[09:32:33] JeroenDeDauw: have you ever tried creating that bottom visualization for mediawiki or wikibase? [09:57:12] Is https://tools.wmflabs.org/wikidata-todo/multibeacon.php supposed to work? [10:17:06] Now I have doubts on where I'm supposed to (not) use the Q/P prefix [10:57:14] srsly https://www.wikidata.org/wiki/Q26932615 [10:57:47] yup [11:31:15] addshore: I suspect I did for wikibase at some point.... [11:31:24] and that it made my browser hang for a bit [11:31:33] let me try again [11:32:04] haha, my Wikibase clone is 3 months out of date [11:38:19] addshore: http://pasteboard.co/67CNO7M3p.png wikibase [11:39:21] ooooh [11:59:16] DanielK_WMDE: https://commons.wikimedia.org/wiki/Commons:Threshold_of_originality [12:45:20] https://www.wikidata.org/wiki/Wikidata_talk:SPARQL_query_service/queries#Select_qualifier_string_content [13:19:45] so, https://www.wikidata.org/w/index.php?title=Q4115189&action=history some of the revisions have a nice summary and some don't. [13:19:53] is this in the wikidata UI code? [13:19:55] or somewhere else? [13:20:06] the ones that weren't were generated by a pywikibot test script [13:21:07] * YuviPanda tries to catch addshore's eye [13:25:53] YuviPanda: most of those are auto-generated summaries [13:26:13] aude: autogenerated where? [13:26:59] in wikibase + sometimes a script, bot or tool appends extra summary [13:27:08] "Created claim: population (P1082): 2,347±0" would be autogenerated [13:27:14] then #quickstatements [13:27:17] from the tool [13:27:26] right [13:27:30] some summaries like "Updated item" aren't so great [13:27:30] so this is something in pywikibot [13:27:43] pywikibot can add something [13:27:54] ( amisha also found this because hitting the API directly creates them properly ) [13:27:59] aude: it seems to be 'resetting' them [13:28:12] https://www.wikidata.org/w/api.php?action=help&modules=wbeditentity [13:28:34] has summary param which i think appends the summary, though not sure ... maybe it replaces the summary [13:29:01] amisha: in your API call are you setting 'summary'? [13:29:12] aude, yep, there is option to override default summary i.e. updated item [13:29:16] I think Amir1 when he is around, is best to ask [13:29:59] YuviPanda, In API call, I din't set summary exclusively [13:30:19] if you don't set it, then it uses an autogenerated summary from wikibase [13:30:20] amisha: can you try setting it to see if that causes the pywikibot behavior? [13:30:25] right [13:30:41] what the summary is depends somewhat on which api module is being used [13:34:15] Right , Earlier I was performing wbsetaliases and wbsetlabel, which appended auto summary [13:34:22] aude, [13:35:14] Amir1: would be the best person to ask but I think he's at a conference now [13:36:18] amisha: can you pastebin the code you're using with pywikibot? [13:37:04] yep. [13:39:29] YuviPanda, https://pastebin.mozilla.org/8949676 [13:43:06] Thiemo_WMDE: is https://gerrit.wikimedia.org/r/#/c/323556/8 okay? [13:43:30] if so, we can get this in swat [13:45:37] so pywikibot is using wbeditentity [13:45:43] and not wbsetlabel [13:46:02] aude: I just wrote a comment there, and added a +1. That's all I can do. I don't have +2 rights there. [13:46:18] ok [13:46:22] thanks [13:47:09] amisha: so pywikibot is using wbeditentity and not wbsetlabel / wbsetalias [13:47:28] amisha: can you try calling wbeditentity directly to see if that also gives you bogus autosummaries? [13:47:35] aude: This part of the migration is now much more obvious than before: all it does is moving the setting out of our default settings. Thanks for splitting the patches so well. I would +2 this now if I could. [13:49:31] Thiemo_WMDE: i'll put it in the next swat [13:49:41] just want to get this moving [13:52:04] amisha: if it does, then we can file a bug against wikidata :D if not, we can file it against pywikibot. [13:52:37] amisha: if the code you showed me is all that you need, you can probably just call the wikidata API manually... [13:53:35] YuviPanda, lol, true. But