[15:38:20] phuedx: i removed the old recurring bi-weekly series (the one in the cal that would be right now and conflict with early backlog grooming), in favor of the updated tri-weekly meeting you arranged with me. [15:46:01] dr0ptp4kt: 👍 [15:49:13] niedzielski: o/ [15:49:19] I'm having some trouble running MockWebServer tests... [15:49:25] in fact I'm not sure how they're supposed to work, after the cache upgrade. [15:49:53] dbrant: \o in Android Studio or the command line? [15:49:58] either [15:51:37] dbrant: that's weird. checking on the caching. i thought it was disabled for MockWebServerTest [15:52:36] i may be wrong, but it looks like it's because OkHttpConnectionFactory has a static Cache object that requires a WikipediaApp instance to be initialized. [15:52:59] dbrant: i fixed the MockWebServer caching issue a few days ago: https://gerrit.wikimedia.org/r/#/c/347084/1/app/src/test/java/org/wikipedia/test/MockWebServerTest.java . i think you just need to rebase your patch [15:54:31] dbrant: mot MockWebServerTests fail in my version of Android Studio lately due to T161161 (details on a workaround inside) [15:54:31] T161161: Robolectric tests that require raw resources fail in Android Studio - https://phabricator.wikimedia.org/T161161 [16:15:51] niedzielski: hmm, in that case it must be a Windows-specific issue. The tests work on my mac... [16:16:17] dbrant: what is the error you're seeing? [16:17:51] niedzielski: https://pastebin.com/ac42vVs9 [16:21:15] dbrant: i get that error in android studio when i don't use the two workarounds in T161161 [16:21:16] T161161: Robolectric tests that require raw resources fail in Android Studio - https://phabricator.wikimedia.org/T161161 [16:21:34] (on linux) [16:22:30] dbrant: so i just have those two lines changed and it AnnouncementClientTest seems to work fine then [16:35:35] niedzielski: yeah, with the workarounds I still have a bunch of failures (running from the command line, on Windows). [16:36:37] dbrant: sorry, to be clear the workarounds break command line usage but fix android studio usage. the tests should pass from the command line without any changes but it sounds like that's not the case [16:55:33] it looks like robolectric is definitely reading AndroidManifest correctly, but then crashes when trying to parse resources (on windows). Anyway... i'll just use the mac for tests for the time being. [16:55:55] i haven't actually run the tests on Windows in quite a while, so I'm not sure when the last-good run was. [17:04:57] that's lame :/ i believe i have a w7 and xp vm i can test on if you want help debugging later. if you haven't already, maybe try not modifying robolectric.properties since it's reading the manifest properly but do change the TestFileUtil since it can't read the resources [17:10:52] (when I say resources I mean things like strings.xml (not even talking about raw resources yet)) [17:27:12] geez [17:30:39] coreyfloyd: bearND tgr|away i'll be late. start without me [17:30:55] k [19:25:11] bearND: ok, i see /feed/trending/edits on the REST API spec. i imagine it's not cached in RESTBase, right? otherwise it wouldn't be up-to-the-minute as it's intended to be. [19:27:59] mdholloway: AFAIU trending edits results are cached for 30 minutes but not stored in RB Cassandra (cc: Pchelolo in case I misspoke) [19:29:17] bearND: Pchelolo: cool. yeah, i imagine that endpoint's responses are cached in varnish (30m) as usual but not RB/Cassandra [19:29:50] bearND: that's correct. Only Varnish for 30 mins, no RB storage [19:33:49] niedzielski: cool, looks like switching from +2 -> -1 blocks the merge after all even though CI keeps running. [19:34:06] phew!