[00:29:43] <James_F>	 Do we want https://gerrit.wikimedia.org/r/#/c/393929/ to land for REL1_31? If so, who's going to merge it?
[00:37:55] <James_F>	 TimStarling: Did you see on https://github.com/facebook/hhvm/pull/8139 that HHVM want you to re-write your patch to match their line length standard and then they'll merge it?
[01:31:43] <Krinkle>	 anomie: btw, did you see the ini_set error on PHP 7.2? https://travis-ci.org/wikimedia/mediawiki/jobs/368430010 
[01:31:54] <Krinkle>	 Looks like it doesn't have a test yet, but might affect SessionManager testing 
[01:32:04] <Krinkle>	 Currently warns from PHPSessionHandlerTest
[03:12:59] <legoktm>	 hi
[03:13:06] <legoktm>	 I suggested php >= 5.6.99
[03:13:14] <legoktm>	 that'll work with HHVM and require PHP 7
[10:45:09] <Reedy>	 https://gerrit.wikimedia.org/r/#/c/427633/
[15:22:06] <anomie>	 Krinkle: I wonder why PHP 7.2 started doing that. Especially since session.use_trans_sid doesn't actually depend on headers.
[15:25:02] <anomie>	 Anyway, we could wrap those calls in `if ( !headers_sent() ) {}` easily enough, or use at-ease to shut it up.
[16:53:56] <Krinkle>	 anomie: I take it the actual behaviour is unaffected? Or maybe the tests pass even if the settings fail and/or not needed?
[16:55:12] <Krinkle>	 Reedy: legoktm: Nice. it's a version number that cannot be reached by Zend PHP except if 7.0. Nice
[16:55:34] <Reedy>	 or 5.6 becomes a security clusterfuck :P
[16:55:55] <Reedy>	 they're already on .35
[16:57:15] <anomie>	 Krinkle: No idea, really. The actual cookie-related one makes no difference in CLI mode, and the other is a not-really-recommended setting anyway AFAIK.
[16:57:34] <Krinkle>	 Reedy: He, true
[16:57:56] <Krinkle>	 They might reach .99 before this Christmas (at EOL)
[19:20:58] <Reedy>	 Why is phpcs failing on REL1_28?
[19:20:58] <Reedy>	 https://integration.wikimedia.org/ci/job/mediawiki-core-phpcs-docker/8441/console
[19:23:15] <Reedy>	 same line is fine on master
[20:24:59] <anomie>	 Reedy: Master uses a newer version of phpcs that's smart enough to realize that long URLs can't be broken?
[20:25:09] <Reedy>	 pr
[20:25:12] <Reedy>	 maybe, yeah
[20:25:52] <Reedy>	 or, our codesniffer package
[20:25:57] <anomie>	 https://github.com/squizlabs/PHP_CodeSniffer/commit/d21a61ab12ddf08f940c90bf6e991bff184fb439
[20:27:01] <Reedy>	 though, it uses the same version as 1.27, which passes
[20:39:15] <anomie>	 Reedy: It looks like that's because you backported Ib47ffed117 to 1.27, 1.29, and 1.30, but not 1.28 for some reason.
[20:40:36] <Reedy>	 Not supported?
[20:53:24] <Reedy>	 looks like mw-cs did something
[20:53:25] <Reedy>	 https://github.com/wikimedia/mediawiki/commit/255d76f2a13a8378ded9f0cf1c2bb172f7f07a5b#diff-4df539c3ace6af2cbe7f857c2cfe4c18
[20:53:44] <Reedy>	 /brought in that patch you linkded
[21:16:05] <Reedy>	 heh, MatmaRex is readying is PHP 5.6+ all the things
[21:20:28] <hashar>	 well it would be nice to not intentionally break php5.5 / hhvm support in master until 1.31 is released in a month
[21:20:42] <hashar>	 that is going to make backport slightly harder when they encounter new code
[21:22:08] <Reedy>	 1.31 is now php 5.6.99+ :P
[21:22:44] <Reedy>	 hhvm won't be going away for a while
[21:22:49] <Reedy>	 5.5 goes away when ops get done
[22:02:04] <Krinkle>	 1.28 is EOL, right?
[22:02:08] <Krinkle>	 Yeah
[22:03:30] <Reedy>	 j
[22:03:32] <Reedy>	 ja
[22:04:10] <Reedy>	 Any reason jenkins still runs test for it though?
[22:14:05] <James_F>	 Do we still run tests for REL1_26?
[22:20:34] <Reedy>	 god knows
[22:21:32] <Reedy>	 Can we get rid of .php5 files when we only have PHP7 support?
[22:21:59] <Reedy>	 James_F: Yes
[22:22:01] <Reedy>	 427823 is in zuul atm
[22:22:26] <James_F>	 Reedy: Yes, .php5 support is partially removed in my patch.
[22:23:45] <Reedy>	 MW doesn't care :P