just to avoid validation and stuff, I considered using pywikibot [13:55:07] amisha: yeah, I agree tho. but most validation seems to be done serverside - I looked trough the pywikibot code that you're calling (editLabels) and it's a super thin wrapper around the API [13:58:58] hi YuviPanda ! [13:59:45] hi addshore [14:05:42] YuviPanda, Or for now, We can send some custom summaries using Pywikibot as well [14:05:58] amisha: you could, but they won't be as nice tho. [14:06:14] you can try adding 'summary=None' to the editLabels call [14:06:16] to see if that helps [14:06:41] * amisha is trying [14:09:34] sigh... what's the point of having novalue to mean no value when certain people refuse to use it and instead create special items to mean "novalue" [14:09:40] YuviPanda, That doesn't help. Still saying updated Items [14:10:52] amisha: ouch, ok. I think it's time to file a bug against pywikibot now [14:11:16] to me it seems completely wrong to do that, but nobody else ever seems to comment on it... am I missing something? :/ [14:13:34] like on https://www.wikidata.org/wiki/Property:P1086 apparently the right item for units of atomic number is Q21027105 ("quantity property without units") but that's not true, the right item is *no* item [14:14:01] YuviPanda, Talking to Amir1 [14:14:48] amisha: nice [14:17:45] https://www.wikidata.org/w/index.php?title=Q20471851&type=revision&diff=412941009&oldid=224708943 ... [16:10:22] hi - Quick question. About a week ago I started getting 429 code responses (too many requests) on the SPARQL endpoint. Did a policy change? what is considered too many? [16:11:01] I am doing requests in parallel, but not that many, and it used to work. I'm just wondering if something changed. [17:05:43] Tobi_WMDE_SW_NA: tut jetzt! [17:05:51] jep! danke! [17:05:56] DanielK_WMDE: [17:06:08] Tobi_WMDE_SW_NA: unser Mails laufen ja alle noch über den MX bei Domainfactory. Man muss jede adresse doppelt anlegen. [17:06:19] DanielK_WMDE: o_O [17:22:50] SMalyshev: see Eric's comment above [17:23:07] also gehel [17:23:52] Eric: yes there was a change but I don't know the details. SMalyshev and gehel will know more [17:24:10] Eric: yep, we started limiting parallel queries. The current limit is at 5 concurrent requests per server (which mean that the actual limit is between 5 and 10 per user) [17:25:42] Eric: we're still experimenting with the actual limit and the impact it has, so your feedback is more than welcomed! (And feel free to tell us we are completely wrong with those limits) [17:26:14] Lydia_WMDE: thanks for the ping, I don't monitor this channel as much as I probably should... [17:26:41] gehel: don't worry :) [17:26:58] information overload is a killer... [17:27:03] it is [17:36:44] Lydia_WMDE: ich hab noch zwei dinge für morgen... T119536 (bzw T150290) und T128667+T128486 [17:36:45] T128486: [Story] Purge Special:EntityData JSON after edit - https://phabricator.wikimedia.org/T128486 [17:36:45] T150290: add CORS to all redirecs in chain from https://www.wikidata.org/entity/{Q...} - https://phabricator.wikimedia.org/T150290 [17:36:45] T119536: [RFC] should wikidata.org/entity/Q12345 do content negotiation, instead of redirecting to wikidata.org/wiki/Special:EntityData/Q36661 first? - https://phabricator.wikimedia.org/T119536 [17:36:45] T128667: Special:EntityData with flavor is cached but not purged properly - https://phabricator.wikimedia.org/T128667 [17:36:49] huch [17:36:57] Lydia_WMDE: soll ich die einfach mit ins dokument schreiben? [17:37:49] DanielK_WMDE: jep [17:41:52] ok [19:23:54] SMalyshev: Any idea which hash function is fasted in SPARQL? It doesn't need to be secure at all, just be fast. Now I used SHA512, quite secure, not very fast ;-) [19:24:34] multichill: in what context? I'm not sure how hash & sparql come together :) [19:24:42] They do [19:24:55] SMalyshev: http://tinyurl.com/znwcs3j [19:25:17] To randomize a bunch of results [19:26:08] ok, got it [19:26:24] My best guess would be MD5 [19:26:25] I think out of