[00:55:43] _joe_: You have fan mail: https://www.mediawiki.org/wiki/Topic:Usx7uerq380mzwq3 [11:25:12] <_joe_> ok I can confirm the issue with the error page handling with php7 - see https://en.wikipedia.org/wiki/Special:ProtectedPages?limit=1000 [11:26:20] <_joe_> with php, you get back a text-only "MediaWiki Internal error" [20:29:50] tgr: do you plan on looking into https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/475065/ sometime? [20:39:10] AaronSchulz: at some point but not soon [20:39:28] TBH I feel that choosing the structure test approach was a mistake [20:40:24] determining in a structure test that's run with a MySQL setup whether the Oracle update scripts are safe is going to be very fragile [20:43:42] tgr: maybe the LoadExtensionSchemaUpdates hook call could be wrapped in calls that disable/re-enable queries to $db [20:43:45] * AaronSchulz shrugs [20:44:56] I don't see any need for things to use the $db to decide what updates to add to the array (they should just be idempotent callbacks) [20:47:56] well at a minimum $db needs to be used to determine the engine type (and so which update file to run) [20:49:12] oh, nvm, you mean SQL queries, not calls to the $db object [20:49:29] yeah that's probably a better approach [20:49:55] there is already a readonly mode for DB handles, right? [20:50:20] I can look into that when I have some free time [21:07:28] TimStarling: Have you also experienced that when using a mwdebug server via XWD that initially the first one or two requests seem to always time out (60s) before it starts working? [21:07:46] I find this somewhat concerning, and this didn't used to be the case ~ 6 months ago. [21:07:53] (HHVM, default settings) [21:08:06] I wonder what changed. [21:09:45] I have also seen that a couple times (not consistently, I think) [21:19:59] I can't get anything to load currently in my main browser. [21:20:10] Been trying to view https://meta.wikimedia.org/wiki/User:Krinkle/global.js 5 time now, keeps timing out [21:22:36] when it happened to me, it happened again after every scap pull, consistently [21:22:50] but at other times when I did SWAT it didn't happen [21:23:58] Hm.. [21:24:20] My issue also seems strangely specific to one of my Chrome windows. And only when loading the page as view-source. [21:24:37] Meh, nevermind, must be Chrome related. Although only happens when XWD is on.