[00:54:05] AsimovBot: [00:54:12] -? [00:54:12] Asimov v. 2, By jem (IRC) / -jem- (Wikimedia), 2010-16 - Bot IRC de apoyo a los proyectos y al movimiento Wikimedia programado en PHP - Las órdenes deben escribirse precedidas de alguno de los prefijos admitidos (-) - 13Lista de órdenes: -ord - 13Problemas o sugerencias: -sug - 13Ayuda: -? 15,02orden / -ic - 10http://wikimedia.es/asimov?uselang=en [08:35:49] Adrian_WMDE: https://github.com/wmde/WikibaseDataModel/pull/639 [08:46:33] Does every article on wikipedia have a corresponding wikidata article? [08:47:19] No. [08:47:21] i.e, if you just wanted basic info like a list of names of all medicines on wikipedia, do you need to scrape wikipedia, or can you just look at wikidata? [08:47:54] They should have one, but there are many cases that a item wasn't created yet. [08:48:41] what are the cases when an article might not be made sjoerddebruin ? is wikidata scraped automtically from wikipedia, or is it all done manually? [08:49:14] Some projects create items automatically. [08:49:28] See https://tools.wmflabs.org/wikidata-todo/duplicity.php? for some stats [08:49:45] But it seems like the tool is down again :( [08:49:52] yeah. [08:50:26] sjoerddebruin, basically i'm trying to create a list of all wikipedia articles related to medicine, disease, health, etc [08:50:39] just wondering if this can be done by looking thru wikidata alone [08:50:53] or if i should build a classifier for wikipedia articles [09:12:42] Thiemo_WMDE: are you preparing the DM release? [09:12:58] or shall I just do it? [09:23:25] hoo: https://integration.wikimedia.org/ci/job/mwext-testextension-hhvm-composer/1499/console [09:23:39] http://www.commitstrip.com/en/2016/01/18/what-idiot-wrote-this-code/ [09:33:05] Adrian_WMDE: https://gerrit.wikimedia.org/r/272687 [09:40:11] https://www.wikidata.org/wiki/Q6703647 [10:16:59] Adrian_WMDE: did we already think about extracting WikibaseView from Wikibase.git and giving it its own repo? [10:17:09] That would allow us to have stable releases for that component [10:17:11] Sure, we already have gerrit project [10:17:19] so what is it blocked on? [10:17:32] Hasn't been prioritized yet [10:17:47] but would be ready to go? [10:17:48] https://phabricator.wikimedia.org/T75020 [10:18:04] I think so [10:18:21] At least a year ago PHP dependencies were quite clean, JS was acceptable [10:18:45] Would create a circular dependency between wikibase.git and wikibaseview.git, though [10:19:19] what does WikibaseView use from Wikibase? [10:19:39] maybe we can do some abstraction there or have dependencies in Wikibase DM Services [10:20:03] lib [10:20:17] I can have another look at it [10:21:04] https://phabricator.wikimedia.org/T75020#850415 documents what was true at one point in time [10:21:07] oh, mainly the formatter [10:21:16] which should be extracted from lib anyways [10:21:51] I'd be willing to put another sprint into this [10:22:41] we can discuss that next week when I'm at the office :O [10:22:42] * :) [10:22:47] Yeah [10:23:04] and maybe pick it into the sprint [10:49:02] Right. https://www.wikidata.org/w/index.php?title=Q4141192&type=revision&diff=306507676&oldid=305135071 [10:55:15] DanielK_WMDE: hi, got a question [10:55:22] where should the factory method for an EntityView go? [10:55:47] a callback in WikibaseRepo.entittypes.php, inside of EntityContent or in EntityHandler? [11:06:09] benestar: EntityHandler produces HTML for a given entity. That's what uses the EntityView. I think we currently inject an EntityViewFactory - if we want that, the factory should be based on WikibaseRepo.entittypes.php. It's a bit tricky - the EntityHandler can and should know the concrete EntityView it needs, but injecting all the services that would be needed to have EntityContent instantiate the EntityView would be annoying... [11:08:17] so, logically, EntityContent/EntityHandler knows the concrete EntityView class and can instantiate it. But doing it that way may be impractical. You could inject a factory callback into EntityHandler, without wrapping it in a EntityViewFactory... but since PHP doesn't have a decent type system for callables, that's not very nice. [11:10:00] DanielK_WMDE: can't find the place in EnityHandler where the HTMl gets generated [11:11:00] benestar: It's in EntityContent (sadly. that was a mistake I made when designing the interface). [11:11:05] ah, it's in EntityContent [11:11:07] EntityContent::getParserOutput [11:11:35] a lot of global state is involved there :S [11:11:36] Ideally, that would just delegate a simmilar function in EntityHandler [11:12:32] EntityContent shouldn't know services really, it should be a value object. I screwed that up when I introduced the Content