[00:01:38] <hoo>	 broken?
[00:02:15] <hoo>	 nvm, I see
[00:02:37] <jzerebecki>	 hoo: i'm confused, had an item page render nearly empty
[00:02:42] <jzerebecki>	 now its there
[00:02:50] <hoo>	 yeah, same here
[00:03:03] <hoo>	 very weird
[00:03:16] <jzerebecki>	 I see nothing in logstash
[00:04:25] <hoo>	 Probably something screws the parser cache
[00:05:34] <jzerebecki>	 this is also new "Gadget "Primary_sources_tool" was not loaded. Please migrate it to use ResourceLoader.  See <https://test.wikidata.org/wiki/Special:Gadgets>."
[00:06:17] <jzerebecki>	 we bumped the parser chache epoch
[00:08:54] <jzerebecki>	 never mind the primary sources thing is not new
[00:09:46] <hoo>	 can't reproduce anymore
[00:13:19] <jzerebecki>	 hoo: happend here, don't purge https://test.wikidata.org/wiki/Q138 
[00:15:09] <jzerebecki>	 ok 2nd time, edit item with no statement to add is instance of q42, then force refresh 2 times (don't ask)
[00:16:28] <jzerebecki>	 ok on https://test.wikidata.org/wiki/Q449 it didn't happen
[00:17:28] <hoo>	 I'm trying locally... but can't reproduce
[00:17:46] <hoo>	 No matter what state the parser cache is before or how the item was before :S
[00:18:08] <jzerebecki>	 possibly unrelated: the icons on editing statements are not there on test but are on beta
[00:19:11] <hoo>	 They are there for me
[00:21:47] <jzerebecki>	 how? the html for the icon isn't even there, even when not logged in
[00:22:00] <hoo>	 huh?
[00:22:22] <jzerebecki>	 <a href="#">edit</a> vs <a href="#"><span class="wb-icon"></span>edit</a>
[00:22:48] <hoo>	 I keep seeing the second one on testwikidata
[00:30:16] <jzerebecki>	 can't reproduce broken pages on beta
[00:30:29] <hoo>	 I can no longer reproduce it anywhre
[00:38:42] <jzerebecki>	 same here
[00:45:02] <hoo>	 I have a suspicion re the parser cache
[00:48:10] <hoo>	 hm
[00:48:35] <jzerebecki>	 mh?
[00:49:30] <hoo>	 The job runners are not picking up on changes immediately (as they are run for a period of time, to take advantage of the JIT)
[00:49:50] <hoo>	 maybe we had an edit from wmf8 go through a wmf7 job runner
[00:50:08] <hoo>	 And maybe that messed with the parser cache in horrible ways
[00:50:34] <hoo>	 But that's just a theory, I couldn't verify that locally by having the jobs run on wmf7 :/
[00:50:54] <jzerebecki>	 isn't there a scap stage that clears the hhvm cache?
[00:51:29] <hoo>	 Not sure, but I don't think we restart the job runners immediately
[00:51:58] <hoo>	 or did you mean the cache that's also the JIT bytecode
[00:52:40] <jzerebecki>	 was thinking of the later
[00:53:21] <jzerebecki>	 but if each job is not a new php process...
[00:53:29] <jzerebecki>	 version switch was at 23:47
[00:53:31] <hoo>	 it's not
[00:53:47] <jzerebecki>	 <!-- Saved in parser cache with key testwikidatawiki:pcache:idhash:1543-0!*!0!*!en!*!*!wb3 and timestamp 20151209001356 and revision id 23472
[00:54:00] <jzerebecki>	 that timestamp is from https://test.wikidata.org/wiki/Q792
[00:55:12] <hoo>	 that's 22 minutes
[00:56:57] <jzerebecki>	 *sigh* wasn't there a way to link a logstash query?
