[08:01:16] <nikki>	 abian: done
[09:11:27] <abian>	 Thanks, nikki! :)
[09:55:55] <Thiemo_WMDE>	 leszek_wmde: I would love to pick the Geo update and fix it. Are you working on this?
[09:56:17] <leszek_wmde>	 Thiemo: not really after the patch which is up there
[09:57:01] <leszek_wmde>	 Thiemo_WMDE: I mean the test issue in https://gerrit.wikimedia.org/r/353022 - I have not looked into that
[10:54:36] <multichill>	 nikki: See http://tinyurl.com/m6pc3ut I want to know which paintings have ?painting wdt:P195 ?location (instead of ?collection), but that seems to time out, any smart suggestions? 
[11:50:26] <multichill>	 http://tinyurl.com/mzps88z seems to do it, still needs a bit of tweaking
[17:11:30] <pharos>	 hi Wikidatans
[17:19:08] <pharos>	 can someone tell me what I'm doing wrong here https://en.wikipedia.org/wiki/Template:Mbabel/MET/ref ?
[20:25:29] <halfak>	 Hey folks.  I want to query wbs_propertypairs from api.php.  Is there a way to do that?
[20:38:12] <halfak>	 Essentially, I'm hoping to give it an item and have it respond with a pid2 and probability fields that correspond to recommended properties (whether or not they exist in the item)
[20:42:47] <multichil>	 halfak: https://www.wikidata.org/w/api.php?action=wbsgetsuggestions&search=&context=item&format=json&language=en&continue=0&entity=Q29653349 seems to be the API call
[20:42:58] <halfak>	 glorian_wd, ^ 
[20:43:03] <halfak>	 Thanks
[20:43:05] <halfak>	 looking at that
[20:43:11] <multichil>	 No documentation
[20:43:28] <multichil>	 https://www.wikidata.org/w/api.php?action=help&recursivesubmodules=1#wbsgetsuggestions
[20:43:51] <glorian_wd>	 thanks halfak & multichil :)
[20:44:40] <halfak>	 glorian_wd, looks like "rating" will work for us. 
[20:45:20] <multichil>	 You found the full table on toollabs halfak?
[20:45:34] <halfak>	 multichil, yeah, but I'd like to avoid using that if possible. 
[20:45:42] <halfak>	 We'll be accessing this from ORES so I want to use the api
[20:45:49] <multichil>	 Might be useful if you want to use it to train something
[20:45:54] <halfak>	 multichil, do you know how the properties param works here?
[20:46:04] <halfak>	 Oh!  I see
[20:46:09] <halfak>	 Property-properties
[20:46:31] <halfak>	 glorian_wd, I think we should query based on the entity ID even though we're targeting a specific revision. 
[20:46:43] <multichil>	 https://www.wikidata.org/w/api.php?action=wbsgetsuggestions&properties=P21&context=qualifier
[20:46:46] <halfak>	 we should expect instance-of and subclass-of to be slow moving
[20:47:05] <multichil>	 I like this one https://www.wikidata.org/w/api.php?action=wbsgetsuggestions&properties=P217&context=qualifier
[20:47:27] <multichil>	 If you add an inventory number, we want to know in what collection it is (P195) and that's the only suggestions ;-)
[20:47:39] <multichil>	  "rating": "0.9999566674232483",
[20:47:40] <halfak>	 https://www.wikidata.org/w/api.php?action=wbsgetsuggestions&entity=Q42
[20:47:52] <halfak>	 Looks like this is empty. 
[20:47:57] <multichil>	 No suggestions left
[20:48:01] <multichil>	 Take a fresh item
[20:48:10] <halfak>	 Which is OK, but I'd really rather get the top suggestions even if they exist. 
