works with FILTER(), replace works with REPLACE(), but not regex. I'm looking for a regex replace in a field. For example "NG(\d+)", "\\1" to replace NG123 with 123 [13:04:46] nikki: To extract the integer from http://tinyurl.com/zvvsvdf and sort by it [13:04:57] multichill: REPLACE accepts a regex, doesn’t it? [13:05:20] I definitely know there’s *some* regex-replace function, I’ve used it (backreferences in replacement with $1, $2 iirc) [13:05:21] Does it? What kind of syntax? [13:05:31] Are just no backreferences? [13:05:38] it does allow a regex in the find bit, not sure about backreferences [13:05:40] no, either with () or  [13:05:45] and backreferences with $1, $2 [13:05:56] $1? Ah, that explains it ;-) [13:06:55] heh yeah, I was about to try $1 [13:07:28] And everything double escaped. Joy [13:07:33] but within the regex, backreferences are still \ (that is, \\) [13:07:46] http://tinyurl.com/hbm3uaa [13:08:39] oh, perl does that (well, not the double-escaping) [13:09:38] O [13:10:13] I'm pretty sure I RTFM'd for the regex part, but missed the replace part being $1 instead of \\1 [13:10:27] yeah, you have to follow the link to the XPath spec for that :D [13:10:48] because the SPARQL spec just says “this is fn:replace” and that’s it :D [13:12:07] Yeah, the people who designed SPARQL are quite horrible [13:12:12] ;-) [13:12:42] WikidataFacts / nikki : Thank you! The sorting at https://www.wikidata.org/wiki/User:Multichill/Paintings_in_the_National_Gallery is now much better [13:12:53] :) [13:12:57] (ignore the first couple, those don't have an inventory number yet) [13:13:18] * WikidataFacts waits until the browser stops hanging to look at the change [13:14:21] ah, nice :) [13:16:07] A lot of their works went to Tate. Tate was nice enought to keep the same number so now you can see how https://github.com/tategallery/collection/tree/master/artworks/n fit nicely in the holes of above list [17:39:09] Hi, guys. I'm going to make a couple of categories in ruwiki and bound them to wikidata elements. Should I get bot flag or something in wikidata before doing that? [17:41:13] I mean, I'm going to do it automatically, using bot. [17:47:14] Never mind, I found info. [18:03:39] Hello everyone. I am part of a research project into sparql-queries posed to wikidata. We want to analyze these queries using an open-souce java library (looking at Apache Jena and OpenRDF at the moment) but ran into a problem with this kind of query: "SELECT ?item WHERE { ?tree wdt:P31 ?item . BIND (wd:Q146 AS ?tree) }", where our parsers complain about the variable ?tree being bound again in the BIND-statement. If I en [18:03:39] ter this into query.wikidata.org it works, so I was wondering if anyone could point me to the library wikidata uses to parse the input query string. [18:08:57] arnad: the backend is BlazeGraph [18:09:45] though I’m not sure why BlazeGraph allows this – the spec quite clearly states that the variable must be new [18:25:14] That's whats throwing me of as well. The weird thing is that we are using the sparql-grammar maven dependency from blazegraph and are parsing queries with org.openrdf.query.parser.QueryParserUtil.parseQuery(), but this does not accept the above query. If you have any idea on which package is used to parse the queries (or a pointer where I could look) I'd be grateful. [19:24:11] I wonder how to best model "Summer 1886"..... [21:28:59] is the a pythonic way to get the last created Q-number of wikidata? [21:39:51] E-doderoo: As in, using pywikibot? I would use the new articles generator I think [21:40:22] ah, yeah, that is a good way [21:41:14] -newpages:1 just gives the last one I think [21:41:33] What do you want to do exactly? [21:42:16] I have some queries to go along all items, backwards... then you need the newest one to start with [21:43:08] I now look it up manually and enter that in the source ... but that is the amateur way of programming ;-) [21:44:05] Actually, it seems to be a wrapper around site.recentchanges(changetype='new', namespaces=[0], total=1)