[00:57:26] <bd808>	 jzerebecki: second icon on the top right
[00:57:38] <bd808>	 box with an arrow poking out of it
[00:58:05] <hoo>	 bd808: Do you have a stalkword on logstash? :D
[00:58:13] <bd808>	 yes
[00:58:15] <bd808>	 and scap
[00:58:21] <bd808>	 and vagrant
[00:58:47] <jzerebecki>	 bd808: I don't see that icon
[00:59:28] <bd808>	 hmm... do you see the little gear at the far right?
[00:59:33] <jzerebecki>	 next to time is reload, home, load, save, ...
[00:59:48] <jzerebecki>	 yes gear right next to save
[01:00:18] <bd808>	 weird. share should be between save and configure
[01:00:52] <jzerebecki>	 yes! its not visible
[01:01:19] <hoo>	 jzerebecki: http://pastebin.com/i1GpjzLb
[01:01:31] <hoo>	 Look at mText (very top)
[01:02:00] <bd808>	 jzerebecki: heh. so stealthy
[01:02:04] <jzerebecki>	 hoo: that https://logstash.wikimedia.org/#dashboard/temp/AVGEQTvJptxhN1XaK61h is the time around that timestamp, not sure it matches
[01:05:25] <jzerebecki>	 ok so why are for me in firefox the save icons missing, but not in chromium
[01:06:06] <hoo>	 jzerebecki: Think you're off by an hour there
[01:06:50] <jzerebecki>	 hu?
[01:07:07] <jzerebecki>	 shouldn't both logstash and that timestamp be utc?
[01:07:17] <jzerebecki>	 oh you are right
[01:07:44] <jzerebecki>	 damn the time selector is in local time?
[01:07:58] <bd808>	 yuck. it may be actually
[01:10:53] <bd808>	 jzerebecki: that's the dumb tool trying to be smart. It uses javascript to munge the time into local time on the client
[01:12:17] <jzerebecki>	 logstash at the correct time isn't more enlightening
[01:13:47] <jzerebecki>	 wtf the private window doesn't have the missing edit icons
[01:13:59] <jzerebecki>	 the local storage resource loader cache?
[01:14:41] <hoo>	 possibly
[01:18:21] <jzerebecki>	 hoo: yup it was local strage. does that mean it is a bug in the resource loader cache?
[01:18:42] <hoo>	 I guess so
[01:19:00] <hoo>	 RL cache should purge after at max 5 minutes
[01:22:02] <jzerebecki>	 oddly the sitelinks were correct
[01:24:12] <jzerebecki>	 also why were the [] not around the edit?
[01:27:40] <hoo>	 yeah, that's odd indeed
[01:32:55] <hoo>	 jzerebecki: I skimmed (almost) the full diff between wmf7 and wmf8 and couldn't find anything interesting
[01:33:47] <hoo>	 (except of the stuff that breaks the ArticlePlaceholder)
[01:35:50] <jzerebecki>	 ok i'll call it a night
[01:35:55] <jzerebecki>	 thx for the support
[01:37:00] <hoo>	 Good night
[01:37:02] <hoo>	 cu tomorrow
[10:53:34] <hoo>	 aude: Already in the office?
[10:57:18] <hoo>	 ah right, you're not even in Berlin
[10:57:43] <hoo>	 DanielK_WMDE__: Around?
[11:01:07] <DanielK_WMDE__>	 hoo: hey
[11:02:38] <hoo>	 DanielK_WMDE__: We noticed a strange thing during the deploy of wmf8 to testwikidata yesterday
[11:02:40] <hoo>	 and I ran out of ideas
[11:02:50] <hoo>	 We got empty ParserOutputs (and those were cached)
[11:02:55] <hoo>	 but only briefly after the deploy
[11:03:09] <hoo>	 I wasn't able to reproduce that locally, no matter how I mixed the two versions
[11:03:25] <hoo>	 even running jobs created in wmf8 in wmf7 and stuff like that didn't help
[11:04:01] <hoo>	 And as told, maybe 20m after the deployment we weren't able to even reproduce it on testwikidata anymore
[11:04:02] <hoo>	 no errors in the log, nothing
[11:04:05] <hoo>	 any more ideas what to poke at?