[20:48:31] <halfak>	 I want to be able to tell the difference between "no likely properties" and "no more suggestions"
[20:49:12] <multichil>	 You can ask the api for the number of statements on an item
[20:49:15] <multichil>	 That gives a hint
[20:49:41] <halfak>	 Hmm...  Essentially, we're trying to get signal for "How complete is this item"
[20:49:52] <halfak>	 And working from suggested properties seems like it ought to be effective. 
[20:50:12] <multichil>	 If the number (different) statements is > 15 and no suggestions, it's pretty complete
[20:50:17] <halfak>	 Sometimes we need to make historical predictions. 
[20:50:38] <halfak>	 So I might ask "When did Douglas Adams(Q42) reach a high level of completeness?"
[20:50:46] <halfak>	 This won't work for that. :/
[20:50:56] <halfak>	 Because we don't know which properties were important. 
[20:51:21] <halfak>	 I think it might be time to file a bug :) 
[20:51:25] <halfak>	 *feature request
[20:51:32] <halfak>	 glorian_wd, do you follow?
[20:51:54] <multichil>	 Constraint violations are generally also a good indication of completeness
[20:51:59] <glorian_wd>	 halfak: hey, reading it
[20:52:06] <multichil>	 Missing P<something> violations that is
[20:52:37] <halfak>	 multichil, can we query for those. (preferrably historically)
[20:52:44] <halfak>	 Missed a "?" in there
[20:54:03] <multichil>	 halfak: Reports are wiki pages with a history?
[20:54:29] <halfak>	 Top recommended property for my item is P570 "Date of Death" -- Wikidata wants me to die. 
[20:54:31] <multichil>	 And I think we have a different constraint reports system too, but I haven't really played aroudn with it
[20:54:36] <halfak>	 multichil, damn. 
[20:55:11] <halfak>	 Would you ever add P570 with a null value for living people?
[20:55:13] <multichil>	 For example https://www.wikidata.org/wiki/Property_talk:P650 / https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations/P650
[20:55:31] <multichil>	 And especially for you: https://www.wikidata.org/wiki/Wikidata:Database_reports/Complex_constraint_violations/P650#Probably_death
[20:57:00] <glorian_wd>	 halfak: hmm, I guess that we still need revid
[20:57:20] <glorian_wd>	 Because, there might be a change after the item has been labeled
[20:57:43] <halfak>	 right.  This is a problem for us.  But it wouldn't be a problem if we could get all relevant property weights -- not just the missing ones. 
[21:02:34] <Amir1>	 I'm heading bed
[21:02:38] <Amir1>	 https://phabricator.wikimedia.org/T164984
[21:02:42] <Amir1>	 halfak: FYI :(
[21:03:10] <Amir1>	 o/
[21:03:13] <halfak>	 o/
[21:06:20] <yurik>	 SMalyshev, is it possible for runUpdate.sh  to keep showing identical items? It seems like it got stuck in a loop
[21:06:40] <SMalyshev>	 yurik: hmm shouldn't happen
[21:06:46] <SMalyshev>	 that may be a bug
[21:06:54] <glorian_wd>	 halfak: so, I guess we would opt using wbgetsuggestions right?
[21:07:03] <SMalyshev>	 what exactly it says?
[21:07:10] <glorian_wd>	 but as you said, we will use entity-id instead of revid
[21:07:18] <halfak>	 glorian_wd, I don't think it'll be sufficient
[21:07:43] <halfak>	 We'll need to be able to either get all recommendations whether they exist or not or we'll need to be able to ask for recommendations historically. 
[21:07:59] <halfak>	 So, I'd like to have you file a task and look into what it would take to implement that. 
[21:08:07] <yurik>	 SMalyshev, https://gist.github.com/nyurik/966ff6469163ebf0d9dcb51a13c0c3a2
[21:08:45] <SMalyshev>	 yurik: hmmm definitely not good. Are you using the latest checkout?
