[08:44:00] hi, is here anyone who can help me to correct a wikidata entry? [08:45:34] Maaaybe. [08:45:45] https://www.wikidata.org/wiki/Q1190970 [08:46:01] What’s broken about it? [08:46:03] the property capital Saujil [08:46:22] the coordinates do not work with GeoHack [08:47:18] They do? [08:47:23] and look little bit groovy with the four zeros [08:47:30] I’m pretty sure those should be on the object for the capital itself. [08:47:57] (the ones there are less precise) [08:47:57] shouldn't they be a qualifier [08:48:01] ? [08:48:26] but more precise coordinates do not work too [08:48:33] with GeoHack [08:49:27] very precise: -28.1813264,-66.5140362 [08:49:55] Try it now. [08:50:24] that's 28°10'52.7750"S, 66°30'50.5303"W [08:51:32] I have tried now, GeoHack shows: 28° 0′ 0″ N, 0° 0′ 0″ E [08:53:12] You’re ‘very precise’ coords look too far west to me, actually. [08:53:28] https://tools.wmflabs.org/reasonator/?q=Q2165549&lang=en <- shows toe coords you gave me. [08:53:41] *the [08:53:42] it looks like a bug when using coordinates as qualifier [08:53:54] but I agree that they shouldn't be a qualifier there, they should just be on the item for saujil [08:53:54] i think so too [08:54:05] I didn’t change them on the department, just on the city itself. (shrugs) [08:54:52] (took them off as a qualifier) [08:56:36] doctaxon: https://tools.wmflabs.org/geohack/geohack.php?params=28.181326388889%20S%2066.514036194444%20W <- your coords still look too far west [08:56:58] yes I saw [08:57:01] but [08:57:19] I will correct it [08:57:19] It could easily be an alignment error in the map itself. [08:58:42] https://nominatim.openstreetmap.org/details.php?place_id=523692 <- FYI. [08:58:57] Gives the center point. [09:00:09] * nikki created https://phabricator.wikimedia.org/T118230 [09:01:08] these are good ones: 28°10'25.7110"S, 66°12'41.3518"W [09:04:05] thx for phab task, I just subscribed [09:04:35] Yeah, those come up in Potlatch as dead center in the Plaza. [09:05:32] (Potlatch is FYI the OSM editor that lets you overlay sat images) [09:06:59] about the middle of the village [09:07:33] Those coords are exactly in the middle of the square in the center of town, within a few dozen feet. :) [09:07:52] yeah hai [09:08:16] I love wikidata, wanted to have more time for it [09:08:42] what is a qualifier really meant for? [09:08:57] Like, when something started or stopped being true. [09:09:05] When you want to specify a year for a prize for example [09:09:33] and I specified a village with coordinates [09:09:43] seems to be okay I guess [09:10:51] On the Poznan Dept item, a qualifier on that property could be something like the year it became the capital. [09:10:56] *would be [09:11:52] but it are not temporal qualifiers only, are they? [09:12:10] but they are not temporal qualifiers only, are they? [09:12:20] one place where coordinates get used as qualifiers is for "mouth of the watercourse" [09:13:20] because there you want to indicate the specific point on the target item [09:13:31] https://www.wikidata.org/wiki/Special:WhatLinksHere/Q18615010 -< “Wikidata property for use as qualifier”… not that it’s only those, but gives you an idea. [09:13:54] Lydia_WMDE: can you state please, what property qualifiers standing for? [09:14:40] have you seen https://www.wikidata.org/wiki/Help:Qualifiers? [09:14:42] that has more examples [09:15:40] https://www.wikidata.org/wiki/Q5794 = Everton F.C. -> property “participant of” “1934 FA Cup Final -> qualifier “number of points/goals scored” = 3 [09:16:03] (as something totally random) [09:28:44] Always wondering why people forget P31:Q5 and P21 when describing humans. [09:31:23] sjoerddebruin: You happen to know if there is any plan to be able to get artwork metadata from WD anything soon? Wondering if writing crud onto file pages is a waste of time. [09:32:03] Revent: See https://phabricator.wikimedia.org/T89594 [09:32:57] I noticed someone futzed with Authority Control the other day, I guess in lua, didn’t know who to nag at. [09:33:31] Common’s version of that is now showing bullet points between empty fields. [09:33:42] Hm [09:33:54] https://commons.wikimedia.org/wiki/Creator:William-Adolphe_Bouguereau <- as a random example. [09:34:37] I didn’t see any edits to the actual ‘template’ when it changed [09:35:10] (I happen to have been the last person to edit the template itself, and my change didn’t cause it) [09:35:24] Can't test because of the high protection. [09:35:50] On the template itself, you mean? [09:37:16] (or on something it uses) [09:39:03] Oh, derr, actually