[11:04:47] <DanielK_WMDE__>	 hoo: you could sprinkle wfDebugLog throughout the view code
[11:05:11] <DanielK_WMDE__>	 And perhaps a few Assert calls to check for null.
[11:05:26] <hoo>	 DanielK_WMDE__: I wasn't able to reproduce it a single time
[11:06:13] <DanielK_WMDE__>	 but with the asserts, we might be able to trace it next time it happens
[11:06:27] <hoo>	 and I couldn't see anything obvious which I would like to assert
[11:06:27] <hoo>	 (reproduce it locally, that is)
[11:06:38] <hoo>	 http://pastebin.com/i1GpjzLb that was one the messed ParserOutputs
[11:06:51] <hoo>	 (as retrieved from ParserCache)
[11:08:31] <hoo>	 hm
[11:08:37] <hoo>	 what would you assert? That mText is never empty=
[11:08:39] <hoo>	 * ?
[11:09:14] <hoo>	 That's the only weird thing I can spot from that ParserOutput (offhand)
[11:10:32] <addshore>	 hoo *facepalm* over the article placeholder stuff >.>
[11:10:39] <hoo>	 addshore: I know
[11:10:54] <addshore>	 this is becoming a hell
[11:10:59] <hoo>	 But there's no easy solution right now which doesn't totally suck
[11:11:08] <aude>	 hoo: sending a mail about parser cache
[11:11:19] <aude>	 in my daily, but i don't have a solution yet
[11:11:21] <addshore>	 no, I'll try and rebase those patches and merge them today
[11:11:28] <addshore>	 but blargaghjhafs
[11:11:28] <hoo>	 We also considered duplicating the code in question… but that seemed even worse (several classes) 
[11:11:40] <hoo>	 aude: Oh ok, so you poked as well?
[11:11:47] <aude>	 hoo: i think refreshlinks is messing with cache
[11:12:12] <hoo>	 I thought that as well, but couldn't reproduce
[11:12:14] <aude>	 if i run refresh links, such as after adding a statement, then reload and *bom*
[11:12:18] <aude>	 boom*
[11:13:55] <hoo>	 On wmf8?
[11:13:55] <aude>	 but do uselang=de and ok (since it's different cache)
[11:13:56] <aude>	 not just wmf8
[11:13:56] <hoo>	 on MediaWiki master?
[11:13:56] <hoo>	 Maybe I should poke more
[11:13:56] <hoo>	 and also add that assert (and backport)
[11:13:56] <aude>	 no , i think this has been an issue before
[11:13:56] <hoo>	 I was poking at the jobs quite a lot yesterday
[11:13:56] <aude>	 don't know why it's not been reported though and maybe not happening in practice so much
[11:14:11] <hoo>	 didn't try wmf7 Wiki with wmf8 job runner, although that should not happen ever in production
[11:15:33] <aude>	 hoo: https://test.wikidata.org/wiki/Q1608 (for example, has no content) :(
[11:15:44] <aude>	 scary
[11:15:58] <aude>	 i also have no icons sometimes
[11:16:16] <hoo>	 I'll poke at this some more, I guess
[11:16:47] <aude>	 ok
[11:19:46] <aude>	 hoo: https://github.com/wikimedia/mediawiki/blob/master/includes/jobqueue/jobs/RefreshLinksJob.php#L207-L215 is always getting executed for me
[11:20:19] <aude>	 don't know if this is new, or always happened before, or what or if it is the problem
[11:20:42] <hoo>	 that sounds suspicious
[11:20:47] <aude>	 if i comment out the $generateHtml check in EntityParserOutputGenerator, then i don't have the issue
[11:22:00] <hoo>	 Argh, doh
[11:22:17] <aude>	 but this code doesn't appear new to me or such
[11:23:12] <hoo>	 but it makes sense that sometimes goes awry
[11:23:31] <hoo>	 Patch for that is trivial
[11:24:54] <hoo>	 ah, why didn't I see that before
[11:26:40] <aude>	 hoo: which way do you think we should fix?
