[09:40:28] Lucas_WMDE: the link in that lexemes that need senses topic would be more useful if it only listed ones without senses :< [09:40:40] I thought maybe we could use the query service, but I can't seem to get it to work [09:40:58] one query I came up with isn't supported by the mwapi stuff, the other one just returns empty rows [09:48:13] nikki: what part do you want to use mwapi for? [09:48:22] “any lexeme without senses” should be possible without it [09:48:30] “your own” I don’t see how that could be done, even with mwapi [09:48:30] any lexeme I created [09:48:34] ok [09:48:52] (perhaps it’s possible if you hard-code the user name… hm) [09:49:00] you can turn the link you made into an api query, but the query service doesn't support it because it doesn't seem to be compatible with generators :( [09:49:20] (unless I'm just confused by the documentation) [09:49:54] well, changing the username in the query would be easy enough for people to do :) [09:52:31] can you perhaps send me the API request if you already have it? [09:52:35] I’m not as familiar with the regular API :) [09:53:13] yes, need to get it off my other computer first though [09:54:02] okay, nevermind then, I think I have it already [09:54:13] generator=usercontribs, I assume [09:54:50] https://www.wikidata.org/w/api.php?action=query&list=usercontribs&ucnamespace=146&ucshow=new&uclimit=500&ucuser=usernamehere is the one I don't know how to use in the query service [09:55:42] https://www.wikidata.org/w/api.php?action=query&generator=allrevisions&garvlimit=50&garvnamespace=146garvuser=usernamehere should work with the query service (for ones the user edited, not just created) but I get empty rows [10:06:10] nikki: got it, http://tinyurl.com/yaoyjvlh [10:06:17] thanks for the allrevisions hint, I couldn’t get it working with usercontribs [10:06:43] no results for your username btw… does that mean I have a bug or did you just clean up after yourself? :) [10:07:26] must be a bug, there are definitely some I haven't done yet [10:07:34] hm [10:08:10] ah, usercontribs just isn’t available as a generator, that’s the problem with that ^^ [10:08:33] I wonder why it isn't [10:08:38] it would've been useful for us! [10:09:00] if I comment out the MINUS line, I get two lexemes for your username [10:09:05] Giraffe and Fledermaus [10:09:07] and nothing else [10:09:11] wat [10:09:49] those are the two I just edited [10:09:51] odd [10:10:04] ah, that explains another anomaly [10:10:15] I had just gotten an extra result for Nikki_ or _Nikki, which was really bizarre [10:10:24] but then that’s just another edit you just made? on Phereknong? [10:10:28] yep [10:10:29] *Pherekhong [10:10:31] ok [10:13:28] trying it out in the API sandbox, I get the same few results [10:13:33] and then a continuation [10:13:41] and the continuation returns no results, but I can continue it again [10:13:48] and after a few more continuations I get some more results again [10:14:10] I wonder if WDQS just terminates at the first empty result and ignores the continuation? [10:15:20] hm, sounds plausible [10:18:26] * nikki resorts to the good old-fashioned method of copy/paste/apply-regex-to-extract-IDs method for now XD [10:19:29] I can’t even find the MWAPI source code :/ [10:19:42] where the continuation should be implemented [10:23:45] ah, it helps if I update my copy of the source code… continuation support was only added later :D [10:26:45] haha, yes [10:38:21] https://phabricator.wikimedia.org/T209034 [10:41:13] thanks :) [11:15:50] I really want a tool for lowercasing labels... [12:38:57] Lucas_WMDE: since you know more about the development stuff than me, do you have any idea why https://gerrit.wikimedia.org/r/c/mediawiki/core/+/412337 hasn't been accepted yet? [12:39:15] * nikki finds gerrit very difficult to understand [12:42:47] no idea, to be honest [12:43:12] to me it looks like the discussion on Phabricator might not be settled yet, but I’m not sure [12:44:46] oh? [13:03:50] nikki: about that change, I would ask Nikerabbit or Siebrand [14:02:35] nikki: there is no consensus, and creating en-us has implication that has not been discussed or taken care of by the patch [14:06:57] Nikerabbit: are those implications mentioned anywhere? [14:07:14] I'm wondering because we keep getting requests for en-us [14:07:32] who is requesting it and what for? [14:08:41] people editing