[09:16:44] <_joe_>	 can anyone point me to where we set Cache-control in MediaWiki?
[09:16:45] <_joe_>	 else I'll go code grepping
[09:56:14] <legoktm>	 _joe_: OutputPage probably
[09:56:46] <legoktm>	 https://codesearch.wmflabs.org/core/?q=cache-control&i=fosho&files=&repos=
[09:57:20] <legoktm>	 so OutputPage (for index.php), ApiMain and ResourceLoader
[09:57:26] <_joe_>	 legoktm: yeah found that, but since you're here... I'm now looking at purges
[09:57:38] <_joe_>	 and it seems to me by looking at purges happening in prod
[09:57:58] <_joe_>	 that we send the purges for action=history even when purging linked pages
[09:58:22] <_joe_>	 so you change a template used in a page => we send a purge for that page's action=history, which AIUI is useless
[09:58:37] <legoktm>	 https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/core/+/master/includes/Title.php#3546
[09:58:57] <_joe_>	 yep
[09:59:53] <_joe_>	 why would we be sending purges for action=history when a template/a wikidata item used in the page changes? that won't appear in the page history, right?
[10:00:52] <legoktm>	 I'm git blaming
[10:01:14] <_joe_>	 I mean it makes sense to purge that page for direct edits
[10:01:32] <_joe_>	 but I don't think it makes sense for purges because of changes in linked pages
[10:02:05] <legoktm>	 https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/core/+/21ca86666fc%5E%21/#F0
[10:02:18] <legoktm>	 yep
[10:02:32] <legoktm>	 I think people just started using that function for other things because it was convenient to use
[10:03:23] <legoktm>	 seems like it would be worth refactoring
[10:03:57] <_joe_>	 <3 I'm preparing a task
[13:38:16] <_joe_>	 can someone tell me if I'm missing something fundamental here? https://phabricator.wikimedia.org/T250261
[13:53:34] <Reedy>	 Definitely if there's no edit to the page itself, there's no benefit to purging it's history...
[13:54:24] <Reedy>	 Obviously if the page is edited itself, I guess we're explicitly purging the history to prevent being stale if someone browses to it immediately after editing it
[14:02:00] <_joe_>	 so sure, for direct edits, that makes sense
[14:48:59] <Reedy>	 I'm guessing git blame doesn't really yield anything useful
[14:54:38] <_joe_>	 Reedy: the purge of action=history was added by Gabriel in 2004
[14:55:00] <Reedy>	 heh
[14:55:43] <_joe_>	 so my guess is
[14:55:54] <_joe_>	 it made sense to be done unconditionally back then
[14:55:57] <_joe_>	 not anymore :P
[14:59:41] <Reedy>	 heh