[11:27:50] <hoo>	 A ParserOutput without html mustn't be cacheable
[11:28:31] <aude>	 agree
[11:52:47] <Amir1>	 addshore: hey, I was asleep
[11:53:50] <Amir1>	 the most important things is, I wrote this in one day
[11:53:50] <Amir1>	 I know it has flaws
[12:00:37] <Amir1>	 I would love to work on this and make it uniform 
[13:06:21] <hoo>	 aude: RE your mail
[13:06:32] <hoo>	 alternatively we could fix the ParserCache stuff in core
[13:07:40] <hoo>	 Well, not really
[13:07:47] <aude>	 hoo made https://phabricator.wikimedia.org/T120935
[13:08:00] <hoo>	 Core's AbstractContent could also disable caching
[13:08:08] <hoo>	 but we don't call out to it in ItemContent
[13:08:19] <aude>	 we could change refreshlinks, but there must have been some motivation to have it do caching
[13:08:35] * aude wonders when this would make sense to do
[13:08:43] <hoo>	 Yeah, sure
[13:08:48] <hoo>	 that is pretty obvious
[13:09:05] <hoo>	 Well, in case you don't have something in the cache, but just generated the whole thing (for some reason)
[13:09:28] <aude>	 w/o html
[13:09:54] <hoo>	 well, not everything even supports that, I presume
[13:10:12] <aude>	 oh really?
[13:10:17] <aude>	 what does wikitextcontent do?
[13:11:56] <hoo>	 It only takes the flag into account for redirect
[13:11:57] <hoo>	 s
[13:12:02] <aude>	 i see
[13:12:12] <hoo>	 the parser has to do the heavy lifting anyway otherwise (find links, images, …)
[13:12:32] <hoo>	 I guess the parser's not modular enough to scrape these without also generating html
[13:12:51] * aude thinks refreshlinks should request html + cache or not cache (if not requesting html) or differentiate the cache key
[13:13:08] <aude>	 anyway, i made the bug and maybe aaron can weigh in
[13:13:20] <aude>	 (and patch in wikibase is ok for now)
[13:13:38] <hoo>	 Maybe ParserOutput should know about whether it has html?
[13:14:35] <aude>	 sure
[13:14:45] <hoo>	 Would be false in most of these cases, except the wikitext one where we (through side effects) have the html (in some cases)
[13:15:16] <hoo>	 Meh, to many layers involved for an easy fits it all case :S
[13:15:45] <aude>	 yeah :/
[13:15:56] * aude checks out the patch
[13:17:06] <hoo>	 I'll be at Hauptbahnhof in a few moments, will be back later on (after also having lunch)
[13:17:12] <aude>	 ok
[13:17:24] <aude>	 still need to figure out the issue with icons
[13:17:26] <aude>	 and caching
[13:54:26] <addshore>	 aude or someone please please please +2 https://gerrit.wikimedia.org/r/#/c/257878/
[13:55:53] <aude>	 addshore: can you add a space after the comma? :)
[13:56:06] <addshore>	 ya
[13:56:16] <aude>	 ok :)
[13:56:18] <addshore>	 wait, which comma? :D
[13:56:59] <aude>	 between 'itself' and 'but'
[13:57:25] * aude is slightly OCD about commas
[13:57:28] <addshore>	 so you mean add a comma too? ;)
[13:57:55] <addshore>	 {{done}}
[13:57:55] <AsimovBot>	 How cool, addshore!
[13:57:56] <aude>	 i think
[13:58:28] <aude>	 ok
[13:59:00] <addshore>	 frimelle: ^^ or you can +2 it :D (also, AP should be fixed on master again once those 2 are merged)..
