[06:25:44] hoo, multichill: thanks for the confirmation :-) [06:52:57] Is there an automated way to archive discussions on a talk page ? https://www.wikidata.org/wiki/Wikidata_talk:WikiProject_Informatics/FLOSS is growing and it would be nice if topics with no activity are archived after a while. I think to remember there is a bot doing this but I'm not sure. [09:51:00] Adding references manually reminded me how bad our suggestions are. [09:53:20] sjoerddebruin: +1 [09:54:41] The different sorting per property also makes me sad. Sometimes "stated in" is at the top, sometimes at the bottom. [12:00:57] DanielK_WMDE: when / if you have time, I would like to ask you a few questions about bot integration testing. [13:14:30] @ChanServ ... [13:34:09] 20M -> 23.8M just now! [13:34:57] what, items? [13:35:09] ? [13:35:09] anyone able to assist with petscan additions? I am looking to do item creations with additions, and when I hit the process commands, it sits there with the big red stop button, and does nothing [13:35:11] https://phabricator.wikimedia.org/T144687 [13:35:13] whoa, [13:35:53] that's cheating :p [13:36:19] No, the former displayed number was misleading. [13:37:21] finally correct [13:37:37] :) [13:39:54] sDrewthedoff: for some time might help [13:40:19] testing with one, and ten minutes later I am still waiting [13:41:17] DanielK_WMDE: I have faced a strange situation for this patch: https://gerrit.wikimedia.org/r/#/c/311483. The tests work in my localhost but they fail in Jenkins. Any ideas how to debug such cases? [13:41:33] how does it look like? or do you have a link to a prefilled form? [13:43:13] matej_suchanek: https://petscan.wmflabs.org/?psid=283919 [13:43:55] and I was just trying to run the first [13:44:07] that gives some results... where's the problem? [13:44:45] hmm, it isn't for me; let me try the other PC [13:45:14] 24 [13:46:16] Ohh, I can get the results, it is the addition/creation to WD that is failing for me [13:47:14] failing like you can't see it? [13:47:32] or it doesn't work? [13:48:02] the green process button turns red, says running (3 of 3 left) and nothing else happens, no action, no creation [13:48:44] strange, indeed [13:49:09] that sounds like "ask Magnus" [13:49:11] I'm not testing as it would start adding data [13:49:21] happy for you to test one [13:49:48] sDrewthedoff: what are you trying to add? [13:49:56] same problem as you [13:50:04] Petscan only accepts combination of Property and Item [13:50:18] instance : biographical article && published in (name of the work) [13:50:21] there must be a problem when creating a new item... [13:50:45] Oh, new items. Never did that with petscan before [13:51:17] instructions that I am following at https://meta.wikimedia.org/wiki/PetScan/en [13:51:43] "enwikisourcewiki" is the valid sitename? [13:51:47] sjoerddebruin: I know not how else to create new items except for manual, and have thousands of those [13:52:09] yes, sjoerddebruin they are all ....wiki [13:52:21] shouldn't that be just "enwikisource"? [13:52:53] let me try [13:52:58] that also works and I belive now it should work [13:53:00] https://dumps.wikimedia.org/enwikisource/latest/, https://www.wikidata.org/wiki/User:Pasleim/projectmerge/enwiki-enwikisource [13:53:28] [[Q15156406#P1800]] [13:56:15] hmm, not for me [13:56:40] Q26945047 created [14:00:19] okay, now it works for me, I needed to have as enwikisource [14:00:28] after your change [14:00:45] thanks for your help [14:00:56] you are welcome [14:01:18] matej_suchanek: if it was your change to the aliases, we may need to do that for lots of wikis [14:01:26] wikisources [14:01:27] those "-wiki" suffixes are wrong except for Wikipedias and Magnus should know that... [14:01:50] it appears that way in some databases [14:01:51] in fact I did nothing [14:02:01] hmm, okay [14:02:10] PetScan just allowed that wrong input [14:02:19] and worked