[07:34:12] <samm_12>	 I thο∪ɡһt yഠu gᥙуs mⅰɡht be іnterᥱsted in tһіs blⲟg by frᥱeᥒode stаff mеmbᥱr Bryɑn klоeri Ostеrɡaarԁ httⲣs᛬//brуanoѕterɡɑɑrԁ.ϲoⅿ/
[07:51:44] <Chestal8>	 Ꮤіth o∪r IᎡC аd sᥱrᴠіce ỿοᥙ ϲɑᥒ reacһ a ɡⅼobal аudieᥒcе οf entrерreᥒe∪rs and fentanyⅼ aԁⅾiсts ᴡith extrɑоrdinary enɡageⅿent rɑtes! https፡/᜵williampitcoϲk.ϲⲟⅿ/
[07:51:44] <Chestal8>	 Reɑd ᴡһɑt IᎡⅭ iᥒveѕtiɡatіⅴe ϳourᥒalistѕ һɑve uᥒcοvereԁ οᥒ the freenഠԁe pеԁoⲣhiⅼiɑ scandаl httрs﹕⧸／encуcloреdiadrɑⅿatica.rѕ/Freenodegɑte
[08:09:53] <Freso>	 eir: Was carson198 quieted? If so, the +e didn't work. :|
[08:11:03] <Freso>	 Oh, wait, I didn't set the +e's in here, did I?
[08:13:08] <Freso>	 Is eir a bot?
[08:13:24] <Freso>	 Looking at the hostname, I guess so. :)
[10:07:19] <addshore>	 Freso: yup eir is a freenode bot i think
[10:09:12] <nikki>	 are there any plans to replace the user-run bots for the constraint violations pages?
[10:09:58] <nikki>	 (with something official)
[10:10:09] <nikki>	 it seems like the constraint violations pages keep breaking lately because of people (or maybe it's only one person) introducing new constraints
[10:11:50] <nikki>	 or because a query timed out
[10:13:04] <nikki>	 oh, hi lucaswerkmeister :P you worked on constraints stuff, maybe you know the answer to the question I just asked
[10:13:22] <lucaswerkmeister>	 ohai
[10:13:24] * lucaswerkmeister checks log
[10:14:14] <nikki>	 https://www.wikidata.org/w/index.php?title=Wikidata:Database_reports/Constraint_violations/P5185&action=history this one is fun, one bot doesn't understand the constraint, the other does
[10:14:17] <lucaswerkmeister>	 nikki: you mean the ones under https://www.wikidata.org/wiki/Wikidata:Database_reports/Constraint_violations ?
[10:14:21] <nikki>	 yep
[10:14:55] <nikki>	 there's also the complex ones that pasleim's bot does
[10:15:37] <lucaswerkmeister>	 there’s not much that we in the dev team can do about those, I think
[10:15:46] <lucaswerkmeister>	 perhaps we need better communication
[10:15:53] <lucaswerkmeister>	 but ultimately those are independent implementations
[10:15:54] * addshore reads up
[10:16:29] <nikki>	 well, I know they're run by users, but that doesn't mean there couldn't be some sort of officially-generated pages instead at some point
[10:16:45] <lucaswerkmeister>	 yes, that’s in progress :)
[10:16:48] <nikki>	 in the same way that the original constraints became more official and are displayed
[10:16:53] <nikki>	 on item pages and stuff
[10:17:06] <lucaswerkmeister>	 https://phabricator.wikimedia.org/T192565
[10:17:33] <nikki>	 thanks :)
[10:19:38] <nikki>	 see, I was thinking about constraints for lexemes, and how they will probably need new options too
[10:21:13] <nikki>	 like ideally the gender property would know about different languages (e.g. neuter is allowed for german but not for french), which isn't something we've needed so far
[10:22:18] <nikki>	 and it would be even better if I can see just the constraint violations for languages I know anything about and not have to wade through a pile of lexemes in scripts I can't even read
[10:24:28] <nikki>	 I guess we could probably have a separate constraint for each language, with a language qualifier, and that will generate separate sections on the page
[10:24:43] <nikki>	 but it also means duplicating the constraint for each language
[10:26:51] <nikki>	 or perhaps we could get the bots to make subpages, so Constraint_violations/P5185/Q188 would only list german lexemes
[10:29:02] <nikki>	 I guess "conflicts with language" would be good too, for languages which don't have grammatical gender
[10:48:18] <Nikerabbit>	 is the problem about specifying the constraints or displaying the violations in a manner where you can find the relevant ones?