[13:59:12] <aude>	 {{done}}
[13:59:12] <AsimovBot>	 How cool, aude!
[13:59:19] <aude>	 :D
[14:03:04] <aude>	 o_O
[14:03:07] <aude>	 is jenkins broken?
[14:03:12] <hoo>	 aude: I think so
[14:04:03] <hoo>	 the failures I see are permission ones and stuff like that
[14:04:13] <hoo>	 nothing related to Wikibase or even MW
[14:05:07] <jzerebecki>	 hoo: hashar told me yesterday that that was fixed :(
[14:05:25] <jzerebecki>	 chmod: changing permissions of ‘/mnt/home/jenkins-deploy/tmpfs/jenkins-3’: Operation not permitted
[14:05:30] <jzerebecki>	 https://integration.wikimedia.org/ci/job/mwext-Wikibase-repo-tests-sqlite-hhvm/6972/console
[14:05:41] <hoo>	 yeah, that
[14:05:44] <jzerebecki>	 on integration-slave-trusty-1016
[14:05:49] <hoo>	 blocking various changes of mine
[14:08:03] <addshore>	 bah why is it broken :P
[14:08:20] <addshore>	 also, hi hoo!
[14:09:34] <hoo>	 :D hi
[14:16:51] <JeroenDeDauw>	 Adrian_WMDE: regarding that blog post... the vote on that PSR finally closed today https://groups.google.com/forum/#!topic/php-fig/dSw5IhpKJ1g
[14:16:52] <JeroenDeDauw>	 Adrian_WMDE: all the drama though
[14:17:09] <Adrian_WMDE>	 :D
[14:17:36] <Adrian_WMDE>	 »Unanimous -1 from Doctrine among all core developers.«
[14:20:42] <JeroenDeDauw>	 Adrian_WMDE: which includes the guy that wrote the NyanCat PHPUnit result printer, so yeah, that's very serious
[14:56:44] <addshore>	 hoo: / jzerebecki / aude composer on the cluster is still a no no correct?
[14:57:48] <aude>	 addshore: at least in jenkins, i think there is a fixed version used
[14:58:05] <aude>	 what do you have in mind?
[14:58:08] <addshore>	 oh yes, I think there is some composer gerrit thing now maybe
[14:58:11] <addshore>	 I'll have a look*
[14:58:35] <aude>	 https://github.com/wikimedia/integration-composer
[14:58:39] <addshore>	 We'll I might potentially want to install https://github.com/JeroenDeDauw/JsonDumpReader on an analytics machine as part of the metrics stuff
[14:58:51] <aude>	 not sure to what extent we use that currently
[14:59:14] <JeroenDeDauw>	 addshore: install? you mean pull in into another app? cause its not an app on its own ;p
[14:59:21] <aude>	 addshore: just make a "build" like we do for wikidata
[14:59:24] <addshore>	 jeryes
[14:59:43] <addshore>	 aude: yeh, I may have to, still trying to decide if I need the dumps right now
[14:59:48] <JeroenDeDauw>	 addshore: can you file a small issue there describing the thing you found could be done easier?
[15:00:05] <JeroenDeDauw>	 Where you pass in this bz2:// path or whatever this looks like
[15:00:10] <addshore>	 why wont my client autocomplete names right now... BAH!
[15:00:33] <addshore>	 JeroenDeDauw, is there a ticket up about using compressed dumps?
[15:01:02] <JeroenDeDauw>	 addshore: https://github.com/JeroenDeDauw/JsonDumpReader/issues
[15:01:17] <addshore>	 no then :P
[15:05:27] <addshore>	 JeroenDeDauw http://php.net/manual/en/wrappers.compression.php
[15:06:25] <hoo>	 addshore: That has recently been implemented in hhvm as well (for bz2 at least)
[15:06:41] <hoo>	 but AFAIK reads only a single bz2 stream as well
[15:07:08] <JeroenDeDauw>	 addshore: https://github.com/JeroenDeDauw/JsonDumpReader/issues/4
[15:07:43] <addshore>	 Also, does anyone know how I can do a sparql query, but only for entities with ids > 0 < 100000 ?