interface. But we can at least have EntityContent just rely on EntityHandler [11:13:11] benestar: when the code was written, it wasn't possibel to inject anything into ContentHandler. Now you can. [11:14:35] DanielK_WMDE: what about just passing the EntityView into the EntityParserOutputGenerator and creating it separately? [11:15:30] we could inject a ViewFactory into EntityHandler and then let subclasses create the view [11:15:43] which can then be injected into the ParserOutputGenerator [11:15:48] Entity* [11:16:24] yes, that's probably the most streight forward approach. [11:16:58] Food? https://www.lieferheld.de/lieferservices-berlin/restaurant-thai-huong-snack/98/ @hoo @Thiemo_WMDE @DanielK_WMDE @aude @Adrian_WMDE [11:17:33] Jonas_WMDE: not for me thanks. I have to run out for a bit anyway, i'll grab something on the way [11:18:24] benestar: actually, don't inject a ViewFactory,. Inject a ParserOutputGeneratorFactory. [11:18:35] roter curry mit tofu + vegetarisch fruhlingsrulle [11:18:39] Jonas_WMDE: ^ [11:18:43] that as well [11:18:51] benestar: why both? [11:19:04] because we have to do dispatching on the EntityView level [11:19:16] and I don't want to pass callbacks into the PaserOutputgenerator [11:19:45] didn't think that through yet [11:20:40] benestar: we don't need any dispatching. the dispatching is already done when getParserOutput is called. it's already called on a specific type of entity. [11:21:09] we *can* do dispatching in the factory. in the ParserOutputGeneratorFactory, probably. [11:21:10] but where do you get the EntityView from then? [11:21:18] but we don't have to, we can have one factory per entity type [11:21:36] benestar: ParserOutputGeneratorFactory hjas a EntityViewFactory injected [11:21:52] yes, but that currently contains information about entity types [11:22:02] and I'm about to refactor it so it doesn't containt that [11:22:10] and only provides newItemView and newPropertyView [11:22:20] that seems backwards [11:22:26] huh? [11:22:28] it should just have getEntityView( $type ) [11:22:45] nope, how should it know about MediaInfoView? [11:22:57] or Wiktionary...View [11:23:00] Jonas_WMDE: Süß-Sauer Tofu [11:23:16] In my mind, EntityViewFactory is a wrapper around a map of factory callbacks [11:23:33] DanielK_WMDE: sure about that? [11:23:53] I'm just trying to get away from the factory callbacks for views [11:24:04] because we *can* do it in EntityHandler [11:24:33] but maybe it's easiest to have separate factory callbacks for the views [11:27:19] benestar: we can go that way, but you will need to inject a lot of services into EntityHandler then. In that case, EntityViewFaqctory and ParserOutputGeneratorFactory can just go away. [11:27:19] We don't need them at all [11:27:19] EntityHandler would know how to construct the right EntityView and POGen [11:35:17] @aude @Thiemo_WMDE food? [11:39:01] Jonas_WMDE: Gelber Curry - Tofu - 5,70. [11:39:09] looks like there's only 58k-ish articles on wikipedia without a wikidata item [11:39:12] roter curry mit tofu + vegetarisch fruhlingsrulle [11:39:18] that's only like 1%-ish right? [11:39:45] voidDotClass: actually that should be way below 0.1% [11:39:54] nice [11:40:32] this is really cool, i wonder why wikidata isn't more popular in machine learning circles [11:41:23] Adrian_WMDE: no scrollbar please [11:42:55] Jonas_WMDE: roter curry mit tofu + vegetarisch fruhlingsrulle [15:01:57] DanielK_WMDE: reminder: you wanted to improve the description of https://phabricator.wikimedia.org/T126445 (e.g. adding the fact we want an extension entry point..) [15:46:39] jzerebecki: Lydia_WMDE https://wikitech.wikimedia.org/wiki/Deployments#deploycal-item-20160223T1900 [15:47:18] * aude notes that there is no train later [15:50:53] Tobi_WMDE_SW_NA: done, thanks for the reminder [16:11:18] aude: I think you also need to clear the parser cache when adding a new site group, but not sure offhand [16:11:29] has been a while since we last did that [16:37:36] hoo: we do :/ [17:15:09] aude: so for the messages for wikiversity a full scap needs to happen for wmf.13 and wmf.14, right? so can you do the Wikidata update of wmf.14 at the same time? [17:24:54] jzerebecki: so we want https://gerrit.wikimedia.org/r/#/c/272531/ ? [17:25:41] can probably do [17:29:42] Is Lydia still in the office? [17:30:40] * aude not [17:41:45] hoo: no. will be online later [17:42:11] aude: yes that is the patch [17:42:39] Ok... I'll have to wait then [18:31:05] The big purge is done hoo, did anything break? [18:32:03] * hoo looks around... [18:32:10] Looks all fine to me [18:32:21] Are the tables now actually fully populated? [18:32:28] or rather the table [18:41:47] Sure looks like it :-) [18:44:33] Awesome [18:45:44] hoo: when are you at the office again? [18:46:33] aude: Maybe March 10/ 11 [18:46:43] and/ or after the Hackathon [18:46:47] and then after the hackathon [18:46:48] ? [18:47:24] Yeah, after the Hackathon... total free from University for almost a week after that [18:47:39] :) [18:48:12] sjoerddebruin: You might be interested in https://tools.wmflabs.org/multichill/queries/wikidata/new_items_with_sitelink_without_claim_nlwiki.txt [18:48:26] :) [18:49:57] If I just take new items I only get 75 hits [18:50:42] Yeah, trying my best. [18:55:08] I was thinking about doing some automatic classification sjoerddebruin . Take for example https://www.wikidata.org/wiki/Q22572484 . A machine could figure out it's a church [18:55:28] Yeah, or the creator should fix his shit. [18:55:53] (like I already asked) [19:07:17] multichill: the stats were from https://www.wikidata.org/wiki/User:ArthurPSmith/Identifiers/0 (also pages /1 and /2) [20:28:57] nikki: Oh, that's quite useful! It's a shame how messy some properties got [20:29:16] Take for example https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations/P709 [20:29:46] take a look at the mandatory violations... [20:29:47] * nikki nods [20:30:33] Just random example https://www.wikidata.org/wiki/Q6820745 is about the house, but both http://portal.historic-scotland.gov.uk/designation/LB15110 and http://portal.historic-scotland.gov.uk/hes/web/f?p=PORTAL:DESIGNATION:::::DES:LB15112 link to it [20:34:18] sjoerddebruin: https://tools.wmflabs.org/multichill/queries/wikidata/new_items_with_sitelink_without_claim_nlwiki.txt ziet er nu wat beter uit [20:34:50] Ik wou dat ik dat kon zeggen over de ongekoppelde pagina's. [21:41:06] If somebody has time could they look at https://www.wikidata.org/wiki/Wikidata:Project_chat#ESRB_ratings I'm basically wondering if I've done it right before I start writing a bot to import [21:42:54] Like use Retrieve URL or Official Website? [21:49:32] which P-number is Retrieve URL? [21:51:09] Sorry reference URL (P854) and official website (P856) [21:51:45] P854 is meant to be used for URLs for sources [21:52:03] where P856 is the official website for a wikidata-item. [21:52:17] 854 you use to source a claim ... [21:52:25] No distinction between Primary and Secondary sources? [21:52:34] not that I'm aware of [21:53:03] if the source is a certain language wikipedia, there is a different P-xxx for it [23:11:40] PROBLEM - wikidata.org dispatch lag is higher than 300s on wikidata is CRITICAL: HTTP CRITICAL: HTTP/1.1 200 OK - pattern not found - 1559 bytes in 0.231 second response time [23:11:47] looking [23:12:07] oh, obvious [23:12:16] aude: ^ the new wikis making trouble [23:12:24] Shall I fix or will you? [23:12:46] M�s antiguo frwikiversity 304254 586596 5252 d�as y 37 minutos [23:12:56] hoo: if you want [23:12:59] Yeah [23:13:08] Learning Spanish on Wikidat anow? [23:13:11] * aude is stuck with things in spanish :) [23:13:15] or has spanish [23:15:10] Ok, recovering [23:15:34] good [23:16:04] * aude forgot how long it takes to enable wikibase on new wikis [23:16:21] mostly my fault to forget wikimedia messages until today [23:16:56] Take a harmless scap... just a time eater [23:17:21] RECOVERY - wikidata.org dispatch lag is higher than 300s on wikidata is OK: HTTP OK: HTTP/1.1 200 OK - 1538 bytes in 0.248 second response time [23:17:27] yeah, slooooooooooooow [23:17:39] and will take some time for translations to get in [23:17:49] Not that long [23:17:59] we still do nightly l10n updates, don't we? [23:18:02] yeah [23:21:10] first links incoming: https://www.wikidata.org/w/index.php?title=Q3938&action=history [23:21:50] \o/ [23:22:19] * aude resists starting with adding stuff to the main page item [23:22:24] just in case something was broken [23:22:38] probably ok now [23:22:43] Better wait for an hour or so just to make sure all kinds of weird caches cleared up [23:23:12] yeah :) [23:23:16] https://en.wikiversity.org/wiki/Wikiversity:Sandbox looks good, though [23:23:26] yep [23:24:00] Do they have their own other projects sidebar? [23:25:25] seems so [23:25:53] mh... on both en and fr... but not de [23:26:06] not on spanish [23:26:26] ugh [23:26:42] i see duplicate interwiki links [23:26:59] ugh [23:30:35] site_identifiers table seems broken [23:30:54] yikes [23:30:59] did you only just populate it? [23:32:14] it was already populated [23:32:29] i can rerun the script for wikiversity (already fixed eswikiversity) [23:32:30] I once ran the populate script for all wikis [23:32:42] but didn't re-run it ever after [23:32:52] (only for the client wikis) [23:32:58] that's better [23:33:08] the whole table is such a mess -.- [23:34:00] :( [23:34:12] all should be good now