[03:23:57] JustHannah: I left a small cleanup patch for you to review at https://gerrit.wikimedia.org/r/c/mediawiki/extensions/MultimediaViewer/+/963855 [04:08:19] TimStarling: ok to tag a release, or want to hold back for something else? [04:30:51] sure, release it [04:32:29] Krinkle: you want to respond to Danny before I +2 it? It is technically October 6 now [04:32:40] not where I am :) [04:34:09] yeah but you didn't specify the timezone, so fair to assume UTC [04:35:22] ok, it won't match the commit and tag meteadata in that case, but fine either way [04:36:27] amended [04:41:43] I don't think it's helpful for readers to make the date match the timezone in the git commit metadata, who is looking at the metadata anyway? [04:56:56] I don't know, but I guess the only scenario in which someone would know the date I tagged it on either by by looking at git log, git show, or gitiles, or something like that in which case it'll tell you that and not something in UTC. Or in a GUI like Gerrit where it's localised to your personal preferences and wrong in the other direction. It's only outside both git cli and a GUI where one could make the infrance that it "should" be UTC, [04:56:56] determine that, and then decide if its' right or wrong. That seems like more effort to me, but that may be because I don't live with a UTC clock around much. [09:58:09] I just used the new ResourceLoader source map support locally for the first time [09:58:13] and it mostly worked, which was pretty cool [09:58:42] but it was quite slow (as in, as soon as a `debugger;` statement was hit, dev tools would pause for like 5-10 seconds while the source was loaded) [09:58:55] and also my laptop tried to turn into a helicopter during this time :S [09:59:20] so I’m wondering whether I should try to optimize my setup in some way [09:59:53] or am I doing it wrong, and source maps are intended for production, but locally I should still prefer ?debug mode when possible? ^^ [10:00:37] that’s a question for… Krinkle or TimStarling, I guess? (once you’re online :)) [15:35:01] Krinkle: by shutdown do you mean __destruct? It looks like we mostly registered something there because that's what upstream does. Their justification was this ticket: ttps://github.com/Seldaek/monolog/issues/91 [15:36:05] essentially that __destruct isnt called on fatal errors, so those logs would be lost [15:36:25] (may or may not be true 10 years later) [17:12:22] ebernhardson: yeah, makes sense. [17:12:41] ebernhardson: so.. removing our override, worth trying? [17:13:58] Lucas_WMDE: it's pretty snappy for me. not sure where it would be spending that time. Do you have a sense of where that time was spent? Is it the sourcemap URL being fetched from a (slow?) localhost? [23:01:01] When using action=parse with the en.wiktionary.org api, fetching a section can take 6x longer than fetching the the whole thing. example "head" or "feel": section 1 takes 3 seconds, full 0.5 secs. https://en.wiktionary.org/w/api.php?action=parse&format=json&page=head§ion=1 [23:14:17] Lucas_WMDE: in my experience, source maps are faster than debug mode. Would be good to have some profiling of whatever the slow part is