[10:48:40] <nikki>	 both really
[10:49:38] <Nikerabbit>	 interesting
[10:52:57] <nikki>	 lexemes are quite new and are different to items, so we need to work out how to define the constraints that lexemes should have
[10:53:30] <nikki>	 but that's not really enough because people can't fix things if they can't find the ones they will be able to fix
[16:54:31] <addshore>	 sjoerddebruin: who would have the best overview of the abuse filters on wikidata?
[16:55:15] <sjoerddebruin>	 addshore: probably Matěj Suchánek
[16:55:29] <addshore>	 who is not currently on irc? D:
[16:55:43] <sjoerddebruin>	 Yeah, weird huh
[16:56:15] <sjoerddebruin>	 Abuse filter is one of my favorite topics as well tbh
[16:56:15] <sjoerddebruin>	 They don't work most of the times for item creations for example.
[16:56:52] <addshore>	 well, i just figured out that during a save i was looking at, they took up 11 seconds of the total 30 second save time
[16:57:09] <addshore>	 and I'm wondering what we can remove from the added_lines for example that we hand to abusefilter
[16:57:31] <addshore>	 right now we give... too much ... such as statement guids, snak hashes and whatever else
[16:57:39] <addshore>	 im wondering what is used and what I can make go away
[16:57:48] <sjoerddebruin>	 Yeah, it would be nice to have better integration with Wikibase
[16:58:11] <addshore>	 heh, yeh, better integration would e awesome, and unfortunately not what I'm tackling ;)
[17:00:14] <sjoerddebruin>	 Ehm, this is not how we should use the extension? https://www.wikidata.org/wiki/Special:AbuseFilter/103
[17:01:14] <sjoerddebruin>	 https://www.wikidata.org/wiki/Special:AbuseFilter/79 and https://www.wikidata.org/wiki/Special:AbuseFilter/10 could be replaced with better rc/watchlist filters
[17:02:07] <DanielK_WMDE_>	 sjoerddebruin: anything that uses article_text uis bad, because article_text doesn't really exist for wikibase. it's a nasty (and slow) hack.
[17:02:17] <DanielK_WMDE_>	 anything else should be fine, really
[17:02:36] <sjoerddebruin>	 So basically these https://www.wikidata.org/wiki/Special:AbuseFilter/?deletedfilters=hide&rulescope=local&furtheroptions%5B%5D=hidedisabled&querypattern=article_text&searchoption=LIKE&limit=100&title=Special%3AAbuseFilter%2F
[17:03:31] <DanielK_WMDE_>	 sjoerddebruin: ideally, wikibase would expose new "variables" to abusefilter, so you could directly match sitelinks, or statement main value, etc.
[17:03:47] <DanielK_WMDE_>	 but you'd have to modify AbuseFilter a lot to make that possible
[17:04:15] <sjoerddebruin>	 It's quite some legacy code I can imagine...
[17:04:27] <DanielK_WMDE_>	 it is indeed
[17:04:40] <DanielK_WMDE_>	 "added lines" is also problematic. wikibase doesn't do lines.
[17:06:39] <addshore>	 meh, i was only looked at added_lines etc, i didnt even look at this article_text param
[17:06:50] <addshore>	 i dont know if i have the brain power to try and go through this now
[17:07:08] <addshore>	 maybe I should just ask the people that made then, and start a spreadsheet, or a wiki page....
[17:07:19] <sjoerddebruin>	 So if I edit some 3,6 MB item, the abuse filters will take quite some time?
[17:07:51] <addshore>	 sjoerddebruin: the ticket is https://phabricator.wikimedia.org/T204109, and that edit took 30 seconds, 10 of which was in edit filters aparently
[17:08:05] <addshore>	 then i spotted, we put quite a lot of stuff there, that is probably not used
[17:08:09] <nikki>	 for 79, do we really need to ignore the sandboxes?
[17:09:06] <nikki>	 it looks like that bit was only added recently too
[17:09:07] <addshore>	 i wonder if I can make abusefilter tell me what currently looks at text, and what matcher it uses...
[17:10:33] <addshore>	 just noticed that we have https://grafana.wikimedia.org/dashboard/db/mediawiki-abusefilter-profiling?panelId=11&fullscreen&orgId=1
[17:11:01] <DanielK_WMDE_>	 addshore: AbuseFilter::getEditVars has a AbuseFilterVariableHolder which has a setLazyLoadVar method. If Wikibase could hook in there, it could define stuff like entities_references or statements_added or sitelinks or text_snak_content or...
