[08:47:59] DanielK_WMDE: ping? [09:34:46] hoo still not in the office [09:34:55] :/ Ok, thanks [09:59:26] a couple of weeks ago or so, something was added so that we can do sparql queries for the number of sitelinks... but now I can't find any mention of it anywhere, does anyone remember what it was? [10:00:11] nikki: https://www.mediawiki.org/wiki/Wikibase/Indexing/RDF_Dump_Format#Page_properties [10:00:20] thanks [10:00:27] I was starting to think I'd dreamt the whole thing :P [10:01:00] I only heard about it at yesterday’s office hour, somehow missed the original announcement :D [10:12:33] Jonas_WMDE: Is Thiemo in? [10:15:30] hoo thiemo is on holiday, daniel will be there in 30 min anything I could help you with? [10:16:20] Ah ok… no, thanks :) [10:51:44] hoo: did we ever get https://gerrit.wikimedia.org/r/#/c/309477/ into master ? (fix/revets lua changes) [10:51:49] * aude sees your new patch [10:52:10] :) [10:52:54] i would prefer the revert in master (so code is deployable) and then the new change [10:54:07] it's a "kittenen" https://en.wikipedia.beta.wmflabs.org/wiki/Kitten [10:54:11] hm… I don't see the point now… we have an actual patch [10:54:27] as long as we merge it quickly, then ok [10:54:48] but would like to deploy next week [10:54:54] Yeah, same here [10:54:59] but I guess we can get it merged till then [10:55:01] at least I hope [10:55:17] I have quite some other stuff I want to do regarding to Lua/ the parser function [10:55:22] ok [10:55:35] hope that I wont end in rebase hell :/ [10:56:27] * aude wonders why there was no wikidata build today? [10:57:13] We don't have any retries in the script [10:57:22] a single failure at any point could have made it fail [10:57:59] it's usually something actually broken [10:58:14] like composer has a conflict [10:58:31] hm [10:58:36] * hoo hits the process of manually [10:58:57] looks fine: https://gerrit.wikimedia.org/r/313385 [10:59:29] oh [11:00:04] :) [11:00:11] [Composer\Downloader\TransportException] ^[[39;49m^[[31m [11:00:12] >> ^[[39m^[[37;41m The "http://packagist.org/p/wikibase/data-model-services%2412d65ec19c0d6467ccb8c9aaa10009e3c5695dbf4e91db6719eb19b7fba69cf8.json" file could not be downloaded (HTTP/1.1 302 Found) [11:00:15] From the logs [11:00:23] so it was just packagist being flacky [11:00:27] ok [11:01:56] aude: https://www.mediawiki.org/wiki/Extension:Wikibase_Client/Lua#mw.wikibase.entity:getLabel Do you think that should also respect the user language if the wiki is multilingual [11:02:04] like mw.wikibase.label/ .description does [11:02:08] seems consistent to me [11:02:20] maybe [11:02:34] hm [11:02:39] consistency is good [11:03:16] Indeed, but this is also a (slightly) breaking change [11:03:23] but doubt that's actually relevant [11:03:39] or a problem [11:03:49] yeah [11:07:59] relevant ticket https://phabricator.wikimedia.org/T143277 [11:08:31] ok [13:44:40] jakob_WMDE: could you file tickets for creating "dispatching" services we discussed today? [13:44:52] i'm a bit worried that we are not really tracking our activity [13:45:32] DanielK_WMDE: sure! [13:57:07] jakob_WMDE: thanks [13:57:25] jakob_WMDE: please add them to the TLA board [13:58:07] DanielK_WMDE: will do. I'll create one ticket for each EntityLookup, EntityRevisionLookup and EntityIdComposer [13:58:12] leszek_wmde: apropos tickets... i was just thinking, we should probably have a ticket for the writing foreign id spec, too. just for tracking [13:58:37] (the ladder one not being a dispatching service, but I started working on that today) [13:59:44] jakob_WMDE: oh, please also EntityStore, to prevent storing foreign [14:00:05] ok! [14:00:12] regarding EntityIdComposer... i was just thinking... it's not dispatching, but we might need it per repo. [14:00:47] EntityIdComposer is used when reading from a db table that only has numeric IDs. if we read such IDs from another repo's DB, the EntityIdComposer needs to add the prefix for that repo. [14:01:28] jakob_WMDE: is that what you have been working on?.. [14:02:34] DanielK_WMDE: no, i was working on disallowing using EntityIdComposer for foreign ids but what you're saying makes sense [14:04:38] I think the code for EntityIdComposer might not even need to be changed to work with foreign entity ids since it uses callbacks anyway [14:06:14] jakob_WMDE: well, it also never gets an EntityId as input, so there's nothing to disallow [14:07:31] jakob_WMDE: EntityIdComposer would just take the repo prefix as a constructor param, with '' as the default [14:08:12] jakob_WMDE: the callbacks would get it as the second param, i guess... newFromNumber may also need it as an extra param? [14:08:33] ...which needs another release of the DataModel component. [14:08:40] this is becoming complicated already :P [14:09:32] ugh... [14:09:56] or drop newFromNumber, and rely on callbacks only [14:10:54] and then have composers in the constructor for each foreign repo? i guess that works but might get ugly [14:11:47] DanielK_WMDE: i think for now i'll remove the red color from EntityIdComposer and add a comment to our document of services ;) [14:12:59] what i started to work on actually was to prevent it from returning foreign ids but that felt super silly so I was already wondering whether that even makes sense [14:14:49] yea... not really :) [14:15:17] and yes, a different composer would get injected for each repo. We are already injecting it, so this should just fall into place [15:17:24] https://www.wikidata.org/wiki/Property:P3222 :D [18:53:47] aude: here? [19:07:55] looks like aude's not around :) [19:10:31] I figured on my own… and damn, SQLite is weird [19:38:08] DanielK_WMDE: ping [21:24:08] hoo: hey. sorry, crazy day [21:24:11] what's up? [21:53:08] DanielK_WMDE: hey [21:53:20] Various things piled up [21:53:41] First of all https://phabricator.wikimedia.org/T143277 [21:53:51] Any obejctions? I want to do it [21:53:56] and aud.e is ok as well [21:54:10] Will need some announcement, but the impact should be manageable [22:01:01] hey how do i find a list of enwiki articles w/o wikidata pages? [22:05:15] Zppix: have a look at https://en.wikipedia.org/wiki/Special:UnconnectedPages [22:16:22] hoo: sounds good to me. does it split the parser cache, though? [22:16:45] we only have allowDataAccessInUserLanguage enabled on commons, right? [22:17:51] hoo: tbh, i'm too tired for anything complex. send me an email, and i'll look into stuff later. i'm on vacation starting tomorrow, but i'll be doing stuff in the evening. [22:25:42] DanielK_WMDE: the parser cache is already split by then [22:26:15] I'll compose an email either today or tomorrow with the things I want comments/ opinions/ CR on [22:27:31] ok, thanks