[00:18:23] you broke Wikipedia? [16:01:22] dbrant: new standup time! [16:02:09] mdholloway: 1min [17:50:17] mdholloway: bearND got a quick one up that we want to get in before the deployment this afternoon: https://phabricator.wikimedia.org/T145830 [17:50:31] dbrant: FYI - patch above to extend the announcements passed today ^ [17:51:08] coreyfloyd: looking [17:53:07] coreyfloyd: we'll need to extend the Android date way past Dec 10. We haven't released to beta yet. [17:55:32] dbrant: what date do you propose? [17:59:09] dbrant: or feel free to add a comment with your proposed time (in UTC) to https://gerrit.wikimedia.org/r/#/c/324506/1/routes/announcements.js [21:22:08] mdholloway bearND: o/ sorry, my irc bouncer is having trouble today so you might get a question like 30 minutes from now or who knows when :] [21:22:28] mdholloway bearND: here's the question in REALtime: does vcr mode still work in the content service? i haven't been able to get anything to record in a long while which is a bummer since the tests are so slow without it [21:23:45] niedzielskii: i seem to remember some discussion in some venue about turning record mode off from running automatically but can't find an email or commit [21:23:53] niedzielskii: but yeah, not working at all for me either, lately [21:24:35] mdholloway: hm but i thought explicitly writing VCR_MODE=record npm test should work [21:24:45] niedzielskii: same here [21:24:58] mdholloway: like i don't even see files being cached [21:25:26] mdholloway: ok, well glad it's not just me. i'll look into it another time [21:25:43] I'm not using any special settings but I tend to `rm -rf fixtures` quite often [21:26:10] bearND: so it works for you without even an environment variable? [21:31:29] niedzielskii: it's enabled by default. I'm actually considering turning this behavior off by default [21:32:22] bearND: geez i wonder what's busted on my machine. that caching saves me a ton of time when it's working right [21:34:13] niedzielskii: Ideally we would just have it cache the responses from external requests but not the ones from localhost [21:38:32] bearND: +1 to that. maybe we can use one of the filtering or cassette options? https://www.npmjs.com/package/sepia [21:39:08] bearND: i think that would fix most caching issues [21:40:25] niedzielskii: Cool. I'll make a task for it [21:40:46] bearND: thanks that'll be awesome for development! [21:41:00] niedzielskii: Do you have a case in mind where that wouldn't solve the issue? [21:41:22] bearND: you mean the localhost filtering? [21:41:38] niedzielskii: yes [21:42:11] bearND: oh, i think the only time we'd ever have a problem would be if the restbase content was purged and a local cached copy existed [21:42:21] bearND: should be pretty rare i'd think [21:43:51] niedzielskii: yes. That's true [22:30:00] niedzielskii: Sorry, the caching is busted on my end as well. In the meantime you can restrict the tests to just the test you're interested in by adding ".only" after the "it" or "describe" functions. So, "describe(foo)" would become "describe.only(foo)". The drawback is that it's important to remember to take this out before running the final test run [22:30:01] before the commit or all the other tests would be skipped by CI and others as well. [22:30:58] bearND: ok cool. thanks for checking it out! [23:17:27] coreyfloyd: forgot to mention this earlier. The announcement change has been deployed. [23:31:50] bearND: omw [23:32:04] cool [23:33:14] tgr, hey yt? Can you link to a documentation on the order of wikitext -> HTML transformation? [23:33:33] tgr, are parser functions invoked before other stuff? [23:33:53] similar to templates being expanded before parsing the whole page [23:34:52] bmansurov: not sure if that's well documented anywhere [23:35:12] pres-save transform, then transclusions, then parsing AFAIK [23:36:09] tgr, thanks [23:36:47] if you want a definitive answe probably best to ask someone from the Parser team [23:37:49] tgr, ok. I'm writing up what I found so far regarding https://phabricator.wikimedia.org/T151276, and hopefully someone will correct me if I made a mistake. [23:37:50] it can be more complicated than that, such as xml tag style extensions returning content that needs to be expanded again [23:38:17] tgr, does https://phabricator.wikimedia.org/T151276#2836821 sound about correct? [23:41:50] bmansurov: huh, that sounds like a fun bug [23:42:15] yeah, -style parser extensions run out of order [23:43:01] the parser processes them before other stuff, then replaces them with a placeholder token, then does normal processing, the restores the placeholders [23:43:17] xml-tag-style extensions which return raw HTML, that is [23:43:41] not sure about how #tag affects that whole situation [23:44:34] ok thank you [23:44:50] I think #tag moves stuff to a later processing phase, at the same time as templates/parserfunctions [23:44:58] but that's still before image processing is done [23:45:13] i see