[16:10:11] TimStarling: meeting in 3hr, right? [16:10:19] Updating the RFC [16:37:06] hexmode: isnt it a little later? [16:37:06] https://phabricator.wikimedia.org/T76575#806428 [16:37:48] 21:00 UTC (now it is ~ 16:30 UTC ) [16:38:21] I thought so as well Helder [16:41:51] Helder: ok.... well, better early than late. [18:00:10] :D [18:00:15] hey everyone :) [18:00:22] Hey [18:00:24] time for the wikidata office hour [18:00:26] Hey [18:00:27] hey [18:00:28] who's here for it? [18:00:34] <-- [18:00:34] yay! people! :D [18:00:51] We need to do the formalities first... [18:01:02] oh my :P [18:01:03] which ones? [18:01:20] Wasn't there a bot here or something? [18:01:36] yes but i fail at using it... [18:01:43] does anyone know how to use it? [18:02:00] also it seems it is gone [18:02:06] nvm then [18:02:12] :) [18:02:17] alright [18:02:27] so big stuff that happened lately: [18:02:29] I think we need to settle the current problem, deletion vs redirection. Some admins redirect and some delete. I think rfd should be consistent. Either redirect or delete [18:02:41] Ehm, deploy today? [18:02:56] heh ok then those first [18:03:06] Jianhui67|cloud: there is an open rfc that needs closing by an admin about this [18:03:11] not sure why it wasn't closed yet [18:03:22] sjoerddebruin: yes deploy today for statements on properties [18:03:26] Yep. It has been opened for 2 months [18:03:29] Maybe we don't have active uninvolved admins? [18:03:43] possibly [18:03:47] i didn't check [18:03:48] Oh, my name is no where on that page. [18:03:52] hah! [18:04:10] But never closed a RfC before. :/ [18:04:49] ohhhhh [18:04:51] but but but [18:04:54] then it can be your first [18:04:57] exciting! :D [18:05:17] Jianhui67|cloud: Have you asked people like Epìdosis why they are still deleting? [18:05:33] And Cycn. [18:05:53] I think its because they simply dont want to loose the rights (due to inactivity) [18:06:08] Bad reason... [18:06:15] Lol [18:06:27] yeah ... btw hi Lydia [18:06:30] Cycn already showed his disagreements [18:06:46] hey matej_suchanek :) [18:07:00] Oh yes, I can't find Cycn because of the stupid signature. [18:07:02] * DanielK_WMDE wibbles [18:07:02] Jianhui67|cloud: what did they say? [18:07:06] I told some of the people who request deletion to redirect instead of requesting deletion [18:07:45] cool [18:08:09] sjoerddebruin: do you feel like you can/want to close the rfc or should we see if we find someone else? [18:08:37] I don't think I'm neutral anymore. :/ [18:09:15] fair enough [18:09:36] Jianhui67|cloud: want to find another admin then? or post it on the admin noticeboard? [18:09:47] good idea [18:09:48] Stryn, why don't you do it [18:09:57] Or matej_suchanek [18:10:20] There is also a backlog regarding property deletions. [18:10:22] I was involed into it and I have never closed an RfC before [18:10:34] I think it's a good point to start 2015 clean. [18:11:53] sjoerddebruin: is the issue a lack of discussion about the properties or about making the decision and doing it? [18:12:33] There is enough discussion, but many people participated. [18:12:55] i see [18:13:39] And I'm very busy with all backlogs... [18:13:44] *nod* [18:13:50] thanks for working on those! [18:14:10] I sometimes think I'm the only one that is beating vandalism. [18:14:17] oh I have just reminded... [18:14:24] I'm lazy :p [18:14:36] Lazy to do anything lol [18:14:37] mh the merge gadged has "Request deletion for extra item on RfD" as a default, probably should be changed when that RfC is closed [18:14:45] Jianhui67|cloud: how about becoming unlazy and posting about the rfc on admin noticeboard? :D [18:14:54] Sure [18:14:57] \o/ [18:15:12] jzerebecki: adaption already in progress i think according to project chat [18:15:19] the filter 32 (Special:AbuseFilter/32) doesn't work anymore, same issue as with the redirects [18:15:35] matej_suchanek: which one is that? [18:15:49] tagging merging items [18:15:53] Uncompleted merges: https://www.wikidata.org/wiki/Special:ShortPages [18:16:14] it's been so since September 30th [18:16:36] matej_suchanek: is this something the dev team can/has to fix? what's the issue? [18:17:14] I don't know but there was a bug that