[07:19:14] https://www.wikidata.org/wiki/Q3404311 I can not add statements.. is that a known bug ? [07:20:02] tried it in three different browsers [07:31:59] Donno what this all about [07:43:13] the item has no statements and I cannot add that it is an award [07:47:17] this is already known [07:47:28] threads on WD:PC and WD:AN [07:47:44] some software bug likely [07:48:11] workaround: purge it manually and it should be fine again for editing [07:57:01] how do you purge ? [07:57:12] and what is PC and AN? [08:02:35] append ?action=purge to the URL in your browser, and it will offer a dialogue to rebuild the item page on the server [08:03:09] PC=project chat, AN=administrators' noticeboard [08:03:15] Thanks :) [08:04:00] links: https://www.wikidata.org/wiki/Wikidata:Project_chat#Item_pages_not_loading_fully and https://www.wikidata.org/wiki/Wikidata:Administrators%27_noticeboard#Missing_translation_fields [08:07:15] It's all a secret plan to lower the dispatch :P [08:07:51] I doubt that [08:07:59] it is just a bug [08:08:10] Hence the :P emoji [08:08:30] as far as the dispatch is concerned, I do think that disabling recent changes is to be preferred over less edits [08:08:32] and I still doubt that the dispatch problem is primarily related to the edit rate ;-) [08:09:02] Well, it's still going down. [08:09:10] wikidata is not there for Wikipedia, it is a project in its own right.. [08:09:17] yes, it certainly helps to lower the edit rate [08:09:45] but I don't think it was caused by the edit rate [08:10:20] (lower the *dispatch lag*, sorry) [08:10:54] I agree with that. [08:11:23] Still, doing like 800k edits on one day without approval is wrong. [08:11:51] according to the grafanas, something happened between 21:15 and 21:30 on 2017-06-28 which caused the dispatch problem [08:12:35] correct, that should not happen [08:14:03] sjoerd.. exposing changes and approval is not the same thing [09:27:30] If you click https://www.wikidata.org/wiki/Q24074537 , do you see edit buttons in the statements? [09:29:13] multichill: no [09:30:12] Probably hitting https://phabricator.wikimedia.org/T170668 ? [09:30:48] I have this error indeed [09:37:28] Lucas_WMDE: are you around? (are you working on saturday? o_O) [09:37:35] I need help again, you see :P [09:47:15] multichill, pintoch: I manually purged this item, now there are edit buttons again. Purging is the most simple workaround for this problem at the moment [09:47:58] nikki: I just wanted to check out what happened on Project Chat, that’s why I opened my work laptop… let me reply to you from my private system :D [09:48:10] hi! [09:48:22] hey! [09:48:56] I have this query http://tinyurl.com/y8r7zlxe and when I uncomment the geof:distance line at the bottom, I get an error I don't understand [09:50:18] from the original query (with the line commented out) it looks like ?coords1 is a value node (wdv:…), not a simple value [09:50:27] and I guess the distance function only supports simple values [09:50:31] ohh [09:51:23] fixed :D [09:51:24] thanks! [09:51:56] damn, you’re faster than me :D here’s my fix: http://tinyurl.com/ybgvb6nt [09:52:15] I’m a bit surprised that there’s no way to directly get from the value node to the simple value, apparently [09:53:54] ah... I just added wdt:P625 but yours is more accurate I assume, since mine doesn't make sure I'm getting the simple value from the same statement [09:54:28] yeah [09:58:31] MisterSynergy: I left it in the state so someone working on the problem has an item to test with [09:59:14] WikidataFacts: https://www.wikidata.org/w/index.php?title=Property_talk%3AP650&type=revision&diff=521397187&oldid=492650825 <- eh? [09:59:20] Is that the thing you were working on? [09:59:38] it's happening to lots of people, if someone needs something to test with, it shouldn't be too hard to find another [10:00:10] multichill: yup [10:00:24] though I didn’t do the actual migration [10:00:57] Is the goal to comment out the templates? [10:01:20] well the goal is to have the constraints on statements [10:01:29] How do I find https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations/P650 if all the templates are commented out? [10:01:29] some of the boxes on the talk page are already automatically generated from the statements [10:01:31] and as part of that, they should probably be removed from the talk pages, otherwise it’d be confusing [10:01:32] (most? all?) [10:01:49] nikki: yeah, that too [10:02:24] WikidataFacts: Better, just a template