[11:03:15] * gmodena waves
[11:03:57] <gmodena>	 Hi folks! I'm compiling a list of WDQS-related airflow dags we should move off Search and into Wikidata Platform new instance 
[11:04:19] <gmodena>	 could you help me sanity check if I got the list right, or if I might be missing something: https://phabricator.wikimedia.org/T414426
[11:04:28] <gmodena>	 no urgency! I'm just doing some grooming
[11:29:21] <dcausse>	 gmodena: hey! sure will take a look
[11:29:32] <gmodena>	 dcausse thanks!
[11:35:54] <dcausse>	 lunch
[14:22:33] <inflatador>	 <o/
[14:29:46] <dcausse>	 o/
[14:32:03] <pfischer>	 o/
[14:35:40] <pfischer>	 I am trying to match requests logged to `event.mediawiki_cirrussearch_request` against logstash access logs (https://logstash.wikimedia.org/app/dashboards#/view/018bde90-a08d-11ed-8137-c3b9b9c0225e?_g=h@e014bf4&_a=h@87b0727) but I simply can’t find an access log entry for one particular row. Are the access logs sampled somehow?
[14:40:17] <dcausse>	 pfischer: I don't think I ever used these access logs
[14:46:50] <ebernhardson>	 \o
[14:47:17] <dcausse>	 o/
[14:52:03] <ebernhardson>	 pfischer: hmm, yea i don't think i've tried to join those before either
[14:52:24] * ebernhardson avoids responding "you could just return false here" to the patch that changes ArrayCirrusSearchResult to have an isRedirect() function
[14:55:48] <ebernhardson>	 hmm, maybe throgh comp suggest?
[14:56:47] <dcausse>	 for the redirects I'm pretty sure "core" should know this since it can resolve redirects?
[14:57:23] <ebernhardson>	 they basically just do $this->getTitle()->isRedirect(). But at least from the primary index that should always be false, redirects can't be a search result
[14:57:35] <ebernhardson>	 comp suggest has redirects, so maybe that's their aim?
[14:58:55] <dcausse>	 yes... but not totally sure that's the right place, the SearchResult interface does not have a isRedirect() method
[14:59:27] <ebernhardson>	 indeed
[15:01:28] <dcausse>	 seems to me a lot easier to do close to api response
[15:01:58] <ebernhardson>	 yea, i suppose will try and suggest that instead.
[15:56:07] <ebernhardson>	 of course...running the integration tests on cindy a second time (without environment reset) passes the test suite...but not my patch directly :S 
[15:58:22] <ebernhardson>	 tempted to sleep before building the comp suggest index...but such a poor solution
[16:03:39] <dcausse>	 the index building might have different code paths if the index already exist iirc
[16:04:02] <inflatador>	 ryankemper we're in stand-up if you're around
[16:40:53] <ebernhardson>	 (paused cindy while figuring things out, it wont be voting)
[16:45:47] <ebernhardson>	 oh! Maybe this is unexpected dependencies of the @suggest hook.  It passes most comp suggest tests, but it doesn't pass a few.  My test added an @suggest annotation to a different feature file, maybe it's just initializing from a different state that doesn't (yet) have the documents expected 
[16:45:49] <ebernhardson>	 hm
[16:46:34] <ebernhardson>	 would explain why it works without resetting the env
[16:56:41] <ebernhardson>	 tempted to define our own quasi-dag, but not sure it's worthwhile :S Would essentially mean defining a dependencies list, moving all the implementation functions (for BeforeOnce) into a map, and then initilizing them in-order ourselves...But sounds like massive overkill, maybe just be lazy and move the test into a file that already has @suggest
[17:00:10] <dcausse>	 yes... I thought the suggest tests would use a dedicated set of docs but very probable that some assume some existing docs are there :/
[17:02:25] <ebernhardson>	 i'm sure they are supposed to, but over time we've added things.  I guess the alternative "right" solution is to define new docs for all the failing tests and initialize them from @suggest
[17:02:50] <ebernhardson>	 but the comments (i probably wrote) in the suggest init hook basically say "hope this runs last"
[17:05:04] <dcausse>	 :)
[19:05:59] <dcausse>	 dinner
[22:16:27] <inflatador>	 ryankemper just left pairing, hit me up if you want to join. I'm looking at some new WDQS allowlist tickets
[22:16:58] <ryankemper>	 inflatador: whoa totally lost track of time, brt
[22:42:06] <ebernhardson>	 Trey314159: I think the new debug docs should be ready, if you have a chance would appreciate a review. No big hurry, most of it is existing features but some is new stuff still in gerrit so untestable (action=cirrussuggestdump, and action=cirrus-schema-dump): https://phabricator.wikimedia.org/P86859
[22:42:28] <ebernhardson>	 or i guess i could just put it in mediawiki.org today, hmm
[22:54:26] <ebernhardson>	 https://www.mediawiki.org/wiki/Help:CirrusSearch/Debug