the AF couldn't tag creating redirects [18:17:24] ahhhh [18:17:26] even if there was everything correct [18:17:26] i remember [18:17:32] let me search for that [18:17:42] and I see this is a similar issue [18:18:19] https://phabricator.wikimedia.org/T72715 [18:18:25] ok i'll bump this one [18:18:36] thanks [18:18:38] np [18:19:37] other things we should talk about? [18:19:52] maybe I have one more question [18:19:57] shoot [18:20:15] Done! [18:20:39] Jianhui67|cloud: \o/ [18:21:15] it's quite complex [18:21:33] it's related to enforcing constraints [18:21:37] we can handle it :P [18:21:39] ok [18:23:21] I also realise from RfD, there seems to be a problem with creating redirects from there. [18:23:23] basically you have once on the Project Chat stated that if there is an inverse constraint the system wouldn't handle it by making two same edits to both items [18:23:45] yeah [18:23:54] this was something I had imagined how it should work [18:24:29] for example both statements would have the same reference(s), qualifiers etc. [18:25:06] what was the reason not to go this way? [18:25:09] right [18:25:13] just asking [18:25:19] sure :) [18:25:29] this is what this office hour is for, right? ;-) [18:25:36] yeah [18:25:44] anyway: the reason is that the world is unfortunately complicated [18:26:10] and while in most cases inverse properties might actually be inverse there are still those special cases that are "different" [18:26:28] just like we don't specify that sex/gender can only be male and female [18:26:36] because obviously the world is more complex than that [18:26:39] however! [18:26:51] we can still build clever tools to help [18:27:18] ok .. by the way, how's the team doing? [18:27:23] for the sex/gender case we are going to build out the suggester so it suggests the most common values first while still letting you enter other things [18:27:38] may I add to the answer to the inverse question a bit? [18:27:58] yes [18:27:58] and for inverse properties we can simply give the user a small hint saying "this seems to be an inverse property. would you like to also add it to the other item?" or something like that [18:28:10] dennyvrandecic: please :) [18:28:20] imagine the inverse of "country" [18:28:31] or of "nationality", to be a bit more specific [18:29:09] if every statement with a country property added to any song, movie, building, etc. would lead to a respective edit on the object of that statement, ... [18:29:12] ... we would be in trouble [18:29:29] hah that too! [18:29:37] the items for Germany, Russia, United States are already quite rich [18:29:47] but that's nothing compared to the number of inverses they would sport [18:30:16] (Hello All) [18:30:26] matej_suchanek: does that make sense? [18:30:29] I like the idea of the small hint [18:30:40] me too [18:31:04] sweet [18:31:20] I think there should be, once we have simple queries up and running, a link to "show me all items that have a statement with this as an object and the property P" [18:31:22] there are quite a few cases where we can be smarter like this [18:31:35] that would resolve your usecase in most cases [18:31:35] if you can think of some that would be useful do file tickets or tell me [18:31:59] not in all use cases, but in most - e.g. reflexive properties are not handled well, still [18:32:04] but these are fortunately rare [18:32:08] dennyvrandecic: the explanation makes sence for me but the mentioned nationality is not an inverse property so there can't be a problem ... or did I misunderstand you? [18:32:19] Lydia_WMDE: When will I be able to view Wikidata on mobile? [18:32:33] sjoerddebruin: ahhh asking the painful questions :D [18:32:34] no, but the property "nationality" would have an inverse, like "citizen" or sth [18:32:46] It's almost 2015. :) [18:32:47] matej_suchanek: ^ [18:33:00] sjoerddebruin: i can't give you a specific date but we'll make progress with the new UI on that [18:33:01] sjoerddebruin: as soon as there are devs to do it :P [18:33:39] why not start an Open Source alternative Wikidata app? I wanted to do some Android deving anyway :) [18:33:54] having an app would be cool [18:34:16] I'm also dreaming of a working nearby page