[10:53:30] <dcausse>	 lunch
[13:09:35] <inflatador>	 <o/
[13:29:02] <dcausse>	 o/
[13:34:46] <ebernhardson>	 \o
[13:37:58] <ebernhardson>	 hmm, reviewed the first search wish" " insource search operator without the resort to complicated regex, maybe addition of a simple keyword to the string for each case." and "Further would be a keyword to exclude categories"
[13:38:27] <ebernhardson>	 i mean i think it's actually a more complicated ask, but first have to parse out the words :P
[13:39:54] <dcausse>	 yes this one is complicated and putting way too much into insource
[13:41:19] <ebernhardson>	 indeed it would have to be some other solution that's not insource..i would be tempted to intext:// (excluding operational concerns) but that won't help this user
[13:43:08] <dcausse>	 perhaps in the second pass we could do smarter things but this sources kind of complicated, would need to parse wikitext and keep the context of every characters to discard matches from the regex
[13:43:29] <dcausse>	 s/sources/seems
[13:44:57] <dcausse>	 ebernhardson: sorry to bother with this but if you get a moment could you have second look at https://gitlab.wikimedia.org/repos/search-platform/discolytics/-/merge_requests/77 ?
[13:45:19] <ebernhardson>	 yea i'm not really sure how to get crazy with the source matching...i could imagine extracting more text from the page to query sepately (like template params)...but not overloading insource
[13:45:23] <ebernhardson>	 sure, looking
[13:46:46] <dcausse>	 thanks!
[13:47:57] <dcausse>	 yes seems to be worded like "I don't want to write complex regex but I want match specific portion of text from the wikitext"
[13:48:23] <dcausse>	 so definitely insource is not the right tool because this one is mainly used for regex
[16:55:23] <dcausse>	 inflatador: if/when you have a minute could you add a new section in the private repo with the following entries: https://phabricator.wikimedia.org/P89890 right after the "wikimedia_enterprise" entry for airflow-search?
[16:57:17] <dcausse>	 context is https://gerrit.wikimedia.org/r/c/operations/deployment-charts/+/1255778
[17:42:21] <inflatador>	 dcausse sure, looking now
[17:42:30] <dcausse>	 thanks!
[17:42:51] <dcausse>	 not sure what to respond to https://www.mediawiki.org/wiki/Extension_talk:CirrusSearch#OpenSearch_1.3_installation_no_longer_possible
[18:00:43] <ebernhardson>	 personally, i would still accept the old key. But i realize people don't like that :P
[18:01:10] <inflatador>	 I'd ask them to open a ticket on the OpenSearch github or ask in their Slack
[18:01:25] <inflatador>	 Although it is a good reminder that we should probably upgrade prod to 2.x or 3.x soon ;)
[18:01:36] <dcausse>	 yes :)
[18:17:01] <dcausse>	 dinner
[18:21:49] <inflatador>	 dcausse private repo stuff is ready
[19:23:23] <dcausse>	 thanks!
[20:50:08] <ebernhardson>	 Looks like the fix for semantic on Special:Search worked in prod: https://en.wikipedia.org/w/index.php?title=Special:Search&cirrusSemanticSearch=&fulltext=1&ns0=1&profile=advanced&search=Why+do+stars+shine?
[20:51:28] <ebernhardson>	 os 2.19.5 plugins also looks all ready now, will need to merge the update to .deb building, ship it to apt.wikimedia.org, and then we can release the updated image
[21:02:50] <inflatador>	 {◕ ◡ ◕}
[21:41:52] <inflatador>	 Just merged https://gerrit.wikimedia.org/r/c/operations/puppet/+/1254320 , so the OpenSearch readahead stuff should be set automatically (CODFW only, we'll roll out to EQIAD next week if nothing blows up)