[15:25:08] <addshore>	 jzerebecki: that max size thing is merged and deployed right? and the code for it is in test.wikidata.org right?
[15:25:24] <jzerebecki>	 addshore: yes
[15:25:39] <addshore>	 mhhhhm
[15:25:54] <addshore>	 316,935 bytes and still going...
[15:26:16] <JeroenDeDauw>	 addshore: #over9000
[15:31:32] <addshore>	 The UI is actually very usable at that size on test too, for only strings :p
[15:42:28] <addshore>	 oh wait, im a factor of 10 out of course ;)
[15:45:44] <nikki>	 addshore: if you have an existing query and just want to filter it, filter (regex(str(?item), "Q[0-9]{1,5}$")). seems to work
[16:02:37] <addshore>	 nikki: I wonder if you can do a integer comparison insteaD?
[16:09:47] <nikki>	 addshore: uhh... the only thing I can come up with is filter (xsd:decimal(replace(str(?p), ".*Q", "")) < 10000). since I don't know of a way to get just the number
[16:10:22] <nikki>	 I'm not a sparql expert though, maybe someone else knows how to do that in a nicer way
[16:27:32] <addshore>	 xD
[16:39:33] <addshore>	 hah https://www.wikidata.org/wiki/Wikidata:Contact_the_development_team#Something_weird
[17:48:05] <chasemp>	 property-suggester.wikidata-dev.eqiad.wmflabs seems unavailable and has bad ldap info, can anyone get into it to run puppet?
[18:07:08] <Mavrikant>	 what happens merged page's wikidata entities? Do they merge to? If not should I do something?
[18:11:39] <addshore>	 Mavrikant: I dont follow?
[18:12:00] <addshore>	 if the Items are about two separate things then they should not be merged
[18:13:03] <Mavrikant>	 I am talking about double pages on Wikipedia.both pages have diffrent WD entitiy.
[18:17:24] <addshore>	 if the Items are about two separate things then they should not be merged ;)
[18:18:09] <Mavrikant>	 addshore, They are exactly SAME.
[18:18:22] <addshore>	 then they should be merged! :D
[18:19:52] <Mavrikant>	 by system right?
[18:21:52] <addshore>	 nope! youll have to merge them! there is a gadget for it!
[18:23:45] <Mavrikant>	 addshore, Thanks. I know merge.js
[20:31:09] <Migrant>	 what has happened here, is it unfulfilled id-merge ? https://www.wikidata.org/wiki/Q1331068
[20:31:55] <Migrant>	 what has happened here, is it unfulfilled id-merge ? https://www.wikidata.org/wiki/Q1331068
[20:37:23] <nikki>	 Migrant: it looks like someone moved the sitelinks to https://www.wikidata.org/wiki/Q671764 instead of merging the two items
[20:43:37] <Migrant>	 okay... thanks for the answer nikki. will you do the follow up of requesting it for deletion and/and or merge the items ?
[20:44:20] <nikki>	 I've merged them. we don't delete duplicates :)
[20:44:46] <Migrant>	 okay... :) thanks for your followup then...
[21:02:10] <multichill>	 I like the new layout! Minor point: For some reason hitting cancel seems to happen accidentally every once in a while. Probably better to move it a bit more out of the way ;-)
[21:03:00] <hoo>	 multichill: You mean when you actually want to remove a Statement?
[21:03:15] <multichill>	 Nope, when I added something and I want to click save
[21:04:04] <multichill>	 Save is smaller than cancel and the two are glued together
[21:04:43] <multichill>	 So if you're editing lot's of things in a row rapidly you might click the wrong one
[21:05:12] <hoo>	 hm... not sure that's easy to fix
[21:05:18] <hoo>	 space is extremely rare
[21:06:09] <multichill>	 save bit bigger in the font, cancel smaller and maybe a bit of extra space in between?
