[02:34:25] ori: does that mean you new version of your patch is the same for small wikis in the default config as it is for Wikimedia? [04:12:04] robla: yes [04:12:18] cool! [04:12:44] yeah, i forfeited a little bit of the possible performance gains but made it suitable for the general case [04:13:45] yup, that makes sense. Tim made a very similar tradeoff a few months ago for an HHVM upstream patch he made (I'm struggling to remember the details of) [04:15:17] * robla seems to recall talking about the problem in depth with ori in a ride to dinner in San Diego [04:16:56] regexp caching....I think [04:28:12] * ori wonders if TimStarling would be willing to merge [04:31:27] http://au.idigitaltimes.com/malware-found-pre-installed-xiaomi-huawei-lenovo-phones-107190 [04:32:17] i thought lenovo learned something from superfish, guess not [04:35:40] ori: I imagine they did learn "something", but probably not enough, and probably not even the right lesson [16:47:10] ori: re https://gerrit.wikimedia.org/r/#/c/236814/ : I think you may have to bump the lib in operations/mediawiki-config too to get the new version of the classes to actually be used in WMF production [16:48:16] * bd808 now sees that you did [17:07:32] tgr|away: stuck on your commute? [17:10:38] bd808: good seeing you in SF, btw [18:00:01] tgr|away: if you have things for SoS get them to anomie today [18:00:59] ori: good to see you too. thanks for the new hoodie [18:02:49] ori: you might want to take a look at https://phabricator.wikimedia.org/T111836 [18:04:20] yah, looking [18:06:54] no SoS stuff, ori was kind enough to review the sentry patch [18:09:22] bd808: anomie: can one of you review https://gerrit.wikimedia.org/r/#/c/222079/ and https://gerrit.wikimedia.org/r/#/c/236225/ ? I would like to get the HTTPS stuff wrapped up [18:09:41] * anomie will look [18:10:17] i read through both yesterday and didn't see anything too crazy but hadn't tried to actually test [18:10:47] the line length for comments in the first patch annoyed me and there are a couple of # comments [18:11:30] are we supposed to use //? [18:12:59] I think so, yes [18:13:26] No specification as to which comment marker might be preferred in [[mw:CC/PHP]] [18:14:19] yeah, you're right. (/me hates seeing # comments anyway) [18:15:01] only mention is "In comments there should be one space between the # or // character and the comment." [18:15:16] which is agnostic of the line comment indicator [18:37:05] bd808: how can I disable auto puppet runs on a self-hosted puppet master in labs? [18:37:51] I am debugging an issue where the first puppet run fails but the second succeeds and that's pretty much impossible with 5 min auto runs [19:02:13] tgr: hmmm... not sure. You can disable puppet entirely with `puppet agent --disable` but that will stop you from running manually too. [19:02:46] I guess you could patch locally to remove the cron job [19:24:42] csteipp: I was listening to http://www.infoq.com/presentations/security-etsy ("Crafting an Effective Security Organization") on the way to work today. It's pretty good. They have 14 FT security engineers -- one can dream, right? [19:25:28] I can dream [20:15:54] tgr: is https://gerrit.wikimedia.org/r/#/c/199598/ ready from your perspective? [20:31:30] ori / Krenair, either of you deploying right now? [20:31:39] i'm about to push out a change [20:31:41] give me 2 mins? [20:31:54] ori: no problem. let me know when you're done! [20:31:56] csteipp, not me [20:32:03] * Krenair logs out [20:32:12] just had a file open in the background [20:47:45] ori: almost, it needs two puppet runs to set up everything successfully [20:47:53] I'm trying to track that down [23:30:55] legoktm: I wonder what jenkins is up to on https://gerrit.wikimedia.org/r/#/c/236857/2