that generates it on the fly based on the statements on the property [10:02:45] yes, that’s what happens on https://www.wikidata.org/wiki/Property_talk:P650, isn’t it? I still see the constraints listed and links to the lists [10:02:48] Just like the property documentation [10:02:52] (seems to be slightly broken at the end btw) [10:03:39] I'll just remove it [10:05:02] WikidataFacts: https://www.wikidata.org/wiki/User:RKDdata was already confused by it :P [10:05:34] Nice system, looks good! [10:06:50] multichill: cool, thanks! [10:07:02] (RKDdata didn’t recognize that the template was wrapped in a comment?) [10:09:49] That and the report was gone for a couple of days. He is an art historian, not the most experienced Wikimedia. He actually maintains the database we link to and fixes errors based on what we find :-) [10:10:11] sorry, I thought it was a bot, not a person :D [10:10:32] Would be good to have more links to https://www.wikidata.org/wiki/Help:Property_constraints_portal and pages like https://www.wikidata.org/wiki/Help:Property_constraints_portal/Format [10:11:13] Maybe make the bold text on https://www.wikidata.org/wiki/Property_talk:P650 clickable? [10:12:31] WikidataFacts: ^ [10:13:09] multichill: sorry, having breakfast, bit slow to answer :) [10:13:22] if you use the gadget, the help icon next to a report points to the constraint type help page [10:13:37] not sure if I should edit the templates [10:20:30] WikidataFacts: That's brunch by now ;-) [10:20:32] I added an error reports table to a few properties recently... I wonder if it would make sense to put them on a separate subpage [10:21:34] multichill: I cite Q18614232 in my defense :P [10:22:00] :D [10:23:36] I actually sometimes do that before traveling to another timezone ;-) [10:49:01] nikki: More Japanese speaking people seem to be working on https://www.wikidata.org/wiki/Wikidata:WikiProject_sum_of_all_paintings/Possible_paintings , but it's still a lot! [10:49:57] there's a lot I can't help much with, since my extremely limited understanding of japanese doesn't extend to art :( [10:50:15] it's not a topic I know much about in *english* [10:51:09] I have seen quite a few items which are people [10:56:04] Yeah, I think the category tree is quite messy on the jawp [10:57:22] like I picked one fairly randomly - https://ja.wikipedia.org/wiki/%E8%8A%B3%E5%B7%9D%E6%98%A5%E5%B9%B8 that says they're an ukiyoe artist from the edo period (unknown birth/death dates), not sure what to enter (got as far as "instance of human" :P) [10:58:26] That's already enough nikki! [10:58:35] That will remove it from the list [10:59:28] but still! would ukiyoe artist be a new occupation or should it just be artist + genre? [10:59:45] and which property would I use to say they're from the edo period? [11:02:05] there's a category for them https://www.wikidata.org/wiki/Q7028503 at least [11:02:40] https://www.wikidata.org/wiki/Wikidata:WikiProject_Visual_arts/Item_structure [11:02:48] Should be on there somewhere [11:03:35] Look for culture / period [11:04:30] I guess period then [11:04:34] It's a genre [11:04:50] So you can add genre or movement to the person with https://www.wikidata.org/wiki/Q185905 [11:05:05] But I haven't really done a lot in that area yet [11:07:26] adding the Ukiyo-e artists from a category is not that hard to do [11:07:42] if you want me to I do it for you [11:08:30] I'm ok [11:32:25] WikidataFacts: I'm getting constraint errors about invalid regexes (like the "format as a regular expression" statement on https://www.wikidata.org/wiki/Property:P10)... is that a problem with the constraints thing itself? I think so, but maybe I'm confused and someone has added a bad statement somewhere [11:34:10] I can't wait for this problem with the edit links to be fixed... [11:34:59] having to purge most of the pages I want to edit before I can edit them is no fun [11:45:14] nikki: can you show me an example statement? [11:45:16] nikki: just do like everybody else, run high-speed bots instead of using the UI (preferably without seeking consensus first) [11:48:07] is there a pattern in what needs to be purged ? [11:49:13] maybe because I added items [11:49:29] it only happened twice so far [11:50:23] WikidataFacts: I first saw it on a name in kana statement, like on https://www.wikidata.org/wiki/Q30936435 [11:51:04] I thought maybe it didn't support the \p{} syntax, but then when I checked the property, its format as a regular expression statement also had an