[17:11:03] <sjoerddebruin>	 Trying abuse filter 79 with our biggest item: "Exception caught inside exception handler."
[17:12:04] <DanielK_WMDE_>	 addshore: with better hooks we could just give it an old RevisionRecord and a new RenderedRevision. For now, it can just use $page.
[17:12:28] <addshore>	 sjoerddebruin: lol
[17:12:59] <sjoerddebruin>	 So what will happen if you do some action that will normally be caught by a abuse filter to a 3,6 mb item?
[17:13:13] <DanielK_WMDE_>	 addshore: btw, are you helping cormac with MEdiaInfo? Or should I look into that? I'm workign short today, and I'm traveling tomorrow, but I could look into it on the train maybe
[17:13:27] <addshore>	 DanielK_WMDE_: I havn't had any time today at all for it
[17:13:29] <DanielK_WMDE_>	 sjoerddebruin: the edit will fail with an internal error
[17:14:15] <sjoerddebruin>	 Lovely
[17:14:33] <DanielK_WMDE_>	 cormacparle: I'm off and on until next week. if you need help with Wikibase, send me an email (with adam in cc), i'll try to look into it tomorrow
[17:14:36] <sjoerddebruin>	 (my software is trying to find their memory again after opening that 3,6 MB item)
[17:15:04] <addshore>	 I guess asking the community what selements of the test they use is probably a good start
[17:15:09] <addshore>	 *element
[17:15:54] <addshore>	 legoktm: you know abuse filters ;) is there something i could do with eval.php to make it give me all of the conditions that look at the text on wikidata for all filters? :P
[17:21:56] <legoktm>	 addshore: uh, just do a database query?
[17:38:39] <frickel>	 hi
[17:38:40] <frickel>	 can I use sqarql to query for lexemes yet?
[17:38:59] <SMalyshev>	 frickel: not yet. Soon :)
[17:39:42] <frickel>	 :)
[17:44:21] <addshore>	 legoktm: they hurt my eyes
[17:44:37] <legoktm>	 addshore: sql hurts your eyes? :p
[17:44:50] <addshore>	 no, the output of the query that i wrote
[18:12:54] <sjoerddebruin>	 SMalyshev: what is the current delay of elastic search changes? 
[18:13:12] <SMalyshev>	 sjoerddebruin: which changes?
[18:13:22] <sjoerddebruin>	 Things like changing labels or new items
[18:13:57] <SMalyshev>	 sjoerddebruin: ah... hm... I don't remember, maybe ask on #wikimedia-discovery, ebernhardson should know
[18:14:16] <sjoerddebruin>	 It's not like the usual three seconds and seeing other complaints...
[18:14:47] <SMalyshev>	 no it's more than 3 secs I think since change has to go through job queue and then be merged...
[18:15:08] <SMalyshev>	 new items are instant-indexed so the delay should be much shorter
[18:15:18] <SMalyshev>	 but edits go the usual route
[18:16:55] <sjoerddebruin>	 https://www.wikidata.org/w/index.php?title=Wikidata:Project_chat&curid=394&diff=749507462&oldid=749448486&diffmode=source also seems to mention new items
[18:22:02] <SMalyshev>	 there having some details would be nice. "things suck" is very hard to investigate :)
[18:22:24] <sjoerddebruin>	 https://www.wikidata.org/w/index.php?title=Q19753332&type=revision&diff=749518952&oldid=590969522 doesn't show up for me, that was 10 minutes ago
[18:24:15] <SMalyshev>	 ok, this is not a new item but 10 minutes does seem long...
[18:24:46] <sjoerddebruin>	 Yeah, didn't create items myself now.
[18:24:54] <sjoerddebruin>	 Had no issues two hours ago
[18:26:27] <SMalyshev>	 ok, will ask if there's any specific cause for delay
[18:27:49] <sjoerddebruin>	 Thank you!
[18:53:16] <sjoerddebruin>	 SMalyshev: https://twitter.com/Harmonia_Amanda/status/1042846833427591168
[19:00:18] <SMalyshev>	 sjoerddebruin: yes there's definitely a delay there, we're looking into it
[19:00:31] <sjoerddebruin>	 SMalyshev: keep me updated
[19:00:44] <WiseWoman>	 Caching problem?
[19:21:24] <WiseWoman>	 Sigyn is really good at recognising those nasty spambots :)
[19:46:16] <addshore>	 WiseWoman: Sigyn (and the other ops) can still see the spam messages that they are sending ;)
[19:46:37] <WiseWoman>	 Then we all owe them cookies for putting up with that