[00:25:24] SMalyshev: how about enabling conversion on test.wikidata? [00:30:50] DanielK_WMDE_: it needs separate config then... config is linked to real entities, not sure what would happen if they don't exist [00:30:56] maybe it'll be ok... [00:31:15] we could try :) [04:05:50] Q: Is it typical for the query service to be displaying data in certain cases that is more than a week old? [08:57:13] Happy birthday, Q18016466! :D [09:34:38] dispatch lag going up again :( (but dispatch *pending* still going down, slightly) [09:37:00] so I guess that means there’s a big block of old changes that grow old (⇒ lag ↑) faster than they’re being dispatched (⇒ pending ↓)? [10:01:49] lydia is so unreferenced :( [10:02:11] Hm? [10:02:43] earlier abian said "Happy birthday, Q18016466! :D" [10:03:50] ;) [10:09:22] well I added a couple more :P [10:09:26] I was hoping for a bigger drop. https://grafana.wikimedia.org/dashboard/db/wikidata-dispatch?refresh=1m&orgId=1&from=now-7d&to=now [10:09:47] But Emijrpbot is editing at 133 per minute... [10:10:15] And a lot more http://wikidata.wikiscan.org/?menu=live&date=6&list=users&sort=weight&filter=all [10:10:46] Look at that spike. ;) http://wikidata.wikiscan.org/gimg.php?type=edits&date=24&size=normal [10:11:19] it's like whack-a-mole ;_; [10:22:52] nikki: https://www.wikidata.org/wiki/User_talk:Emijrp#Emijrpbot_2 [11:41:02] heh... the map view doesn't cope very well with two points either side of the international date line [11:42:46] rather than two dots very close to each other, I get almost two whole maps of the earth and two dots nowhere near each other visually [12:17:46] nikki: I'm just going to ask every bot that edits items with sitelinks to postpone for one week. [12:19:19] ouch, I thought this mostly affected cebwiki, but apparently the stalest wiki right now is enwiki https://www.wikidata.org/wiki/Special:DispatchStats [12:19:36] Oh dear, didn't notice that. [12:19:51] It was ceb for a while [12:20:29] sjoerddebruin: by this point you might as well announce it on project chat, right? [12:20:38] call it a “voluntary moratorium” or something ;) [12:20:40] Mailing list etc [12:30:54] +1 [12:31:26] Done [12:32:03] "Your post has been submitted and is pending approval by an admin." in the Facebook group... :/ [12:34:13] Should I also make a Phabricator task? [12:35:12] I’m not sure I understand… what would the task be for? [12:35:16] just tracking? [12:35:34] idk [12:42:27] hey hoo [12:42:33] hi sjoerddebruin :) [12:43:26] Are you aware of the dispatch issues? [12:43:38] vaguely, yes [12:44:20] Trying my best to combat it, not sure what is useful. [12:46:34] The average is going down when I ask several bots above 60 edits per minute to stop. [12:56:02] I don't really know what's going on there, to be honest [12:59:41] Who does? [13:08:21] Maybe ladsgroup does? [13:08:55] Amir1 [13:11:18] yeah [13:11:27] I know. Just pinging him. :P [13:22:35] Getting https://gerrit.wikimedia.org/r/336542 in finally would be great [13:24:24] That will help with this? [13:24:49] No, but it's just generally annoying to see things lying in gerrit for months [13:46:38] ohh I finally realised that "change dispatch statistics" means the dispatch statistics *for* changes, not that it's a page for altering the statistics [13:46:53] I was wondering why we would want to fake the stats [13:52:03] sjoerddebruin: hey [13:52:08] I missed your ping, sorry [14:06:15] Amir1: no problem! Do you know something about dispatch? [14:07:33] sjoerddebruin: yeah, we checked that, the median is fine [14:07:50] means only one or two wikis get stuck [14:07:56] enwiki and cebwiki probably [14:08:01] every other one works just fine [14:08:09] but I need to dig deeper [14:08:22] I do it when I have my rights back [14:08:32] Yeah, you had some problems... [15:13:56] Hello, is there a way to use SERVICE wikibase:label for having a ?itemAltLabel list not separated by commas but, e.g., by pipes? [15:24:07] is there a front end to make Wikidata queries for users who don't speak SPARQL? [15:27:10] like, suppose