[04:44:44] Krinkle: Do I remember seeing a commit from you somewhere recently stopping/limiting the versions of HHVM we're testing against on travis? [04:45:10] https://github.com/wikimedia/mediawiki/commit/c7d0fa0ee999fb26823c2873f56c69a867fe9d1b [04:45:13] Hmm. Not quite [04:58:48] James_F: Ah yes [04:58:48] https://hhvm.com/blog/2019/02/11/hhvm-4.0.0.html [04:58:53] HHVM 4.0.0 was released today [04:58:56] This explains it exactly :) [04:59:05] Helpful. :-) [04:59:21] This is of course why we’re abandoning ship. [04:59:25] heh [04:59:44] I wonder if it's something to ask legoktm nicely to add into library upgrader rather than me doing the rest of these repos manually [04:59:53] Or just fix as/when they're a problem [05:00:10] "HHVM is no longer able to execute Composer; Composer 1.8.4 includes support for identifying the HHVM version when executing via PHP, which is how we currently recommend installing dependencies for Hack projects. [05:00:10] We are currently working to move to a package manager that fully supports multiple languages." [05:00:12] oh lord [05:00:17] I’d rather we work on dropping HHVM entirely quicker. :-) [05:00:37] Send decent coffee to the italians [05:00:49] Though, it did show something else... [05:00:52] (Not throw shade at the wonderful SRE.) [05:01:02] We've left PHP <= 5.6.99 on various repos [05:01:10] But bumped codesniffer to a version that needs >= 5.6.99 :( [05:01:36] Oops. Should have been >=5.6.99. [05:01:52] Which is why I did https://gerrit.wikimedia.org/r/#/c/php-session-serializer/+/489948/ [05:02:12] I guess that's a bigger issue if it's wider spread :) [05:02:46] https://travis-ci.org/wikimedia/xmp-reader/builds/491982406 [05:02:54] ^ it's had the intended resolution at least [05:04:31] https://codesearch.wmflabs.org/deployed/?q=%3D5%5C.6%5C.99&i=nope&files=&repos= [05:04:37] I don’t see any… [05:04:52] "php": ">=5.5.9", [05:06:19] There's quite a few of those :) [05:06:22] Yeah, a bunch of extensions will need to get bumped per https://codesearch.wmflabs.org/deployed/?q=%3D5%5C.5%5C.&i=nope&files=&repos= but it’s not on fire, I think. [05:06:44] Nah, it'd probably only be noticed if they're actively running travis [05:08:57] https://phabricator.wikimedia.org/T215862 filed for that too [05:09:28] The other q is how quickly we’re going to drop 7.0.x. :-) [05:09:47] So many tasks to file and fuss over. [05:09:49] heh [05:16:04] <_joe_> hey hey why I am now called "the italians" [05:16:10] <_joe_> :P [05:16:43] _joe_: Because your awesomeness is so great it seems like it belongs to multiple people. :-) [05:16:45] <_joe_> we should limit to HHVM < 3.22 [05:17:14] Are we likely to upgrade 3.18 -> 3.22? [05:17:15] We’re just limiting directly to HHVM 3.18. [05:17:32] <_joe_> Reedy: I'm more likely to convert to american coffeee [05:17:41] It's in various libraries MW itself uses, rather than MW itself [05:17:46] So pinning to 3.18 can stay. [05:17:49] <_joe_> James_F: that's fair, ok [05:18:06] Timo has done 3.18, 3.21 and 3.24 in mw core on travis [05:18:32] We should probably drop the extra testing though. [05:19:06] <_joe_> since we have no idea how many people are actually using HHVM with MW [05:19:22] <_joe_> it's probably fair to wait until we drop HHVM completely [05:19:36] <_joe_> and yes, it was 3.24, not 3.22 [05:37:00] James_F: That's another 7 repos I noticed failed for varying reasons on travis on their last runs :( [05:38:50] Thanks :) [05:39:46] yay for cherry picking across repos to speed up the copy pasta [05:40:29] Reedy: So many semver breaking changes and without even dropping HHVM. :-( [05:40:50] Well, only if we do releases [05:40:56] Many of them we might not need to [05:41:03] But tests not failing on HEAD is nice [15:38:56] James_F: I think I just need to make the cleanup/update patches for at-ease :) [15:39:09] REL1_32 coming out, REL1_30 going away was the blocker [15:39:39] Reedy: Oh, OK, couldn’t easily see when we switched to 1.2.0. [15:39:57] I checked their policy with Niklas then looked at older branches in GH [15:40:15] Right. [15:40:30] OK, but we need to actually write those patches. :-) [15:41:38] I've got ~5 hours on a plane with wifi tomorrow [15:41:50] Adn we've got a few months, so not a major rush [15:43:16] OK, WFM. [17:37:07] Interesting [17:37:07] https://github.com/Wikia/app/pull/16677 [17:37:18] Wikia are basically about to upgrade to 1.32 [17:38:15] oh, nvm [17:40:30] lulz [17:41:02] Reedy got trolled [17:41:13] I know they are working on it :P [18:48:42] Kibana. You enter a query and press return. It now search the past 15 minutes, while it's busy you change the duration to past 30 days. It now (unbeknownst to you) searches the past 30 days for without query, because it amends the last completed query. Then, you press the submit button explicitly on the form to make sure it really does the right thing. Unfortunately, pressing submit while a query is running means "abort current query". [19:20:45] anomie: Did you want https://gerrit.wikimedia.org/r/c/mediawiki/extensions/JsonConfig/+/488996 deployed? [19:26:29] James_F: The patch should be rolling out with the train this week now, so I'm not sure how much it matters. But go ahead if you want. [19:28:07] anomie: I don't know how to test it, sorry. [19:42:47] go core [19:42:53] (forgot the /)