[21:12:57] <hoo>	 that might look very odd
[21:13:02] <hoo>	  but no idea
[21:13:43] <nikki>	 I've accidentally removed statements a load of times when trying to save them, even before it was altered :/
[21:24:10] <nikki>	 and I generally find that there's too much whitespace in places I don't need it, and not enough in places I do
[21:46:59] <hoo>	 DanielK_WMDE__: ping
[21:47:02] <hoo>	 kinda urgent
[21:48:13] <DanielK_WMDE__>	 hoo: i'm in the archcom session, done in 10 minutes or so
[21:48:19] <hoo>	 ok
[21:48:47] <DanielK_WMDE__>	 just tell me, i'll respond when I can
[21:50:15] <hoo>	 DanielK_WMDE__: Is it ok for VT to not have a validator callback?
[21:52:04] <hoo>	 or should the callback throw if that's not a valid thing?
[21:53:25] <hoo>	 (to be concrete: Can there be a validator for a wikibase entities or should we only have these for items/ properties specifically)
[21:56:21] <hoo>	 To be even more concrete
[21:56:21] <hoo>	 https://gerrit.wikimedia.org/r/258040
[22:08:35] <DanielK_WMDE__>	 hoo: we can have validator callbacks for VTs. if there is one for the PT, the VT one will not be used though.
[22:09:04] <DanielK_WMDE__>	 hoo: i don't think we need separate validators for P and Q entity references
[22:09:56] <DanielK_WMDE__>	 not sure if a validator callback should be required... it probably should. if no validation is necessary, we should explicity set a dummy validator
[22:10:22] <DanielK_WMDE__>	 that would safeguard against forgetting to set the validator, or making a typo in the array key (easy mistake to make)
[22:10:37] <hoo>	 DanielK_WMDE__: hm... if we use a single validator, couldn't I mix items/ properties?
[22:10:49] <hoo>	 I mean set ones which aren't supposed to be set
[22:11:05] <DanielK_WMDE__>	 oh, true. we should check that it has the correct type!
[22:11:18] <hoo>	 I don't think there's any further check
[22:11:23] <DanielK_WMDE__>	 i was only thinking of the existance check. but you are right
[22:11:39] <DanielK_WMDE__>	 yea. better have separate validators for the PTs, then :)
[22:13:02] <hoo>	 Amended as  EntityExistsValidator supports entityType = null… but still, we should have specific validators if we can :D
[22:14:03] <DanielK_WMDE__>	 one validator class, but distinct instances for the different types, yes
[22:14:20] <DanielK_WMDE__>	 and all VTs should have validators. do they?
[22:14:55] <DanielK_WMDE__>	 sems like they do
[22:19:18] <hoo>	 DanielK_WMDE__: Yeah, they do
[22:19:25] <hoo>	 I tested it without and it worked fine
[22:19:33] <hoo>	 but still for the sake of completeness, I think we should have it
[22:19:36] <hoo>	 anyway, tests added
[22:28:42] <DanielK_WMDE__>	 hoo: thanks. good catch, nasty bug >_<
[22:30:16] <DanielK_WMDE__>	 hoo: think i should drunk-merge it?
[22:30:36] <DanielK_WMDE__>	 this is causing problems on the live system, no? Can't link to property IDs any more...
[22:30:52] <hoo>	 Yeah, that's how we noticed
[22:31:02] <DanielK_WMDE__>	 ah, just noticed the ticket ;)
[22:31:04] <hoo>	 integration tests :P
[23:31:17] <hoo>	 Wikidata's fine again! :)
[23:41:31] * Platonides wonders why is the empty span needed
[23:42:30] <hoo>	 It's being use for the icon (If we're talking about the same)