[10:43:37] aude: Are you online? Do you have time to merge https://gerrit.wikimedia.org/r/358343 ? [10:45:41] :D [11:22:43] I long for the day when I can enter "instance of island" without having to avoid accidentally saving the statement as "instance of Iceland" :/ [11:23:25] Lucas_WMDE: http://localhost/repowiki/index.php/Special:JavaScriptTest/qunit?filter=entityselector [11:24:06] Lucas_WMDE: $wgEnableJavaScriptTest = true; [11:25:04] is today not a feiertag in berlin? [11:25:16] nikki: just like "avocado" is above the item for lawyers here, because one of the aliases is "advocaatpeer" in Dutch. :) [11:25:21] nikki: not in Berlin, no [11:25:26] Thiemo_WMDE: thanks :) [11:25:38] And I really want to punish those disambiguation pages for hitting the results so high. [11:27:00] haha, occupation: avocado... [11:27:42] I've been adding the ones I come across to https://www.wikidata.org/wiki/User:Nikki/Item_search so that I have a list of examples to test when someone tries to fix it, if you want to add dutch ones to it [11:46:27] sjoerddebruin: does https://phabricator.wikimedia.org/T147746 still behave that way for you? if I click "add reference" and then "insert reference", I get a js error [11:46:52] so I'm wondering if I'm doing something different or if it's just changed how it behaves [11:47:46] Why are you clicking add reference first and then insert reference? [11:48:18] it was an accident [12:03:02] Lucas_WMDE: it seems the constraints thing does know about the exceptions, how does that work? does someone extract them from the talk pages from time to time? if so, how often? and are there any plans to make it easier for users to mark something as an exception? [12:04:05] wondering 'cause it told me I just created a constraint violation and I'm not sure I can be bothered to go mark it as an exception, I'm busy :P [12:27:40] nikki: constraint exceptions are part of the constraint template, and were imported along with that [12:27:57] once we migrate to constraint statements, you’ll add them with P2302 instead [12:28:26] but for now, there’s no way to add an exception (you can edit the template on the property talk page, but we probably won’t reload the templates again before the migration) [12:30:56] hm... that's gonna make some pages huge [12:31:21] yeaaahh… look at https://www.wikidata.org/wiki/Property_talk:P225 [12:31:48] wow, that's even more than my huge list for gnis ids [12:31:59] IMHO that constraint should just be removed, that’s just absurd [12:32:02] but that’s not my call to make [12:32:29] oh, that’s also a huge list [12:32:51] I have no idea about that one, but for gnis ids at least, the constraint is normally true, there's just a specific subset where there are two ids instead [12:34:37] and it has been a useful constraint for me for finding things which needed fixing, even if the list of exceptions is ridiculous [12:35:12] we might need to think about a different way to collect exceptions [12:35:25] do those exceptions have some other statement that identifies them? [12:36:20] you mean another statement which could be used to filter them out? [12:36:32] yeah [12:36:38] instead of just having a list of exceptions [12:37:12] not at the moment :/ [12:37:34] hm [12:38:34] most of them can be identified as being ok because they have the same census code in the source data (which is how I generated the list in the first place, after I got sick of checking by hand :P), but we have no link between the gnis ids and the census codes [12:46:05] Thiemo_WMDE: https://gerrit.wikimedia.org/r/#/c/358343/ can be deployed in swat [12:46:26] unfortunately there is no swat on fridays, but maybe can be done on monday [12:46:46] it would be possible to take the source data and turn it into a file which lists all gnis ids which share a census code (which might be ok as a data file on commons or something) but then it would need to support something like that and I'm not sure if there are other cases where something like that would be useful [12:47:42] errr, today is thursday :) [13:00:59] Lucas_WMDE, nikki: Perhaps it would be interesing to flag the exceptions by using a qualifier for each statement causing the exception, not as a part of the property having the constraint [13:06:52] *for each statement causing the violation, I mean [13:07:21] that might be possible, if we can figure out which qualifier would make sense and how to describe the constraint [13:08:19] We can create a qualifier ad hoc [13:08:42] Something like "it's the same value as in" [13:08:55] ... for the case of unique value [13:10:05] Other constraints can have their corresponding qualifiers [13:12:41] I can't quite picture what you mean :/ [13:13:48] What part? [13:14:24] geonames id is another property which is usually a single value but sometimes has multiple, in particular, they sometimes have multiple ids if something is in multiple countries, since their data model only allows one country per id [13:14:38] I'm not sure what the qualifier and value would be [13:15:18] and how the constraints code would know that it's ok [13:19:12] They would simply be the ones we defined (we can create them for this), and the code should be modified so that, if one of these qualifier-value pair is found, its statement was marked as valid concerning the analysed constraint [13:20:52] Each constraint can have its own qualifier, or we can simple use something like P2303 for all of them [13:21:00] *simply use [17:03:46] nikki: Remember the years, made https://www.wikidata.org/wiki/User:Multichill/Kladblok to get an idea. PokestarFan made quite an incomplete mess [17:18:36] o/ [17:19:08] Get DanielK_WMDE_, I heard