with it as if it was correct [14:02:25] which is sort of wrong [14:02:33] okay [14:06:37] hmm, is there a better tool to utilise to create the names and descriptions first? [14:06:52] QuickStatements [14:07:45] but it needs very precise control [14:07:45] I find quickstatements ##headthunk## territory [14:20:56] Wonder when we can expect surnames between the property suggestions. :( [14:27:24] yeah, I miss them as well [14:27:39] may I do some computations? [14:29:03] Of course. [14:32:20] sjoerddebruin: well, we have 9900 disambig pages/family name to clean before we can start any automatize campaign to add family names [14:32:29] harmonia: yeah, I saw that :( [14:32:59] if you want to help, these 24 are bothering me today http://tinyurl.com/hktes7o [14:33:24] disambig pages + family name + a third P31 [14:33:35] Oh dear [14:33:39] if you like to split all that up... [14:33:48] Not feeling that today, sorry. [14:34:02] (why am i not surprised? :p) [14:34:50] sjoerddebruin: actually, I'm doing a mostly good work, but people keep adding wrong given names [14:35:07] so i'm still at ~200 wrong given names [14:35:09] I know the feeling, trust me. [14:35:09] grr [14:35:24] sjoerddebruin: with gender, it should be very soon [14:35:31] matej_suchanek: <3 [14:35:35] * harmonia wants to eliminate the violations constraints [14:35:53] Maybe request some abuse filter, harmonia? [14:35:56] matej_suchanek: how many are missing? [14:36:11] wait... [14:36:12] sjoerddebruin: hmm, what sort? [14:36:29] btw it should already happen with people [14:36:32] sjoerddebruin: people who add "name" items instead of "given name" ? [14:36:42] that triggers when people add disambiguation pages as value for example [14:36:43] harmonia: would you see that it is simply removing the disambiguation statements? otherwise we may have a scenario where some WPs call them disambig, and others SMITH (surname) [14:36:43] matej_suchanek: it doesn't [14:37:03] :( [14:37:11] sDrewthedoff: no it isn't simply retire disambig [14:37:32] sDrewthedoff: we need to see if it's about a family name, about a disambiguation page, about something else [14:37:42] and move the sitelinks, if needed [14:38:02] Josve05a: Try again the kian game. I'm guessing it's much better now [14:38:37] * sjoerddebruin still wants a better birth date / death date game [14:38:41] sDrewthedoff: these query give us items with at least 3 P31, so only difficult cases [14:38:48] Maybe some game for places too [14:38:53] for soemthing like this https://www.wikidata.org/wiki/Q13554167 WTF knows? [14:39:57] https://quarry.wmflabs.org/query/12648 [14:39:58] 1) who can tell if disambig or not, 2) the family name and Chinese surname are both correct [14:40:23] sDrewthedoff: we keep the most precise P31 when there are both [14:40:31] and when is a subclass of the other [14:40:40] when one* [14:40:46] matej_suchanek: the entity suggester became a whole clusterfuck for me [14:40:59] how so? [14:41:02] so no problem at all with "Chinese surname/family name" [14:41:19] Hey [14:41:20] but we need to know if it is a disambig or not [14:41:21] It was much more predictable before the identifier conversion [14:41:48] why? [14:42:05] datatypes have never been handled specially... [14:42:24] For my feeling we've had better suggestions back then. [14:42:32] But could also be the huge growth of our data. [14:43:20] of course, our data grew up and the suggestions are now less obvious... up to nonsense [14:43:24] i think it's the latter [14:44:09] at the workshop I attended I was explaining quite well how it works [14:44:19] which means it's a really simple algorithm [14:44:58] ah? [14:45:03] We remove certain properties from the calculations now btw. [14:45:09] yep [14:45:27] matej_suchanek: so why don't we have family name in the suggestions for a human? [14:46:14] I will investigate, just need to find a person without it [14:46:44] https://www.wikidata.org/wiki/Q16115994 [14:46:47] https://www.wikidata.org/wiki/Q26256902 [14:46:53] easyyyyy [14:46:57] :p [14:47:14] I was typing into Query Service btw [14:47:25] but didn't recall triples for number of statements [14:49:05] it's there at https://www.wikidata.org/wiki/Q776805 [14:49:17] the most bottom one [14:49:38] ah yes [14:49:42] O_o [14:49:51] So another property must have a very negative influence on the suggestions [14:50:24] as it could be caused by removing thos ext-id ones which have the positive influence [14:52:29] I must say, https://www.wikidata.org/wiki/Property:P166 mostly doesn't show up when a item has a ext-id. [14:53:36] even date of death holds suggestion of surnanme [14:54:29] lets see if I can find a item that probably has a ext-id but not yet [14:55:28] https://www.wikidata.org/wiki/Q26922479 [14:56:05] it's there, isn't it? [14:56:51] I'm going to add a lot of statements, but not the RKD ID. [14:59:20] Well, it's gone now. [15:00:02] isn't there anything more removed together with ext-ids? [15:01:51] but still, why are not the other positive suggestions kept? when there is a match coming from non-identifier property, that one should apply... [15:02:17] Ehm: sport, image, commonscat. [15:02:44] also "member of" and "participant of" [15:02:51] According to https://phabricator.wikimedia.org/T132839 [15:02:55] harmonia: "these 24 are bothering me today http://tinyurl.com/hktes7o ..." {{done}} [15:03:07] sDrewthedoff: thank youuuuuuu [15:03:21] :) [15:03:33] some were ugly, and probably needed a proper cleanup, defaulted to disambig [15:03:34] matej_suchanek: I think "official website" is to blame in this situation, as it is used in various subjects. [15:03:51] I thought that as well actually [15:04:20] But then we end up with removing half of the properties from the calculations. [15:04:38] pain in the arse that Petscan cannot give me a list usable as a base for quickstatements (well not that I can see) [15:05:03] I've added the RKD ID now and "sport" disappeared from the suggestions. [15:05:10] \o/ [15:05:12] sDrewthedoff: did you check the linked pages too? [15:05:31] sDrewthedoff: before defaulting to disambig, were they used as surnames? [15:05:31] sure [15:05:40] sDrewthedoff: you are really great [15:05:43] thank you [15:05:52] wouldn't that mean that there's a "negative feature" to hide matches? that would be new to me... [15:06:09] I didn't split them all, but if some labelled ... (disambig) that was enough even wehn otehrs were soemwhat surname [15:06:15] maybe I know only a half of that magic [15:06:21] You should talk with hoo, matej_suchanek. [15:07:07] sDrewthedoff: well, we'll [15:07:09] see [15:07:25] if the surnames are needed we'll have to create them one day [15:07:40] not stressing about it before :) [15:10:42] dachary: you are adding sources as qualifiers with FLOSSbot [15:11:03] like https://www.wikidata.org/w/index.php?title=Q24448646&type=revision&diff=377673512&oldid=372824266 [15:11:34] we need a good splitting tool [15:17:05] https://www.wikidata.org/wiki/Property:P1412 shows up though, so we need a lot more statements then :P [15:18:17] Always nice to leave a item with 30k more bytes. [15:23:42] I've added this to my list: https://www.wikidata.org/w/index.php?title=User%3ASjoerddebruin%2FEntity_suggester&type=revision&diff=377676954&oldid=376463414 [15:30:31] sjoerddebruin: <3 [15:30:53] worth creating a project in Phab maybe? [15:36:34] ^^ [15:37:39] petscan defeats me; it needs a really good manual [15:38:08] aagh, one of the example https://petscan.wmflabs.org/?psid=148910 [15:38:18] comes back with 502 Bad Gateway [15:38:20] I think there's too much criteria so that one gets lost [15:38:39] there could something like "light" version [15:39:00] sDrewthedoff: if it comes with an error, I cannot debug [15:39:25] just "bad gateway" so presumably a labs issue [15:39:38] not only [15:39:41] giving up and going to bed, need to be up in 4 hours [15:39:42] i think there were issues with labs today [15:39:52] was it? [15:39:54] * aude not sure and doesn't know details [15:40:19] the webservice might need to be restarted or something [15:40:30] sDrewthedoff: "bad gateway" is often a timeout [16:00:38] hello [17:02:38] PROBLEM - Response time of WDQS on wdqs1001 is CRITICAL: CRITICAL: 12.50% of data above the critical threshold [300000.0] [17:07:39] RECOVERY - Response time of WDQS on wdqs1001 is OK: OK: Less than 5.00% above the threshold [120000.0] [17:27:13] matej_suchanek: I may be creating a tracking task soon. Need to discuss first. [19:18:54] DanielK_WMDE: Hey, do you have a minute to check this on your localhost? https://gerrit.wikimedia.org/r/#/c/311483/ [19:19:19] All tests pass for me in my localhost but jenkins complain and I don't know why. [19:19:36] Also any tip on how to debug these cases would be awesome :) [19:21:24] oops you just posted there :D [19:32:59] Amir1: i'm about to leave to office. i'll have a quick look, but testing will have to wait until tomorrow [19:33:13] Thanks DanielK_WMDE [19:33:44] maybe it has something to do with items being in the main namespace (or not)? [19:34:00] i saw you use Title::getText, that doesn't include the namespace... but maybe somewhere... [19:34:35] debugging something that passes locally but fails on jenkins is always annoying. [19:35:09] you can ask someone else to try, or you can add echo() output... not nice :/ [19:35:25] anyway, time to go home. [19:35:26] cu! [20:08:48] What is the right way of querying the current stable version of Firefox from Wikidata (ideally through wikitext, but an API query works too, this isn't for an on-wiki use case) [20:08:49] https://www.wikidata.org/wiki/Q698 [20:09:25] It seems there are several 'software version' entries. One for each release. Some of which are tagged with LTS, Beta, Stable etc. but the labels are not exclusive to a single point. Old releases can be marked stable if they were stable at the time. [20:10:25] Krinkle: usually, the most current version has the Preferred rank, and other statements have the Normal rank [20:10:51] in this case, there’s two statements with the Preferred rank, 49.0 (stable) and 45.3.0 ESR (extended support) [20:11:03] so you’d have to look for preferred rank and qualifier: stable version [20:11:09] Hm.. So it's not an exclusive mark to be preferred. [20:11:20] I see. How would I put that in a query? [20:12:26] http://tinyurl.com/jdprvyn [20:14:36] Ah, using the query service. Nice [20:14:53] I was looking for a way to use the MediaWiki API (api.php) but I guess this is more advanced. [20:14:56] Thanks. [20:15:42] not sure if it’s possible using the API… you’re welcome! [20:21:11] WikidataFacts: Just GET https://www.wikidata.org/wiki/Special:EntityData/Q698.json and retrieve the content of entities.Q698.claims.P348 [20:21:38] (you could also filter with rank: "preferred" in order to retrieve only preferred statements [20:23:04] or more simple: https://www.wikidata.org/w/api.php?action=wbgetclaims&entity=Q698&property=P348&rank=preferred [20:29:01] WikidataFacts: thanks. This helped https://github.com/WordPress/browsehappy/issues/37 [20:29:04] Tpt: or less simple: [20:29:04] curl https://www.wikidata.org/wiki/Special:EntityData/Q698.json | jq '.entities.Q698.claims.P348 | .[] | select(.rank == "preferred") | select(.qualifiers.P548[0].datavalue.value.id == "Q12355314") | .mainsnak.datavalue.value' [20:29:49] (not sure if that’s useful at all, you pointed me at the JSON and I wanted to see if I could get this done with jq :) ) [20:31:22] Krinkle: Tpt’s variant might be better, not sure [20:44:51] Tpt: Interesting. That does send a lot more data though. [20:45:09] yes, that's true [20:45:18] but it's well cached on server side [20:46:21] Tpt: Hm.. good point.