on Wikidata. ;) [18:34:25] matej_suchanek: another issue is that a reasonable inverse might sometimes depend on the actual domain and range [18:34:28] sjoerddebruin: there is a tool for that [18:34:37] sjoerddebruin: let me see if i can find it [18:34:56] matej_suchanek: the property country is currently used on movies, songs, albums, buildings, mountains, etc. it is hard to figure out a good inverse for that [18:35:20] sjoerddebruin: http://www.johl.io/aroundme-wikidata/ [18:35:22] matej_suchanek: but it would be a pity to create incentives to create only properties that have a good inverse [18:35:34] I meant that not all properties need an inverse [18:35:36] Why is it German? :( [18:35:56] It's broken. [18:35:59] sjoerddebruin: you can poke johl on twitter to translate it ;-) [18:36:04] matej_suchanek: true, but there would be task forces to make inverses for all properties ;) [18:36:12] it is working for me [18:36:24] the tool is nice anyway [18:36:29] Will try the iOS-simulator then. [18:36:33] k [18:36:50] Nothing... [18:36:58] hmmmm [18:37:11] are you giving it permission to read your location? i missed that first [18:37:18] Yup. [18:37:27] dennyvrandecic: i'm totally for getting rid of all inverses ;) its redundant data [18:37:28] WONTFIX, works for me ;) [18:37:50] sjoerddebruin: hmm then i don't know sorry [18:38:24] But we have https://tools.wmflabs.org/wikidata-todo/around.html :D [18:38:47] ah that also works ;-) [18:39:02] The Netherlands is crowdy because of all the monuments. :) [18:39:31] nice! :D [18:39:46] I once had the idea to put locations of Subway restaurants, Starbucks, etc. on Wikidata... [18:40:29] So we can become a good POI-provider. [18:41:32] that might be a bit too much and better left to openstreetmap? :) [18:42:15] we are still missing a property to link to stuff on osm that is not a relation there [18:42:15] Me and my crazy ideas... [18:42:28] sjoerddebruin: <3 [18:43:04] Lydia_WMDE: otoh wikivoyage probably has many of those POIs already... [18:43:14] sjoerddebruin: but that would need an item for each starbucks? [18:43:27] dennyvrandecic: Probally. :P [18:43:39] I don't think that's a wacky idea [18:44:20] We also need a business hours property then. :/ [18:44:27] I actually think it would be pretty awesome. Although I would start not with Starbuckses, but "all schools", "all museums", etc. [18:44:50] we have a not too bad coverage for these in some places already [18:44:59] yeah [18:45:17] all libraries [18:45:58] that would be rather neat [18:46:04] long time ago, I got a crazy idea [18:46:14] matej_suchanek: I'd love to hear it [18:46:40] as a user, you can have a sandbox but that's only for wikitext [18:46:56] what about having a sandbox item in the user namespace? [18:47:08] I have a file with a lot of schools in the Netherlands, sadly no license is noted.https://data.overheid.nl/data/dataset/02-vo-adressen-alle-vestigingen--ministerie-van-ocw [18:47:11] there is a sanbox item already in the main namespace [18:47:33] I know [18:47:58] data licenses suck :( [18:48:10] as I said, it was a crazy idea [18:48:13] Sailing schools are public domain. https://data.overheid.nl/data/dataset/zeilscholen XD [18:48:31] matej_suchanek: not crazy ;-) [18:48:45] sorry, leaving for a while [18:49:01] The best public domain dataset I've found is about gas stations. With coords and all. [18:49:16] gas stations? which area? [18:49:23] Netherlands. [18:49:59] (10 mins left) [18:50:34] any other topics we should discuss still? [18:51:38] ok then i'll give a quick list of things the dev team has been working on lately: [18:51:57] * performance improvements - some are already live but more are coming [18:52:10] * UI redesign - currently the sitelink section is still being worked on [18:52:22] * statements on properties - to be rolled out tonight [18:52:36] * language fallbacks - first version to be rolled out in a week or two [18:52:55] * arbitrary access and usage tracking - coming early next year - rollout will be staged [18:53:57] * started working with the team of students to work on data quality tools - they are concentrating on constraint violations and reports as well as checking against 3rd party