I wanted to find everyone whose P21 (sex) was listed as Q6581072 (female and not Q1052281), but who was listed as someone's P22 (father) -- how much coding would that entail? [15:27:57] hi [15:28:15] I don't know if there's anything like that [15:28:20] Pr0methean: not much coding :) http://tinyurl.com/ycgwr97o [15:28:31] we have https://www.wikidata.org/wiki/Wikidata:Request_a_query though [15:28:39] but I’m not sure if we have any interface that would’ve helped you with that query [15:29:00] there is a query helper on query.wikidata.org, but if I’m not mistaken it’s not useful for this query [15:30:50] instead no way for personalizing labels output on the same row? :,( [15:31:39] I propose that we work on building a GUI that people who don't speak SPARQL can use to compose their queries [15:31:45] Ciccio: I don’t think that’s possible with the label service, I think you’d have to get the aliases with triples and use GROUP BY and GROUP_CONCAT [15:32:07] we can call it the Turbo Web Interface for Locating Information Through Execution of SPARQL [15:32:13] or TWILITE SPARQL for short [15:33:08] Ciccio: like this: http://tinyurl.com/yc4htntd [15:33:55] Lucas_WMDE: great! Thank you. : )))) [16:14:59] nikki: your dream https://www.wikidata.org/wiki/Special:Contributions/VIGNERON [16:15:25] ;_; [16:15:42] Magnus is aware now, hope he builds something. [16:16:16] at least they're adding descriptions in a language they speak [16:16:21] * nikki clutches at straws [16:17:27] Average dispatch is still dropping. [16:17:34] or at least not as high as before [16:42:17] Lucas_WMDE: do you know if there's a way to *not* select certain variables, other than selecting everything else? [16:42:28] I think that’s the only way [16:42:30] bah [16:42:54] sometimes I end up with an unwanted variable as a byproduct of getting to some other value, then it messes up my "select *" :P [16:43:19] "unbind" would be nice there [16:43:34] I almost never use SELECT * because it doesn’t work with the label service [16:43:46] and sometimes typing out the SELECT works quite well to plan out the query in my head [16:43:53] I never use that, partly because it doesn't work with select * [16:44:24] but mostly because I like selecting specific languages and having multiple variables for different languages [16:44:34] like right now I have english as ?label and german as ?de [16:44:41] ok [16:44:52] oh, I wonder if you can use the label service multiple times… [16:45:49] haha, it works :D http://tinyurl.com/ybaqlrhf [16:45:56] but not with SELECT *, wtf [16:46:34] oh, cool [16:46:49] it should work with select * if you add patterns explicitly to service [16:47:14] not sure why it doesn't, looks like a bug :( [16:48:00] yeah, I thought the same, it doesn’t have to guess the label associations because I specified the rdfs:label [16:48:09] even more funny: http://tinyurl.com/yc26ymb9 [16:48:18] I'm pretty sure there's some bug there... [16:48:58] huh [16:49:04] disabling the optimizer doesn’t fix it either [16:49:55] nah it's not the optimizer, looks like it's something inside how * works screwed up... will need to look into it [16:50:24] e.g. http://tinyurl.com/yc4sxtsq - very weird result [16:51:04] but explicitly selecting ?en and ?reallyen doesn’t work either [16:51:13] ?en gets a value and ?reallyen stays blank [16:51:39] hmm yeah I wonder what's up with that? [16:52:34] there's a bug where blazegraph reorders the query so that SERVICE is always last... maybe that's also it [16:52:50] i'll look into it [16:52:51] I thought the label service basically happens last, so if you try to do anything with the variables, it never works [16:53:16] (like if you want to extract part of the label) [16:53:41] I’ve created a task for the first bug (label not in *): https://phabricator.wikimedia.org/T171194 [16:54:20] nikki: it doesn't have to always happen last, but I think the bug makes it to happen last [16:54:36] Lucas_WMDE: ah, thank you! [16:59:23] SMalyshev: while you're here, is there a way to make the query service map plot points with