error, and when I checked other properties like the video one, that did too [11:52:58] that’s different issues [11:53:32] ah [11:53:50] `[\p{Hiragana}\p{Katakana}ー・　]+` is not a valid regex for the Wikidata Query Service, because those character groups aren’t available in Java regexes (which is what WDQS currently uses) [11:54:22] and P1793 “format as a regular expression”, as a regular statement, has nothing to do with constraints at all [11:54:32] but someone placed a format constraint on that property as well [11:54:39] they're not? I thought java supported them years and years ago [11:54:49] using a completely crazy regex to recognize regexes, from StackOverflow [11:54:56] (which I don’t think is CC0, by the way) [11:55:02] but which is also not supported by Java, I think [11:55:21] at least I have vague memories of only perl and java being able to do them and wishing other languages would add them, but maybe I'm confusing them with something else [11:55:25] nikki: I just tried it out, “java.util.regex.PatternSyntaxException: Unknown character property name {Hiragana} near index 12” [11:55:37] java supports some \p{} groups, but apparently not Hiragana [11:55:45] that's silly [12:04:12] does IsHiragana work? [12:04:42] oh, apparently it does [12:04:48] but presumably that won’t work for KrBot then [12:05:01] what does krbot use? [12:05:09] PCRE [12:05:14] see also https://www.wikidata.org/wiki/Help:Property_constraints_portal/Format [12:07:02] yeah, `grep -P` doesn’t like the IsHIragana version of the pattern [12:07:38] fun [12:07:55] yay regex flavors [12:09:27] perl doesn't care what you want to use. it's perfectly happy with Hiragana, IsHiragana, InHiragana (iirc that's something slightly different), sc=Hiragana *and* script=Hiragana [12:09:48] (those latter two I found on some page about java... grep doesn't like them) [12:16:45] * pintoch is thinking about adding "property for this type (P1963)" claims based on this query: http://tinyurl.com/y9tpod5m [12:18:39] because there are currently 4154 properties which have not been linked to any type with P1963 [12:19:09] nikki: I guess that must be one of the differences between Perl and PCRE then [12:20:08] yeah [12:20:21] somewhere we say pcre so I guess I'll leave it for now [12:38:15] WikidataFacts: https://www.wikidata.org/wiki/Special:ConstraintReport/Q42 [12:38:38] What's wrong with the constraint report of Q42? :S [12:38:43] abian: https://www.wikidata.org/wiki/Wikidata:Project_chat#Constraint_statements_update [12:38:52] Q42 has a Commons category [12:39:13] Commons category has a constraint with an extremely long exception list, which is truncated in the database [12:39:16] so the constraint check crashes [12:39:29] and I can’t fix this because no deployments on the weekend [12:39:44] *you* can, though, by removing that constraint :) [12:42:31] Done :) [12:42:36] Thanks, WikidataFacts [12:42:58] oooh, that’s great, thanks :D [12:44:39] I also see three ⧼wbqc-constraintreport-status-deprecated⧽ in https://www.wikidata.org/wiki/Special:ConstraintReport/Q42 [12:44:43] Is this also pending? [12:46:14] yes, that’s me being an idiot, sorry [12:46:23] fixed on master, but not severe enough for a backport [12:46:34] hopefully there’ll be a deployment this week [12:46:53] No problem, WikidataFacts [12:48:28] As you submit many patches, it's normal for you (not an idiot) to make a mistake from time to time :) [12:48:58] regarding the Commons category constraint… as a WMDE employee who shouldn’t do too many edits on Wikidata, I’m basically reduced to asking “will no one rid us of this bothersome constraint?” :P ([[Q30675091]]) [12:48:59] 10[2] 10https://www.wikidata.org/wiki/Q30675091 [12:49:08] I hope no one will attack you over removing it [12:51:53] A constraint with so many exceptions means that something isn't going very well... [13:02:39] WikidataFacts: What about contemporary constraint? It's expensive to generate a report with all the violations in Wikidata, but it shouldn't be expensive to know if there's a violation for a given item (constraint report) [13:03:33] not implemented yet, and I don’t know what the exact semantics are [13:03:49] is there some hard-coded list of properties it considers? [13:04:37] I provide an example here... https://phabricator.wikimedia.org/T141859 [13:05:08] But, indeed, I'm not too enthusiastic about the idea of having a hard-coded list of properties [13:05:45] oh! I wasn’t aware of that task [13:05:48] thanks [13:05:59] I