databases [18:54:04] back again [18:54:05] up next: [18:54:13] * finishing the above [18:54:18] rough idea on the stages for the arbacces? [18:54:19] * working with the Foundation on queries [18:54:41] * more work on data quality tools [18:54:47] * units [18:54:55] Units <3 [18:55:04] dennyvrandecic: commons, small wikipedias, big wikipedias, other projects maybe? [18:55:21] sweet [18:55:41] what about wikidata, commons, ... ? [18:55:49] ? [18:55:55] wikidata should be the most friendly place to start this arbaccess [18:56:02] and they won't need it that much [18:56:10] oh it already has it! (just without purging) [18:56:28] we endabled that a while ago because there were bad hacks to make it work anyway [18:56:37] ah, ok, so it will be the first place to get tracking and purging anyway? [18:56:39] good idea [18:56:42] yeah [18:56:56] so it can be tested there, and then it goes to commons, who will really benefit from this [18:57:00] and the smaller wikipedias [18:57:01] right [18:57:05] awesome [18:57:17] and agree with sjoerddebruin: units <3 [18:57:20] :D [18:57:27] all <3 units [18:57:29] Stryn closed the rfc [18:57:33] \o/ [18:57:34] so the arbacc deployment will be split? [18:57:45] wow!!! [18:57:46] matej_suchanek: yeah because we need to see how it scales [18:57:48] sjoerddebruin:, dennyvrandecic: Might I add please add / contribute at some point to this conversation some CC wiki link resources planned for first for large languages, focusing on Sailing schools, All Museums, All Libraries as part of a CC all 7, 106 languages, MIT OCW-centric open wiki World University and School, which is like Wikipedia with MIT OCW ? [18:58:08] Lydia_WMDE: right [18:58:24] Scott_WUaS: yes, sure, go ahead, it's an open forum [18:58:28] every big change goes first to small projects [18:58:53] maybe even small wikipedia before commons [18:59:00] *nod* [18:59:10] because commons can (and should!) go crayz with arbacc! [18:59:15] *crazy [18:59:16] hehe true [18:59:19] Stryn, many thanks to you [18:59:43] Thanks ... still only in English, but planned for large languages and for wiki teaching and learning - here's the beginning wiki Sailing subject / school - http://worlduniversity.wikia.com/wiki/Sailing [18:59:44] it wasn't so hard than I expected [18:59:54] thanks Stryn! [18:59:56] lol [18:59:59] it's time :D [19:00:13] yeah it was very short :/ [19:00:14] It's 3am over here xD [19:00:19] Library Resources planned to aggregate all open online libraries in each of all languages - http://worlduniversity.wikia.com/wiki/Library_Resources [19:00:38] thanks everyone for coming! :) as always if you want to chat more come to #wikidata [19:00:55] was good to chat with you all [19:00:57] <3 [19:01:01] Museums wiki subject page - planned to aggregate all open online museums in each of all languages http://worlduniversity.wikia.com/wiki/Museums [19:01:07] Thanks, Lydia! [19:01:17] thanks Scott_WUaS! [19:01:42] see you on #wikidata [19:01:47] thanks Lydia_WMDE [19:02:06] always [19:02:11] Thanks member:dennyvrandecic and member:sjoerddebruin [19:02:38] Now I feel like in a socialist cadre, being called a member :D [19:02:53] haha [19:03:12] I removed the colon and these technologies printed member :) [20:56:01] 5 minutes [20:56:26] till the Global Wiki discussion? [20:56:59] There's going to be a MediaWiki RfC meeting [20:57:17] ah, thanks [20:57:20] https://lists.wikimedia.org/pipermail/wikitech-l/2014-December/079713.html [20:58:54] #startmeeting RFC meeting [20:58:57] Meeting started Wed Dec 3 20:58:54 2014 UTC and is due to finish in 60 minutes. The chair is TimStarling. Information about MeetBot at http://wiki.debian.org/MeetBot. [20:58:57] Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. [20:58:57] The meeting name has been set to 'rfc_meeting' [20:59:13] #topic Shadow namespaces | RFC meeting | PLEASE SCHEDULE YOUR OFFICE HOURS: https://meta.wikimedia.org/wiki/IRC_office_hours | Please note: Channel is logged and publicly posted (DO NOT REMOVE THIS NOTE) | Logs: http://bots.wmflabs.org/~wm-bot/logs/%23wikimedia-office/ [21:00:51] #link https://www.mediawiki.org/wiki/Requests_for_comment/Shadow_namespaces [21:01:20] so this RFC has