reading the template now… [09:41:00] sjoerddebruin: Would not moving the * until after the | in each line fix it? [09:41:14] Idk, test it. [09:41:53] YEah, the last time I edited that it spammed the crap out of the job queue, need to be careful. :P [09:43:30] Ah, no, breaks the template completely. [09:48:43] sjoerddebruin: FYI, this is what broke it… https://commons.wikimedia.org/w/index.php?title=MediaWiki:Common.css&diff=178599631&oldid=172094738 [09:49:01] Okay.... [09:49:43] So I guess I need to go yell at him, lol. [09:52:53] doctaxon: can you give an example? i am not sure what you mean [10:16:00] Has anyone around today used pywiki bot for a wikibase instance other than wikidata? [11:04:06] addshore: do we have dispatch stats in graphite? [11:04:20] not yet [11:04:52] what are the reasons for not yet? [11:05:28] * aude used to have a tool for this on labs and could hack that up / have graphite in labs for this + experiemnts [11:05:46] but if putting in the real graphite is easier/better, then would prefer that [11:10:52] aude: link me to the code? [11:11:58] Just havn't got around to that one yet in the big list of things to do ;) [11:12:29] * aude never put it on github :( but is trivial [11:12:43] but used rrdtool and i would use graphite today (both similar) [11:12:55] yeh, just an api call every minute right? and spit the result into graphite [11:13:00] yep [11:13:02] I can probably get one up today if you want :) [11:13:20] *goes to remember where it is in the api* [11:14:11] it's somewhat urgent now since we are trying to reduce the number of concurrent cron jobs [11:14:25] okay, can you remmber which api module it is in? [11:14:39] and otherwise do the real fix, of not so many db connections + use job queue [11:15:03] * aude looks at the api [11:15:47] I thought it was this.. https://www.wikidata.org/w/api.php?action=query&meta=siteinfo but no :P [11:16:01] https://github.com/christopher-johnson/check_dispatch/blob/master/check_dispatch#L58 [11:16:33] https://www.wikidata.org/w/api.php?action=query&meta=siteinfo&format=json&siprop=statistics [11:17:05] my tool predated the api, so did horrible screen scraping :o [11:17:14] api is waaaaaaaaaaaaaaaaaaaaaaaaaaaay better [11:20:08] :D [11:20:15] cool, I should be able to get it up reallly shortly ;) [11:20:21] \o/ [11:25:40] ugh :( https://phabricator.wikimedia.org/T118162 [11:26:08] I'm at a training seminar for the rest of the week, so don't know when i'll have time to look into this... [11:26:39] aude: which numbers exactly do you want? ;) [11:27:12] there are 8 right? freshest, meanest, stallets, and avg, pendig and lag? [11:27:36] i wonder why it would open so many connections. is that the issue you were discussing? [11:33:01] aude: so stats like wikidata.dispatch.avg.lag would be cool? [11:37:48] average would be good [11:37:59] DanielK_WMDE: it is the issue [11:38:02] primary issue [11:38:54] DanielK_WMDE: i think the way the dispatcher selects clients could be done differently [11:39:05] but probably not trivial to change [11:39:35] aude: there'S a lot of things that should be done differently... i think the most urgent question (and hopefully easiest to fix) is: [11:39:39] why does it hog connections? [11:39:44] yeah [11:40:12] my guess is that it fails to call LoadBalancer::reuseConnection under some circumstances [11:40:21] maybe when a query times out? [11:40:31] maybe or when the script stops [11:40:31] that would be bad, making an already busy system more busy [11:41:06] when the script stops? that should close all connections by killing the socketes, if nothing else... [11:43:41] aude: itll be in graphite in the next 15 mins ;) then ill like you to where it is ;) [11:43:48] addshore: yay! [11:43:56] i'll be offline in a minute. will come back online some time tonight [11:46:15] hashar, aude - hi [11:46:28] we have test failures in ContentTranslation again [11:46:40] https://integration.wikimedia.org/ci/job/mwext-qunit/7759/consoleFull [11:47:38] aharoni: huh [11:47:49] aharoni: see #wikimedia-releng :) kartik poked about it [11:47:58] apparently the db is too slow, so need to recheck [11:51:08] aharoni: aren't you in #mediawiki-i18n ? :D [11:51:31] aude https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.pending and https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.pending [11:51:37] wait, * and https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.lag [11:51:50] yay! [11:51:57] *goes to hunt for the phab tickets related to this* [11:52:05] super helpful [11:56:02] aude: https://phabricator.wikimedia.org/T75647#1795482 links to the code im using, we may want to try and slot it into puppet at some point, rather than me running it on stat1002 [11:56:23] hashar: not as consistently as I should be [11:57:41] aharoni: it's not clear to me why it broke just now and how to fix :/ [11:57:59] if it's an issue with jenkins or wikibase [12:06:02] think i see the issue now [12:13:29] Is it a bug that I can see (e.g. https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.lag) but I don't have a login for graphite.wikimedia.org. If I knew the right target name would I end up with non public data? [12:19:43] tarrow: nope, the api is public :) [12:19:44] afaik, i think things were changed recently to allow graphite to be public [12:19:47] https://phabricator.wikimedia.org/T108546 [12:20:09] the graphite render api has always been public ;) [12:20:14] ah [12:22:01] its hard to store numbers that constitute private information ;P [12:22:22] also, for nicer graphs see mhttps://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.pending&from=-1hour&lineMode=connected and https://graphite.wikimedia.org/render/?width=586&height=308&target=wikidata.dispatch.*.lag&from=-1hour&lineMode=connected [12:24:04] Ah, so it doesn't store numbers related to specific ips for example? [12:24:11] nope [12:24:19] nothing like that is in there :) [12:42:11] hehe https://usercontent.irccloud-cdn.com/file/7W4RsGBf/ [13:32:49] Hey [13:34:35] I try to get all Version of MozillaFirefox with a query [13:34:37] https://trash.ctdo.de/a/162fzbhk443 [13:34:55] huh... how can two items have the same sitelinks? https://www.wikidata.org/wiki/Q21082332 and https://www.wikidata.org/wiki/Q21082333 [13:35:01] also https://www.wikidata.org/wiki/Q21127593 and https://www.wikidata.org/wiki/Q21127594 [13:35:08] But i wont work, and i am to dumb to find the solution [13:37:50] as far as I understand it, "?version p:P348 wd:Q698" would look for items which have a P348 statement with Q698 as the value. P348 is a string property, so it's never going to link to Q698 [13:38:40] Q58... obviously this had to be one of the first items... [13:39:45] The funny thing is, for some reason my script is chocking on it [13:39:50] >_> [13:41:34] "wd:Q698 p:P348 ?version" might be what you want? although that prints things like "wd:statement/Q698-06dcb02c-4904-b7bc-3117-df64b39b2bdd", not sure how to get the actual text from it [13:44:02] wdt: instead of p: brings just a string [13:44:30] But really just one string [13:44:30] yeah, just tried that, but then I only get one value [13:47:42] ah... https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service/queries says "Such triples, using the wdt: representation of the property, are not created for all statements, but for only statements that have the highest rank for the property on that item and are not deprecated." [13:48:08] the current version is set to preferred and the rest are set to deprecated so that's not going to return them all... [13:53:40] Thanks a lot [13:53:52] Thats exactly what i need [13:54:00] Fits 100% in my case [14:08:22] another pair... https://www.wikidata.org/wiki/Q21162675 and https://www.wikidata.org/wiki/Q21162676 [15:23:15] does anyone know if flow still sends notifications every time someone edits a page on your watchlist? [15:27:40] echo* [15:27:58] And nope, it's not part of that either. [15:28:07] * nikki is confused [15:28:22] It's a pre-notification system thing. [15:28:34] https://www.wikidata.org/wiki/Special:Preferences#mw-prefsection-personal, lowest option [15:29:50] when flow was first enabled, some people whose talk pages I have on my watchlist converted their talk page to use flow and suddenly I started getting the echo notifications every time someone did anything on the talk page [15:30:37] Oh, talk pages. [15:31:09] Yep, you can just remove those from your talk page and only watch the topics that you care about. [15:31:41] which was really really annoying, so at first I took them off my watchlist to make it stop, but then I found the option to disable flow notifications, which meant I could at least add them back to my watchlist (although I still miss most of the edits since they don't always show up on there) [15:32:36] but now I just noticed that I'm not getting notifications for mentions [15:32:44] (on flow pages, works fine on non-flow pages) [15:37:16] anyway I was wondering if it still does that or if it was a bug [18:02:18] re [18:02:30] aude: what's the status of the dispatcher issue? anything i can do?