think I might even have created a duplicate [13:06:05] Simply, there's a violation if there's a null intersection between the range of time existing item A and item B [13:08:24] If A started after B ended, or B started after A ended, there's a constraint violation [13:12:42] "Start" is defined by a set of properties (e.g., {"start time"; "date of birth"; "inception"}), and "end", by another set of properties (e.g., {"end time"; "date of death"; "dissolved, abolished or demolished"}) [13:15:48] yeah, those properties should probably be part of the constraint parameters, not a static set known to the checker [13:16:02] what happens if an item has e. g. both start time and date of birth? [13:19:02] The constraint is "weak", a combination of these possible statements being satisfied is enough to consider the constraint satisfied [13:19:18] That's why I use ORs [13:19:19] ok, makes sense [13:19:30] use the widest possible range on each end [13:21:36] earlier begin date, latest end date, I guess? [13:21:40] earliest, I mean [13:21:45] yeah [13:23:39] Right [13:31:22] PROBLEM - wikidata.org dispatch lag is higher than 300s on www.wikidata.org is CRITICAL: HTTP CRITICAL: HTTP/1.1 200 OK - pattern not found - 1952 bytes in 0.105 second response time [13:37:36] multichill: I've marked a whole bunch as people, japanese should no longer be the one with the most :P [14:18:41] wtf is happening to the average lag wtf is happening [14:18:44] https://grafana.wikimedia.org/dashboard/db/wikidata-dispatch [14:23:30] sjoerddebruin: My fault, I don't why, though [14:23:30] I ran a dispatch by hand and that seems to make things go bad [14:23:53] Hmm [14:27:14] Seems to be working again [14:27:31] You gave me a heart attack :P [14:33:17] Leaving again… the dispatchers are more fragile when they used to be :S [14:34:22] :( [14:38:00] gah [14:38:28] as if having to purge most pages wasn't enough, I'm having to refresh a load of pages multiple times to get the coordinates to be linked [14:41:22] RECOVERY - wikidata.org dispatch lag is higher than 300s on www.wikidata.org is OK: HTTP OK: HTTP/1.1 200 OK - 1951 bytes in 0.144 second response time [14:41:31] it's been nearly two years now since pages stopped rendering reliably [14:41:54] and I've given up on adding stuff to the ticket pointing out that people still keep complaining 'cause what's the point [15:04:08] nikki: Awesome! :-D [15:04:19] WikidataFacts: Any idea why collection is giving a warning at https://www.wikidata.org/wiki/Q27022520 ? [15:05:10] https://www.wikidata.org/wiki/Talk:Q3305213 looks ok [15:06:04] multichill: no idea [15:06:18] damn, I thought the type check would work better now… [15:06:50] https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations/P195#Types_statistics_3 does show painting as green (first entry) [15:19:07] WikidataFacts: Added https://www.wikidata.org/w/index.php?title=Property%3AP195&type=revision&diff=524608438&oldid=523483395 as a work-around [16:10:49] does anyone know what the time limit on Quarry queries is? [16:11:38] it was half an hour, last I checked [16:11:48] jeez [16:11:50] okay, thanks [16:12:03] I need to approach that with a different mindset, then [16:12:59] what're you trying to do? [16:13:27] most-used descriptions, using wb_terms (see thread: https://twitter.com/belett/status/888788937132703744) [16:13:49] ah [16:13:51] I usually rely on the WDQS timeout to kill expensive queries… looks like this will take a while [16:14:05] but at least I’ll be able to save the results when it’s done, I think [16:14:12] so it should not be necessary to run it again any time soon [16:14:22] the top ones are basically disambiguation, category, things like that [16:14:43] might've changed a little bit since I last checked but doubt it will have changed much [16:14:52] yeah, sjoerddebruin suggested Wikimedia category, with four million items [16:15:05] Wikimedia disambiguation page was some 800k or something like that [16:15:26] I bet there isn't a higher one. https://www.wikidata.org/wiki/Wikidata:Database_reports/Popular_items [16:15:40] https://quarry.wmflabs.org/query/2118 that one's quite old [16:16:10] https://quarry.wmflabs.org/query/6508 a little bit older [16:16:13] oh, SELECT CURRENT_DATE is a great idea [16:16:20] yeah [16:16:33] looks like I have much to learn about Quarry [16:21:20] https://quarry.wmflabs.org/query/9313 also japanese but that's probably less interesting if you can't read it :P (starts with: category, disambiguation, disambiguation, template, surname, template) [16:22:11] oh, the version I’m running just doesn’t filter for the language :D [16:23:10] if it gets killed, let me know and I can try doing it from a dump [16:23:35] I'm sure I have a script already somewhere, and if not, well, I definitely something that will be easy enough to adapt [16:23:42] +have [16:23:53] okay, thanks [16:23:56] let’s see what happens [16:33:18] okay, killed after 30 minutes [16:33:49] boo. suspected that would happen [16:34:59] which is why I already dug something out :P my most recent dump is 2017-06-26, I'm simultaneously extracting the descriptions from that and downloading a more recent one [16:46:32] I just found this query in my old browser tabs: http://tinyurl.com/ycqn5d4c [16:46:43] I think I’ll keep it open in case it ever becomes topical again :) [16:49:05] haha [16:49:15] I try to bookmark things instead of keeping them open [16:49:25] doesn't seem to help, I still always have tons of tabs... [20:10:00] WikidataFacts: partial results :P https://pastebin.com/raw/g7JBujND en/fr/de/nl/ja descriptions used more than 1000 times. still thinking of a better way to do *all* languages, since we have 18 gb of descriptions [20:10:40] surprised that nl is top [20:11:00] nikki: thanks, I just started my own dump download :D [20:11:06] I think I forgot fr in there [20:12:31] hm, on WDQS en slightly beats nl [20:12:49] though the difference is small enough that it might just be hitting different servers which are slightly out of sync, not sure [20:13:30] it took a lot longer than expected, since I ran out of disk space halfway through XD [20:14:36] I’m satisfied that apart from Wikimedia stuff, my guess of “scientific article” apparently really is the most common description :) [20:14:51] though “species of insect” comes very close [20:17:47] I also like all the gene descriptions – you can tell that Mus musculus, Homo sapiens, Rattus norvegicus and Macaca nemestrina are all well-studied species :D [20:20:19] heh yeah [20:22:28] some of the dutch ones have two spaces [20:22:44] oh yeah [20:22:48] the Geographic ones, it looks like [20:23:00] eww [20:23:01] but not all of them [20:23:19] basisschool in Japan (two) but spoorwegstation in Japan (one) [20:23:37] wonder whose fault that is [20:26:04] nikki: should I tweet that pastebin link? or do you want to do another version with fr? [20:26:34] wow, there’s 10× as many nl ”single” as en “single” [20:37:11] Hate merging and splitting items, just to know that there are many more work needed... https://www.wikidata.org/wiki/Special:Contributions/Josve05a [20:37:59] I know the feels. [20:42:41] Luckily there are maintenance lists to fin these https://commons.wikimedia.org/wiki/Category:Pages_with_biology_property_incorrect_on_Wikidata [20:48:45] WikidataFacts: oh, sorry, I wandered off to get something to eat. I can do one with french included in a minute [20:49:42] cool [20:52:41] Well, with some good luck average dispatch will be at 15s again tomorrow [20:53:50] https://pastebin.com/raw/0mBBSwDE [20:54:18] nikki: nice, thanks! [20:55:06] looks like the French speaking community can’t decide between «de Wikimedia» and «d'un project Wikimedia» [20:55:30] the sume of the two is almost the same as the nl,en,de version [20:55:40] *projet, sorry [20:57:56] * pintoch clearly prefers « d'un projet Wikimedia » [21:15:05] Can someone plese make https://www.wikidata.org/wiki/Property:P627 an eternal ID/identifier? [21:15:07] external* [22:42:07] soooo, i'm suddenly unable to edit statements. The "add" link is gone. Anyone know why (like if something recently changed that could've broken a script/gadget)? [22:42:28] purge the pages (add ?action=purge to the url) [22:42:48] https://phabricator.wikimedia.org/T170668 [22:43:47] WikidataFacts, thanks, that fixed it [22:44:00] just noticed now it was only one page, and that did the trick :) [22:44:23] well it seems to happen semi-randomly, some people are getting it on a lot of pages apparently [22:44:53] strange [22:44:54] “fix should probably [be] backported next week” [22:45:05] (Y) [23:21:47] This is tedious. https://www.wikidata.org/wiki/Special:Contributions/Josve05a (almost) all are an items of wikimedia categories with a sielink to ptwiki (and commonswiki), with another item with svwiki, and a sitelink on viwki not connected to any item. All merged and added to one category. [23:22:12] Why are there so many duplicate items for categories :/ [23:22:30] one item*