been renamed from "global scripts" and (entirely?) rewritten in the last day [21:01:48] yaaay [21:02:13] I see. I think the idea has merit [21:02:14] yeah, Fiona and I talked about it yesterday and updated it with where we'd like to see it go, it was kind of a mess before [21:03:02] Fiona? not just the RFC that was renamed then? [21:03:20] :P [21:03:20] Hi. [21:03:22] ah, we have Marybelle as well [21:03:38] Via telnet, even. ;-) [21:03:39] Global gadgets was part of the earlier RfC, but that is alreayd mostly implemented and was split out to https://www.mediawiki.org/wiki/Requests_for_comment/Global_user_preferences [21:04:08] Yeah, I didn't like the organization, so I re-did it. [21:04:15] help namespaces by remote loading is an interesting idea, I always assumed it would be done by XML export [21:04:39] Didn't we have Help pull from Meta-Wiki at some point? [21:04:43] The problem with XML exports is that you have to keep re-importing to stay up to date [21:04:54] I actually meant to note that in the RFC. [21:05:11] The other problem with XML is that it's XML. [21:05:31] and importing gets messy with usernames and attribution and if a local user has the same username as a user on mw.o or something. [21:05:32] I don't think anything ever actually pulled in help [21:05:50] I could swear we had some kind of hacky auto-tranclusion thing. [21:06:04] In any case, shadow namespaces could easily be adapted to the Help namespace, I think. [21:06:13] It's relatively easy to fetch a full Content object from another wiki's database. [21:06:23] Oo! so, could this be used to update help pages on non-WMF wikis? [21:06:40] ...if the database is directly accessible. [21:06:49] DanielK_WMDE: we'd want this to also work over the API [21:07:13] hexmode: yes [21:07:22] legoktm: both could work. but i see several issues. [21:07:29] one hard thing is purging [21:07:32] +1 [21:07:35] the idea is to get wikitext from the remote wiki and then do the parsing locally? [21:07:38] the other hard thing is not tripping over a thousand assumptions [21:07:46] hexmode: update is the wrong word though, it would always live on mediawiki.org, but be visible from the local wiki [21:07:58] TimStarling: I think you'd do parsing on the foreign side? [21:08:06] anywaaay.... if I click [edit], what will happen? [21:08:25] If you click edit, I think you'll get an edit notice saying "the page you're looking for is over here." [21:08:26] legoktm: ok, not ideal, but one step at a time [21:08:31] doing the parsing remotely is probably good for help pages, and certain kinds of templates [21:08:31] With optional Mario graphic. [21:08:41] but probably not so good for lua modules [21:09:07] I think you'd want stuff to be parsed locally so parser functions and things are run in local context, not foreign [21:09:34] maybe it should be configurable? [21:09:34] Doesn't that assume a sane local parser? [21:09:41] yeah, that is the difficulty [21:09:47] well, one of many difficulties [21:10:20] If we just spent 30 minutes listing the problems, I think that would be a good use of time. (Not kidding.) [21:10:28] That would allow Lego and I to re-work as necessary. [21:10:31] about the centralized help: https://strategy.wikimedia.org/wiki/Proposal:Wikimedia_Help_%28A_single_wiki_to_centralize_all_of_the_help_content%29#Summary [21:10:43] if you do local parsing, then what do you do when you see a template invocation in the fetched wikitext? [21:11:07] DanielK_WMDE: for purging, I was thinking on the local wiki you'd have a continuous script listening to rcstream and if a page that your wiki depends upon, it queues up htmlcacheupdate jobs [21:11:26] Continuous listening sounds challenging. [21:11:32] or what do you do if you see a require() in a remotely-fetched module? [21:11:50] TimStarling: You could process it like any other template, check locally and then check foreign repos, so it could allow for sub-overrides and stuff. We would probably want some batch lookup process [21:11:50] those two questions are very closely related in the current Scribunto implementation [21:12:16] since require() actually works by calling the parser's template fetch function [21:12:22] ty helder [21:12:27] so that it will be registered in templatelinks [21:12:32] legoktm: i'd much prefer push notifications. keeping a socket open scanning a cvonstant feed is more maintenance than most people are willing to do, i suppose [21:12:38] but possible, yes [21:13:08] I don't think we need to worry at all about editing right now. [21:13:20] For now I'd assume this is read-only on the local side. [21:13:25] or people could poll on regular intervals. Push will require usage tracking [21:13:47] Not sure why we'd need polling. [21:13:57] legoktm: subscriptions, yes. sonme pub-sub mechanism [21:14:00] We could just use caching like we do with Commons/. [21:14:00] legoktm: you would need to be pretty careful to name modules and templates such that they don't get accidentally overridden [21:14:23] hm... local parsing sounded good at first, but i'm a fraid it's a horrible can of worms [21:14:43] well, for modules, local parsing is essential [21:14:45] Can we talk about specific use-cases? [21:14:57] For user pages, local parsing is probably not needed? [21:14:58] for viewing, the remote can do the rendering [21:15:17] for *using* JS/Lua, whatever, no parsing is needed, no link updates apply, and no templates are being resolved [21:15:32] TimStarling: well when you're creating a new template, you'd see that one already exists at that name on a foreign repo, and choose a different name. iirc you can't upload locally over a commons image unless you're a sysop [21:15:38] Link updates kind of apply if we want to track usage. [21:15:49] I think the mechansims for using code on wiki pages, and viewing wiki pages, are rather separate [21:16:00] I'm not worried about name conflicts. [21:16:07] Most wikis don't have much content. [21:16:15] <^d> legoktm: Yes you can. [21:16:21] <^d> (upload over commons) [21:16:48] You're not uploading over anything, exactly. [21:16:53] <^d> True. [21:17:07] And I think the Commons integration works quite well. [21:17:13] Which is why I'm not really concerned about conflicts. [21:17:18] You need the reupload-shared right for that, which is only sysops on enwp [21:17:50] Restricting page creation is trivial. [21:17:58] What are the hard parts here? [21:18:04] For help pages and user pages I think remote parsing is fine. [21:18:25] How do you track usage? [21:18:33] I don't know if I am totally sold on the idea of the local module namespace being shared with the remote repository [21:18:46] in previous design discussions I think I said that I preferred having a prefix [21:18:57] a global module namespace? [21:19:03] You mean a prefix other than "Module:"? [21:19:24] instead of module: or in addition to module: [21:19:47] I don't think the chance of collisions makes it worrth it. [21:19:58] I think the vast majority of Wikimedia wikis have 0 modules. [21:20:09] So you're kind of solving for a problem that doesn't exist, no? [21:20:52] https://bn.wiktionary.org/wiki/%E0%A6%AC%E0%A6%BF%E0%A6%B6%E0%A7%87%E0%A6%B7:AllPages/Module: [21:21:01] I think probably most of them have more than zero [21:21:27] and the consequences of using the wrong code in an include() are potentially severe [21:21:39] and hard to debug [21:22:06] I'm not too sure how usage tracking would work, we could have sites that use our templates/modules/etc could register in our db somewhere, we give them a token, they send authenticated API requests with said token saying which templates they're using, and we send them push updates in return. [21:22:12] <^d> I wonder how many are original creations and not copy+pasted from other wikis. [21:22:33] Why authentication? [21:22:52] So random spammers can't fake usage [21:23:05] ^d: right, a lot of them are copy-pasted [21:23:06] ...and how many are just reinventing the wheel [21:23:31] or forking due to the originals not working without local customizations [21:23:40] but you can't randomly replace copy-pasted modules with modules of the same name but a different version and expect it to keep working [21:24:06] <^d> I would do it like we do files. Local version takes precedence. [21:24:14] Right. [21:24:15] <^d> So you wouldn't be using the hosted copy until you got rid of your local hack. [21:24:32]