wikidata, to be able to enter american names of things [14:09:27] For that you could add en-us to Names.php, which would be a simpler case than what the patch does [14:10:04] Having a language listed Names.php doesn't yet make a language available as an interface language (although it is a requirement for that) [14:10:28] One small detail of that is that the fallbacks cannot be set [14:12:13] I assume people would want to be able to select en-us for the interface too, so that they would be shown the en-us labels [14:17:36] They could do that with babel, I think [14:18:12] like I said above, creating it as an interface language has questions like should we do something about the time formats in 'en' and similar stuff [14:19:53] and translatewiki.net configuration.. I am sure someone over eager will start saving 1:1 translations to en-us with current setup if we enabled it like this [16:21:30] addshore: or here :) [16:26:24] addshore, Lucas_WMDE: any opinions on the idea of having a seperate namespace for references? A la lexemes. There's a lot of de-duplication of references that can be done, but that the community doesn't want to exist as seperate items. [16:26:51] At the same time there is already a lot of bibliographic metadata on the main namespace and it is certainly uesful there. [16:27:43] But for instance, there's a gadget that duplicates references, because often the same reference is used on the multiple statements on the same item. [16:31:22] The answer is probably "no thanks" but thought I'd throw it out there :). [16:41:36] hi mvolz :D [16:41:49] :d [16:41:54] :D [16:42:09] this is one of the big questions [16:42:13] SO BIG [16:42:14] infact, the biggest question [16:42:19] THE BIGGEST [16:42:37] but its also funny, because there are essentially 2 or perhaps 3 possible answers [16:42:50] what sort of deduplication are you thinking of, that we wouldn't want as separate items? [16:44:50] Number of page of a reference, access date, etc. wouldn't deserve their own items [16:45:01] As an example: https://www.wikidata.org/wiki/Q18328015. [16:45:03] Or *entities [16:45:17] It has identical references for every statement. I used the copy gadget there. [16:45:34] There's definitely no reason to make that an item, it's just a url, stated in, and retrieval date. [16:45:50] More broadly for websites in general. [16:45:53] (and because one gadget is not enough, there’s also at least one user script for the same job that I’m aware of ^^ ) [16:46:14] personally I'd rather just have better tools for working with references [16:46:28] yes well that would be part of it, certainly, [16:47:26] the only advantage I can really see to having separate objects for them is that you could edit the object to update all the references (which could also be a really bad thing) [16:47:45] I think the granularity of what you would include in an "entity reference" vs what you would include as part of the reference on the item is an interesting question [16:47:49] where do you draw the line [16:48:13] Yup, and the thing is in citoid where the line is drawn is different for different item types according to Help:Sources [16:48:33] Lucas_WMDE: does it do exactly the same thing, or does it have some other useful functionality? [16:48:55] so, the easy question I was talking bout from earlier is actually the question about where to put all books, journals, scientific papaers / things you want to reference, in the item NS or in another NS :P [16:49:28] this question about what to do with actual references is harder [16:49:54] nikki: it has a different UI (drag+drop instead of clicking “copy” and “paste” links), but AFAIK it does the same thing [16:49:58] I would include everything as items if that wasn't a technical problem :) [16:50:39] Lucas_WMDE: ah, is that one of magnus's things? sounds vaguely familiar [16:50:44] abian: in my eyes its not really a technical problem, more a social problem [16:51:05] if we include 100,000,000 jouranl articles, socialy and within the community as things stand things would break down [16:51:19] nikki: yeah, I think it’s https://www.wikidata.org/wiki/User:Magnus_Manske/dragref.js [16:51:43] is there anyone arguing that we shouldn't use items for those things? and it seems to me that someone is trying to add 100,000,000 journal articles already :P [16:51:54] absolutely no consensus here: https://www.wikidata.org/wiki/Wikidata:Property_proposal/Container_title [16:52:03] and I need to do that approximately 20 more times [16:52:15] If the social problem is the amount of data, we'll probably be unable to manage it no matter the namespace :S [16:52:27] for about 20 more fields I'm not sure should be stored as a string in the reference or should be created as items [16:52:37] and I picked the most clear, simple case of all of those T_T [16:52:51] hmmmmm [16:53:09] 19 million Q13442814 so far [16:53:32] no wonder the random page link mostly takes me to articles [16:53:40] mvolz: do you need https://www.wikidata.org/wiki/Wikidata:Property_proposal/Container_title though if everything you are referencing is an item? [16:53:46] nikki: indeed [16:53:55] nikki: and imagine we had 10* that ;) [16:54:03] just a matter of time probably [16:54:05] addshore: the alernative is to create an item for every website ever. [16:54:17] :) [16:54:25] Are there more websites or are there more journal articles? Hmm. [16:54:33] definitely more webpages... [16:54:35] mvolz: well, every website you want to reference? :P [16:54:42] A slight improvement for references would be to use a hash table for each entity so that each identical reference is stored in the same data structure [16:54:44] addshore: ha, maybe not so bad. [16:54:52] i guess each web page wouldnt need an item, just the "container" of pages? [16:54:55] But just for each entity [16:55:26] abian: interesting idea to bring down disk usage :) [16:55:33] abian: that does solve the immediate issue of the duplication gadget [16:55:41] I guess which is not my main problem right now. [16:55:42] on the query service, something similar actually happens, except that the hash is global, not per-entity [16:55:48] nikki: indeed, people already want it [16:56:42] nikki: thats why I have been thinking of some way to create something that would work where each entity (and entity revision) essentially can be evaluated and assigned to a number of different groups for example [16:56:51] you can kind of think of it in terms of wikiprojects now [16:57:23] the wikiproject for roads might only want to focus on road related items, thats a pretty easy sparql query ti write, but how to tie that back into the rest of wikidata, editing etc [16:57:39] mvolz: well, you'd still need it to be able to link the same reference to multiple statements [16:57:52] and then more complex wikiprojects might have slightly more complicated criteria [16:58:45] Lucas_WMDE: I kind of want to be able to run a sparql query over a single item while being saved, but for every edit, wont scale :P [16:58:55] Lucas_WMDE: https://www.wikidata.org/wiki/User:Addshore/Lense [16:59:14] there are other more crappy ways to figure that sort of stuff out of course, but they are more crappy [16:59:56] hm [17:00:32] with something like that, it doesnt matter if there are 1000000000 journal article, as people that want to focus on them and the changes relating to them can, and people that dont want to can exclude them while they are trying to focus on other things [17:01:01] that could be useful for AbuseFilter too [17:01:18] instead of the current “regex on JSON diff” (I think?), which is pretty ridiculous https://www.wikidata.org/wiki/Special:AbuseFilter/104 [17:01:20] the more crappy way of trying to do something like that, is you could essentially tag items with other items....., perhaps not as a statement, but as something else within the entity, and then filter by them [17:01:31] yeh, lets not talk about abuse filter [17:01:36] :P [17:02:34] re abusefilter i need to get someone to do https://phabricator.wikimedia.org/T205254 [17:04:00] and then the other issue is other wikibase deploys [17:04:09] ? [17:04:09] Where they don't have the db that wikidata does [17:04:13] but still want to reference things [17:04:17] aaaaah [17:04:24] oof [17:04:26] federation, flavour 12532 [17:04:40] :D [17:04:41] InstantWikidataReferences [17:05:05] but that will happen, access to wikidata data from 3rd party wikibase installs over the api [17:05:08] no db access needed [17:05:12] I'm actually thinking too different export formats right now, wikibase, which is wikibase friendly for dates and stuff [17:05:14] and wikidata [17:05:30] which will actually use wikidata canonical [17:05:38] (for citoid) [17:05:50] Today's brainwave. [17:05:52] two* [17:06:06] hmm, wait, export