[01:44:23] /15 [13:16:16] Krinkle: It could probably be done, although the little-used ApiPageSet::getDataSource() might have to change. It seems like it could be a bit confusing for humans looking at calling code though. Is there a use case that would need it? [15:02:06] anomie: https://gerrit.wikimedia.org/r/#/c/427531/12/resources/mobile.watchstar/WatchstarGateway.js [15:14:53] Krinkle: I guess you're pointing to the new "loadWatchStatus" function in there, beginning on line 66? Why is it bothering to query some by page ID when it already has all the titles? [15:37:32] anomie: I don't know sorry :) [18:05:31] So, I'm going to go ahead and cut rc.0. I'm not happy with it, but we need something out [18:28:13] Really.... https://gerrit.wikimedia.org/g/mediawiki/core/+/REL1_31/RELEASE-NOTES-1.31#133 [19:31:43] no_justification: Two bug fixes seems a little light, yes. [19:51:24] :/ [19:54:32] hopefully more than that was fixed, just wasn't reported in the release notes? there should be a way to look at phab for all tasks closed within some period of time as resolved [19:55:55] https://github.com/wikimedia/mediawiki/compare/REL1_30...REL1_31 [19:55:57] 3422 commits [20:06:41] k, time to do some scripting [20:08:54] I'm sure no merges removes a few commits :P [20:09:53] well I was just going to scrape all the commit messages for those for Bug: TXXXXXX [20:10:10] then correlate that with task numbers in the release notes, and show all the tasks that were mentioned in commits but are not mentioned in release notes [20:12:21] Sounds useful for starters :) [20:17:30] once there's a list, we'll probably need to manually filter out any regressions that happened in the 1.31 dev cycle itself [20:26:20] https://phabricator.wikimedia.org/maniphest/query/GLtvbEO0U7vc/#R [20:26:35] Skizzerz: ^ [20:31:33] Hmm. When I run "npm update" or "npm install", it errors out complaining that "'babel/core' is not in the npm registry" and "It was specified as a dependency of 'postcss-html'" [20:31:47] greg-g: also https://phabricator.wikimedia.org/maniphest/query/EtujXP_8iPyd/#R -- forgot we had a bot for that [20:32:12] although need to strip out non-core projects from that list [20:35:44] Skizzerz: yeah. and is the only diff between that query and mine you limited to "resolved" instead of "any closed status"? [20:36:07] yeah [20:36:17] figured "resolved" is more indicative of what actually made it in [20:36:41] if it was closed otherwise, it could've had a patch merged and then reverted, for instance (so the bot would still tag mw-1.31-release-notes) [20:37:24] hmm, my script to parse the logs doesn't handle reverts either [20:46:58] James_F: https://gerrit.wikimedia.org/g/translatewiki/+/master/TranslatewikiSettings.php#35 [20:48:04] MaxSem: OK? Tell Niklas, not me. I can't fix third party installs for them, even if they're third party installs we like. [21:13:36] AaronSchulz: just make sure you saw https://phabricator.wikimedia.org/T193668 (train blocker due to spamming logs) [21:17:53] greg-g: a bit distracted at the moment. Something like https://gerrit.wikimedia.org/r/#/c/430273/ might help. I was looking into this area yesterday due some wikibase test failures. [21:18:20] basically complex callbacks from onTransactionIdle to DeferredUpdate callbacks [21:38:23] no_justification: I made https://gerrit.wikimedia.org/r/#/c/430493/ if you're interested. [21:40:19] James_F: I already tagged rc.0 earlier :) [21:40:36] https://dpaste.de/tOVi is the list that I generated per above (looking at tasks mentioned in Bug: in commits that weren't mentioned in RELEASE-NOTES-1.31) [21:41:26] Ouch [21:41:48] no_justification: I saw. But it doesn't clash (yet). [21:42:17] It'll land in rc.1 :) [21:42:19] no_justification: There'll probably be an rc.1 anyway… [21:42:23] Yeah. [21:42:30] ofc there will [21:42:32] rc.0 is shi [21:42:33] *shit [21:42:46] It's why I've been delaying [21:43:02] well on the plus side, there's probably already more than 2 whole bugs that got fixed [21:43:11] on the downside, it's going to be a lot of effort to figure out what all the other ones are :( [21:44:12] if someone wants the script's source, poke me and I'll pastebin that too [21:45:38] Skizzerz: wanna put that on a wiki page so we can collaboratively go through that list? [21:46:19] sure, mw.o or somewhere else? [21:46:52] mw.o yeah [21:47:08] k [21:47:09] sec [21:48:49] Oh phooey, bad link in e-mail: https://lists.wikimedia.org/pipermail/mediawiki-announce/2018-May/000219.html [21:48:54] But otherwise, announced [21:48:58] cc James_F, greg-g ^ [21:49:24] no_justification: Yay. [21:51:23] legoktm: https://www.mediawiki.org/wiki/Release_notes/1.31/additions [21:52:03] Skizzerz: ty, I'll look through it in a bit [21:52:44] no_justification: Can I drop RELNOT1.31 from master so that people sync properly? Or do we still want it for backports? [21:53:03] Um, I prefer doing changes directly to REL1_31 then forward-port to master anyway [21:53:25] OK, legoktm/Skizzerz, that means you. :-) [21:53:25] (which is an undocumented workflow of mine, obvs) [21:53:38] uh [21:53:50] normally we drop 1.31 rel notes from master once the final release is done? [21:55:07] hmm [21:55:11] I maybe messed up log command [21:55:16] might get a shorter list [21:57:20] ok, I fixed the command and got a whole 2 bugs out of the list [22:19:06] So far, I've identified on commit ( Ib15fcc1b1b ) that touches the release notes for 1.31 but isn't backported. [22:44:18] Skizzerz: https://gerrit.wikimedia.org/r/430508 https://gerrit.wikimedia.org/r/430509 https://gerrit.wikimedia.org/r/430510 as the first few. :-) [22:45:44] nice! [23:30:59] no_justification: cache/.htaccess isn't in the tarball? [23:31:42] Bleh. [23:32:00] Overzealous exclusion list [23:32:29] Easy fix for rc.1 [23:36:28] no_justification: also it's not using mediawiki/vendor...? [23:36:51] I get one answer from one person, one from another [23:36:58] I thought we *removed* the dang submodules? [23:37:04] In favor of composer for tarballs? [23:37:12] * no_justification gives up [23:37:22] it's not supposed to be a submodule (well I don't really have a strong opinion on that) but it is supposed to be in the tarball [23:38:13] https://phabricator.wikimedia.org/T185618#3954292 [23:38:20] > but we will filter down the list of installed libraries to just what is required by everything distributed by the tarball [23:38:30] I think we still need to do that, but it does need to use mw/vendor [23:39:18] Why not do a submodule then? This just complicates it [23:40:25] I'm tired of caring about this stupid tarball [23:40:43] if it's a submodule, can you still use composer normally? i.e. if you don't check out/init the submodule can you still use the vendor directory without causing local changes? [23:40:57] (for users cloning via git) [23:41:13] One or the other folks. Both just confuses it. [23:41:34] But cake! And eating it! [23:41:51] I'm done after MediaWiki 1.31. Somebody else should do releases. [23:42:13] * legoktm hugs no_justification [23:43:41] the end result we want is that people who check out via git use standard composer to get dependencies. And people who use the tarball have dependencies from mediawiki/vendor included [23:47:28] Well, that requires some code changes :)