[21:08:52] <yurik>	 SMalyshev, yep
[21:09:10] <yurik>	 SMalyshev, well, i downloaded the service-0.2.3-dist.zip
[21:09:11] <SMalyshev>	 hmm weird why it does that. Try restarting it
[21:09:18] <SMalyshev>	 0.2.3 is ancient
[21:09:34] <SMalyshev>	 current is 0.2.4 :)
[21:10:20] <SMalyshev>	 yurik: you can take binaries from https://github.com/wikimedia/wikidata-query-deploy
[21:10:31] <yurik>	 SMalyshev, i followed the instructions, and downloaded it from maven
[21:10:31] <yurik>	 https://www.mediawiki.org/wiki/Wikidata_query_service/User_Manual#Standalone_service
[21:11:03] <SMalyshev>	 yurik: take this one https://github.com/wikimedia/wikidata-query-deploy/blob/master/lib/wikidata-query-tools-0.2.4-SNAPSHOT-jar-with-dependencies.jar don't forget git fat pull
[21:11:08] <SMalyshev>	 (I think git fat should work for you, right?)
[21:11:43] <SMalyshev>	 yurik: yeah I haven't updated maven one for a while. Probably need to
[21:11:54] <yurik>	 yeah, or change the docs :)
[21:11:55] <SMalyshev>	 yurik: if you prefer, I can put a zip on archiva.
[21:12:05] <yurik>	 sure, might be easier :)
[21:12:27] <yurik>	 does it mean i have to rebuild everything?
[21:12:32] <yurik>	 i mean - re-import wikidata/
[21:12:34] <SMalyshev>	 yurik: no
[21:12:39] <SMalyshev>	 yurik: also no
[21:13:07] <SMalyshev>	 yurik: replacing the jar is painless, nothing needs to be done except restarting updater
[21:13:17] <SMalyshev>	 if you want a zip I can make one
[21:13:39] <SMalyshev>	 yurik: actually, given the bugfixes probably not a bad idea to release 0.2.4... maybe I'll do that later
[21:14:52] <yurik>	 thanks, that would be good
[21:15:18] <SMalyshev>	 yurik: in the meantime, try using deploy repo and git fat. if that doesn't work tell me I'll deploy a snapshot
[21:18:35] * yurik setting up the fat
[21:33:40] <yurik>	 SMalyshev, i cloned the deploy repo, and did git fat init + pull, and i now get this error when starting blaze graph:  https://gist.github.com/nyurik/966ff6469163ebf0d9dcb51a13c0c3a2#gistcomment-2090188
[21:52:13] <SMalyshev>	 yurik: hmm. You didn't have to replace blazegraph, only updater
[21:52:30] <SMalyshev>	 yurik: but anyway, this looks like you're running old Java. it needs java 8
[21:52:59] <yurik>	 SMalyshev, openjdk version "9-internal"
[21:53:11] <yurik>	 OpenJDK Runtime Environment (build 9-internal+0-2016-04-14-195246.buildd.src)
[21:53:21] <SMalyshev>	 yurik: hmmm no idea what is that... looks like it has incompatible options
[21:54:06] <SMalyshev>	 you can edit runBlazegraph.sh and remove the GC options, but I don't know why they don't work... they work in standard java 8
[21:56:07] <yurik>	 SMalyshev, the two packages related:  openjdk-9-jre-headless/xenial,xenial,now 9~b114-0ubuntu1 amd64
[21:56:19] <yurik>	 and java-common/xenial,xenial,xenial,xenial,now 0.56ubuntu2 all 
[21:56:44] <SMalyshev>	 hmm no idea what they changed in java 9
[22:10:49] * DanielK_WMDE_ remembers when Java 2 whas The New Shit
[22:10:54] * DanielK_WMDE_ feels old
[22:12:05] <SMalyshev>	 yeah I remember one of my first CS courses was in that new hip language - Java, everybody was excited
[22:24:05] <yurik>	 DanielK_WMDE_, yep, good ol days, when msjv was much cooler than java ;)