ones I see in the standard, md5 would probably be the fastest [19:26:32] https://www.w3.org/TR/sparql11-query/#func-hash [19:27:12] The time difference is very minimal so that's why I was wondering [19:27:36] yeah since you do it in projection, by the time it gets there most of the work is done [19:28:01] so unless you have thousands of results it doesn't really matter I think [19:28:14] ah, wait, you're sorting by it [19:28:48] SMalyshev: 48.000 in this case [19:29:07] hmm... try to use md5 and see if it makes it faster [19:29:18] I tried that and both are around 10 seconds [19:29:34] but I wonder if there's a better way to get random results... [19:29:43] That makes two of us [19:29:59] I was helping a friend and she already did quite a bit of googling [19:30:42] I wanted something with an even distribution and would be completely different every run so a hash sounded like the best fit [19:31:17] mysql has order by rand but I don't think sparql does [19:31:38] RAND() just returns one float.... [19:32:19] do you need one image? [19:32:58] Need? This is just an example to return 10 random images of paintings [19:33:24] It's used in a similar way at https://www.wikidata.org/wiki/User:Spinster/Paintings_without_depicted_things [19:41:21] SMalyshev: I asked the lazyweb, let's see what comes back ;-) [19:43:41] I added it to the query examples. I'm trying to have at least one example for every weird/edge case I encounter [19:46:15] @Lydia_WMDE @gehel Thanks! I get these errors if I do about 80 queries in parallel. I will see if I can combine them into groups of 10. [19:47:37] You must have a cache, as if I do a set of 80 queries, I might get 20 rejected; then do it again - 10 rejected; then again - everything works [19:48:16] what are you doing that needs 80 queries in parallel? [19:48:36] See http://www.dageshify.com/timeline?query=Category:Ancient_Greek_mathematicians [19:49:03] Retrieving data on birth/death of mathematicians [19:49:10] one query per mathematician [19:49:29] Batch them! [19:49:36] You can do that all in one query [19:49:47] Eric: note that 10 is best possible limit, you might be block at 5 in // [19:49:50] I should be able to combine them, but not all the way into one query, because is a limit on the number of characters you can have in one sparql query [19:50:22] Firing 80 queries on our poor query engine is not cool [19:50:22] excellent - I will batch them [19:50:29] sorry! [19:50:33] So what data do you actually need? [19:50:40] I will batch them [19:50:44] You have 80 people and you need some data for those people? [19:50:46] dates [19:51:01] And you already have the Qid for those people? [19:51:09] yes, primarily dob and dod [19:51:24] the query gets them based on the name of the wikipedia article [19:51:35] one moment, I'll get the query .. [19:51:39] Eric: and yes, we do have caching as well, so the actual limit may vary [19:52:21] http://tinyurl.com/za9kml3 [19:52:59] I get the qid by saying: schema:about ?subject . [19:53:05] in the where clause [19:53:20] What the heck, that's one huge query! [19:54:40] Eric: I would use VALUES for the " schema:about ?subject ." part to get them all in one go [19:55:02] Or if the query gets too big. Do one query to get the qids, second one to get all this stuff [19:56:00] right - that makes sense .. should have thought .. [20:00:41] ok so I see rate limiting is working ;) [20:02:29] SMalyshev: yep. And from the discussion above, it seems that this was a valid reason to limit :P [22:12:30] Lydia_WMDE: would it be better if I made a phabricator ticket for the multiple valid sitelinks thing I keep pestering you about? [22:12:47] nikki: yes. and sorry! :( [22:13:00] ok :) as long as you don't start hiding every time you see me :P [22:13:01] just juggling too many balls [22:13:05] haha [22:13:06] * nikki nods [22:13:06] not [22:13:09] *no [22:21:09] Eric: Did you manage to update your query? [22:56:58] Lydia_WMDE: there, created https://phabricator.wikimedia.org/T152465 and assigned it to you :)