formats? :P [17:06:12] for citoid [17:06:12] export formats for citoid? [17:06:15] gotcha [17:06:19] right, everything is strings [17:06:21] i should learn more about citoid to be honest :D [17:06:32] right well, all you need to know is, everything is strings [17:07:03] Would that be helpful to put in my wikicite presentation then? [17:07:08] mvolz: yes :P [17:07:27] mvolz: so this is citoid >> https://thumbs.dreamstime.com/z/ball-white-string-2420002.jpg ? [17:07:56] right then, in it goes. I don't even know if I have a presentation slot or how long it will be, did you get anything from the organisers for yourself? [17:08:07] addshore: :D [17:09:02] Sorry, quick question: is there an intention to address T100933 in depth, or a workaround with the special page would be fine? [17:09:03] T100933: [Story] Make it impossible to use Special:NewItem with the same content for label and description - https://phabricator.wikimedia.org/T100933 [17:09:07] addshore: can you clarify about "but that will happen, access to wikidata data from 3rd party wikibase installs over the api" is there a ticket for that somewhere? [17:09:10] mvolz: i got asked if I could do a wdqs / sparql presentation, ask dario, he will know! [17:09:47] abian: Lydia triaged it as high so I would say yes there is intention to fix it :) [17:10:00] mvolz: yes, let me dig it out [17:10:52] mvolz: https://phabricator.wikimedia.org/T48556 [17:11:33] addshore: I mean, changing all the interfaces or just the special page [17:12:20] Maybe I can help with the special page if only that is expected [17:12:53] Lydia also said "let's do this" to some task about adding label/desc/alias in a language that is not your babel two years ago. :) [17:13:24] sjoerddebruin: hehe [17:13:56] :o [17:14:08] I suppose there are too many tasks [17:14:27] Well, actually too few developers ;) [17:14:39] both! [17:14:39] :D [17:14:40] there are definitely many tasks :/ [17:14:49] abian: it looks like there needs to be a decision about what the message should say [17:15:17] Something descriptive like "Error." :) [17:15:20] and if it should be default behaviour, (i kind of feel like default behaviour should be fine)..? [17:15:30] also, should this only apply to items, or to properties too [17:16:52] Lucas_WMDE: hehe, if it shouldnt be in Wikibase, it could also be in the Wikidata.org repo, (evil laugh) [17:17:11] not many people can create properties so it's not really a problem there... and I don't recall seeing people adding the same label and description to existing ones [17:17:49] do we actually want it to error, or is there just something else wrong with the page? :P [17:18:16] * nikki is confused [17:18:32] addshore: What do you mean? [17:18:41] it could for example provide examples for what labels and descriptions should actually be :P [17:18:45] im just thinking out loud [17:20:21] People who currently are editing like this don't read, period. [17:20:44] That's sadly true [17:21:11] another thought, it could error the first time but if the user re submits allow it? i wonder if there is ever a case for that? [17:21:34] I can't think of any case [17:21:50] If you're going to write the same, you can keep the description empty [17:24:47] abian: i got the ball rolling ;) [17:24:47] https://gerrit.wikimedia.org/r/#/c/472500/ [17:26:07] addshore deserves applause :D [17:29:17] wow https://www.wikidata.org/wiki/Q57906199 [17:29:50] (found using http://tinyurl.com/y8w2qt4m) [17:30:02] some of these items actually have label = description in more than one language, eek [17:45:12] Lucas_WMDE: a d they all shouldn't? [17:45:14] *and [17:46:06] I haven’t looked at all of them in detail, but nothing there looked like it was correct [17:47:13] well, take https://www.wikidata.org/wiki/Q8062 for example [17:47:25] de label and description are both “economy of Ireland” (translated) [17:47:37] neither of them is wrong, but I guess the description isn’t useful there [17:47:38] Descriptions are some kind of definition, and it's always wrong for me to define something using exactly the same string [18:07:18] Lucas_WMDE: managed to add senses for all the lexemes I made :D [18:07:32] some are better than others though... [18:08:52] nikki: \o/ [18:08:57] I’m still way behind on mine [18:09:26] you were probably doing things you were supposed to be doing for most of the day though, unlike me :P [18:09:52] well, yes :) [18:11:12] BTW, sorry for interrupting you all with the task about labels === descriptions :/ [18:12:39] Yeh, gawd abian, so interrupting [18:12:39] * addshore is on a 🚆 [18:13:52] is that a bus or a train or a tram? [18:15:29] Good question, train [18:18:54] Right, I'm gonna sign off. Tata for now [18:19:03] see you [18:24:46] Infact, 1 final thing, who here is in the wikidata admin IRC channel [18:24:52] :p [18:26:13] me [18:28:55] * nikki prods addshore [18:28:58] nikki: can you invite me? :0 apparently I haven't been in it since I got my bit back [18:29:25] I have permissions but can't enter [18:30:00] Hehe [18:30:05] did that work? [18:30:39] Or sent invite exception on our hosts etc ;) [18:30:49] how do I do that D: [18:31:58] Mhmmgz I'm really off now, I'll poke you tommorrow :) [18:32:18] sure :) [18:54:47] addshore: how was the party? :D [20:10:21] It's a shame the London underground stations don't have a depth statement [20:14:06] addshore: wouldn't they need multiple statements? they're not all at the same depth [20:14:44] it seems the data is out there, if someone wants to add it [20:21:04] THE DATA IS OUT THERE [20:21:13] [ominous music] [21:07:07] Hello guys [21:55:08] hello, can anyone answer me this question https://www.wikidata.org/wiki/User_talk:Andreasmperu#Q7322 ? [21:57:40] Hi, leoncastro :) [21:57:51] hola abian [21:59:04] Andrea probably hasn't had time to answer yet... [22:00:09] Maybe we can wait a little [22:00:20] Anyway, can I help with that somehow? [22:03:05] well... she had done two hundred editions after that [22:03:26] i only need to understand [22:03:56] and better if you talk spanish [22:04:04] Vale :) [22:04:47] osea todo viene de que me deshizo dos veces una edicion [22:05:19] en la que finalmente argumenta «Either way, do not delete a referenced statement» [22:05:39] Sí, el tema es un poco confuso [22:06:33] Resulta que los datos afirmados en fuentes externas relevantes pero que se sabe que son falsos se suelen poner con el rango obsoleto ("deprecated") para indicar a la gente que, pese a que las referencias digan eso, no es válido [22:07:49] es que partimos de la base de que se asume que la referencia indica un dato, cuando ni siquiera es cierto [22:08:33] Según el traductor de Google dice «marino italiano» [22:08:49] ¿? [22:08:52] Aunque probablemente la Italia a la que se refiere sea la histórica y no la actual [22:09:11] la referencia indica como nacionalidad el codigo "e-it" [22:09:30] pero luego ni siquiera tiene declarado la existencia de dicho codigo [22:09:43] Ah, no me había fijado en el código; yo lo decía por la descripción [22:09:53] "Italiensk sjöfarare" [22:09:59] se supone que "it" es "italiano", pero "e-it"... [22:10:13] Ni idea :/ [22:10:22] Quizá precisamente se refiera al Estado histórico [22:10:43] posiblemente, pero en wikidata se esta atribuyendo al Q de la actual italia [22:11:02] mi cambio en la entrada se correspondia a dejar solamente el Q de la italia de la epoca [22:11:19] Vamos a dejar más tiempo para que Andrea responda, que por nosotros no sea, ¿qué te parece? [22:11:20] que es la republica de genova (concretamente en esa region) [22:11:30] bien [22:11:44] Y, si al cabo de un tiempo no dice nada, miramos cómo ponerlo [22:12:05] pero [22:12:13] entonce no toda referencia vale [22:12:16] entonces* [22:13:04] Bueno... digamos que, conforme el ser humano va avanzando científica y tecnológicamente, las referencias nuevas pueden indicar que las antiguas no sirven [22:13:28] Podríamos sostener que la Tierra es plana con referencias antiguas... pero esa declaración habría que definirla con rango obsoleto [22:13:42] Porque hay referencias modernas, más fiables, que lo desmintieron [22:13:47] cierto [22:14:17] La restricción de coetaneidad solo puede violarse con rangos normal y preferido [22:14:20] pero siguiendo con tu ejemplo, si yo declaro que la tierra es piramidal, y lo referencio con un "google.com"... no sirve, no ? [22:15:16] Hombre... no sé, si es una referencia muy loca, no, :) pero eso como en Wikipedia [22:15:32] Si hubiera alguna publicación científica que lo dijera, se podría sostener junto con la otra declaración [22:15:36] Pero no creo que la haya xD [22:32:32] * Platonides piensa que seguro que hay referencias recientes del flat earth movement [22:32:57] Pero las referencias contrarias al movimiento son también recientes :P