negative longitudes on the right of the map? (i.e. so that I can see things in the pacific without half being on one side of the screen and half on the other) [16:59:48] I tried adding 360 to the longitude since apparently that would make leaflet plot it on the right rather than the left, but that just gave me a java error [17:00:49] and then I ran out of ideas :P [17:04:47] nikki: hmm... java error? what exactly did you do? [17:08:47] http://tinyurl.com/ycbarz7a the commented out line is the one which gives me an error [17:09:06] if I only add 180, it works, so I assume it just doesn't like coordinates being outside -180 to 180 [17:09:49] but that’s an error in the remote service, isn’t it? [17:09:58] is it? hm [17:10:34] I get the same error if I move it outside the service bit [17:10:46] I think so, the query in the error message is just the inner query [17:11:01] really? I tried moving it out and got 0 results (but no error) [17:11:26] but wait, the error makes sense [17:11:29] “BIND clause alias '{}' was previously used” [17:11:35] they’ve messed up their error message formatting :D [17:11:44] but you’re BINDing ?coords2 twice [17:11:55] per SPARQL spec that’s not allowed (though BlazeGraph doesn’t complain) [17:12:02] I'm only using one of the lines at a time [17:12:06] oh, sorry, nevermind [17:12:06] yeah [17:12:07] I guess I should've been clearer about that [17:12:10] I just realized that makes more sense :D [17:12:30] okay, now it looks like the error is on our end [17:12:36] nikki: yeah coordinate should be [-180; 180] iirc [17:14:13] nikki: http://tinyurl.com/y9sxrgfw [17:14:36] geof:distance doesn’t like the adjusted coordinate, so give it the original one but select the adjusted one :) [17:14:43] oooh [17:14:44] <3 [17:14:52] but no more SELECT * :P [17:14:57] I'll live :P [17:16:38] what does that query do, by the way? is it a coincidence that it has only one result, and that result is across the date line? [17:17:27] it selects the coordinates in wikidata and the coordinates in the tgn item linked from wikidata and maps them [17:18:27] so that I can see more visually whether they look like they're for the same item [17:18:44] okay [17:18:55] oh, I didn’t notice the VALUES clause, that’s why there’s only one result [17:18:59] yeah :D [17:22:33] and of course if they're on different sides of the screen, I can't really tell whether they're near each other or not [17:35:18] now I can do things like http://tinyurl.com/y9gbh6zz too :D [17:35:35] (map of all the things in fiji) [17:52:21] nikki: should we warn https://www.wikidata.org/wiki/Special:Contributions/VIGNERON? [17:53:35] I guess so, seems unfair to tell some people to stop editing too fast and not others :) [17:58:31] I added the fiji locations as an example on https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries/examples#Locations_in_Fiji, that's the first time I've added an example so hopefully I didn't break anything in the process [18:06:48] that fiji query helped me spot a couple of incorrect coordinates too :D [20:16:37] Disrespectful cunts. https://www.wikidata.org/w/index.php?title=Q19198&type=revision&diff=524258502&oldid=524254699 [20:18:15] yup [20:28:54] is anyone else getting a bunch of pages without edit links? [20:29:24] not all the time, but I've had to purge a bunch of pages in the last couple of hours to get them back, 'cause refreshing didn't help [21:24:14] nikki: VIGNERON also stopped, it seems. [21:24:35] Go dispatch go! :P [23:16:30] hello room. May someone help a little bit? [23:17:25] a short query with duplicated items, need a help how to remove duplicates without losing useful information [23:18:54] anybody there to have a look on the query? :) [23:28:16] guest_24: I can't, but if you posted your query you'd save one roundtrip of communication :) [23:31:42] you might be right. Here is the query: http://tinyurl.com/ycnovuta [23:32:41] I noticed that it has another mistake with the place of birth, since the items are not limited to the desired countries [23:33:12] will wait for a while :)