that the statement tracking stuff is low priority for the product team. I'm trying to figure out what that means from a practical perspective. [17:19:21] *Get --> Hey [17:19:24] multichill: fun :/ [17:19:42] I was only really looking at labels, since a load of them are wrong [17:21:33] nikki: Wrong in what sense? [17:24:26] some languages use different digits (not sure how *wrong* 0-9 is, but it's at least inconsistent to use those and they should be fixed for that reason if nothing else) and some languages need extra text to turn a number into a year (e.g. 2000 in chinese and japanese is "2000年") [17:25:35] Looks like we might be took late for DanielK_WMDE_, hall1467 [17:26:26] halfak: Yeah, it looks like it [17:28:31] Damn timezones [17:28:59] Haha yep [17:29:07] Essentially, I want to know if this means that Daniel will find zero time to work on it or he's just cautioning us that he may become unavailable at some point in the future. [17:30:13] If this is deprioritized such that we can't move forward, let's talk to Lydia_WMDE about what feeds in to prioritization so we can make sure that the strategic significance of this is considered during their next prioritization activity. [17:30:32] In the meantime, keep working on your proposal and we'll try to get that in front of Abraham :) [17:31:54] Sounds like a good plan :) [17:49:51] DanielK_WMDE_: ping about https://gerrit.wikimedia.org/r/#/c/355316/ [17:49:57] I think we can merge it now [17:54:31] nikki: I see, maybe we should just pick a year in the future (3017?) and have a bot create all the missing items, statements, etc, to get it over with [18:00:45] might work, although it would need to work out whether a year is already there and the existing ones are quite inconsistent [18:01:14] and I'd love something that would undo the edits where people have blindly copied the same text to every single possible language code :P [18:01:23] I also found a bunch where someone used nameguzzler to do that instead >_< [18:01:56] also there's a bunch of duplicates, some are duplicates of existing items, some are duplicates where pokestarfan managed to add the same year twice >_< [18:52:13] SMalyshev: the url encoding patch also bumps the version number of our schema. may want to announce this [18:52:28] i'm good to merge it tomorrow, have to run now! [18:52:37] DanielK_WMDE_: ok, tomorrow is fine [18:52:43] I'll announce once it's merged [20:15:55] SMalyshev, has there been any updates for the proper interlink space incoding? [20:16:15] yurik: hopefully will be merged this week and deployed next week [20:16:25] awesome, thanks! [20:16:42] watch T131960 :) [20:16:43] T131960: "_" character encoded as %20 in Wikidata URI RDF serialization - https://phabricator.wikimedia.org/T131960 [20:17:09] yep :) watching it carefully :) [20:45:18] nikki, do you still work on the future year item flaws, together with multichill? [20:52:58] I took a break to go have some food [20:53:50] and now I have something more important to deal with >_< [20:55:38] Hi, I posted this also to #wikidata-tech, but it was suggested that I ask it here as well: [20:56:35] I've been using the wikidata stand-alone for several months now without much trouble, but since last Tuesday I've been getting a lot of 429 errors. [20:57:09] e.g. Unexpected status code fetching RDF for https://www.wikidata.org/wiki/Special:EntityData/Q17577314.ttl?nocache=1497533660340&flavor=dump: 429 [20:59:01] Has some parameter changed that the runUpdate.sh script would suddenly be generating too many requests? [21:04:12] nikki, there was some help offered at WD:AN, but I had to refuse it [21:11:10] * nikki nods [21:48:35] PROBLEM - High lag on wdqs2001 is CRITICAL: CRITICAL: 30.00% of data above the critical threshold [1800.0] [22:23:36] RECOVERY - High lag on wdqs2001 is OK: OK: Less than 30.00% above the threshold [600.0] [22:53:19] EricInSanDiego: is that problem still happening? [23:06:16] SMalyshev: yes the problem is still there. [23:06:38] EricInSanDiego: werid, I can access that URL just fine. Maybe some kind of per-IP limit? [23:07:33] I am running this from my desk at home. [23:09:43] EricInSanDiego, SMalyshev: fwiw I also get an MWException when visiting that URL [23:09:44] Could this be related: https://phabricator.wikimedia.org/rOPUPf6d26fb482438bfbc32c13c08da28b34664a9926 [23:10:25] possible.... [23:10:47] try increasing batch size maybe? [23:10:56] The problem is still there, but it seems to be happening less often. [23:11:07] that would make less requests to RC API (though the same about of requests to wikidata) [23:12:22] may make sense to talk to #wikimedia-traffic maybe... [23:17:19] Can you remind me how to specify the batch size? [23:17:21] I wonder if we should make API on api/rest_v1/ for export formats... [23:17:31] EricInSanDiego: -b SIZE should work [23:18:17] if you are using runUpdate.sh that'd be runUpdate.sh -n NAMESPACE -- -b 500 or something like that [23:23:06] I'm getting an error: Invalid argument: com.lexicalscope.jewel.cli.ArgumentValidationException: Option only takes one value; cannot use [500] [23:23:52] I remember fixing my local installation to better accommodate rccontinue's, and may have introduced this error myself. [23:24:16] This was about 8 months ago, so I'll need to refresh my memory a bit. [23:26:19] I restarted with the -v option enabled. Will update the phabricator item if I can get traces that shed more light on teh problem. [23:26:59] It's been running 5 minutes now with no problems. [23:35:49] Still no problem. SMalyshev: let me let this run for another hour or two, and I'll get back to you with the results. Thanks for your help.