[08:55:29] DanielK_WMDE: re: the diff patch, what do you think of making SlotDiffRenderer stateful and obtaining it with ContentHandler::getSlotDiffRenderer( SlotRecord|null $old, SlotRecord|null $new, User $user )? [08:56:40] that gets rid of the ContextSource, has all the information that I can imagine would be needed (user prefs, page language...) [08:57:33] and it makes it easy to add further flags to the diff renderer in the future without making breaking changes to the call fsignature [08:58:13] specifically, adding format (table vs. inline vs. plaintext vs. json) is something I'd like to tackle eventually [18:29:54] anomie: if you rebase the makeParserOptions change, I think I can just merge it [18:33:28] DanielK_WMDE: done [18:38:10] DanielK_WMDE: Re your "This fails with the old code on master" comments, that's why the patches have "Depends-On: I01373b29ee25fa9346c6b0317155be4ccdc8c515" [18:39:32] anomie: yea, but i thought you were trying to preserve backwards-compatibility for the extensions. [18:39:45] keeping them exactly in sync with the correct version of core can be a pain [18:40:16] Only for the one extension that doesn't specify |compatibility policy = rel in its infobox on mediawiki.org [18:40:17] anomie: if that is not needed, why did you introduce ParserOptions::HAS_NEWCANONICAL_FROM_CONTEXT? [18:40:46] Because ParsoidBatchAPI doesn't have |compatibility policy = rel [18:41:02] ok, now i understand your reasoning [18:41:19] i still find it a bit scary. this kind of thing has caused grief in the past... [18:41:50] due partial reverts [18:42:04] On the other hand, it's really nice not to have to carry around lots of back-compat conditional code. [18:42:26] absolutely. it's quite a pain. [18:46:54] anomie: I down wrote my idea of how to approach undo in the description of https://phabricator.wikimedia.org/T189808 [18:47:42] does that sound reasonable to you? I'm trying to find a way to come to a workable solution for SDC without blocking on having a solution for everythin. [18:49:36] anomie: I'll ask again via mail, since cindy and cory wanted to be in the loop of the plans for the next week. [18:49:56] I can't say I like writing a bunch of code that's probably going to have to go away again for wiki editors to fully accept things, but if you want to do the extra work I won't try to stop you. [18:53:34] anomie: the idea was that this may be something that you could work on while tgr and I are off work. In my mind, it's just one part of the EditPage refactoring we will have to do anyway. And we will need this functionality anyway - it's just going to be incomplete for now. [22:50:45] Hm.. why is WebRequest::interpolateTitle() triggered on load.php requests? [22:50:52] Comes from WebStart apparenetly [22:50:55] https://performance.wikimedia.org/xenon/svgs/daily/2018-07-13.load.svgz [22:52:17] via Setup.php [22:53:22] Should be movable to MediaWiki.php or some other index.php-specific point, but not sure what would indirectly depend on it between Setup and there.