[15:36:22] Trivial: https://gerrit.wikimedia.org/r/#/c/315841/, also easy: https://gerrit.wikimedia.org/r/#/c/315842/ [15:43:54] Reedy: thx on the former [15:45:04] I'll merge the other in a min [15:45:09] tyvm [15:45:13] just wanted to hit rebase and save a merge commit [15:45:44] oh, no rebase available, thanks gerrit [18:11:52] Does anyone know exactly which subclass of BagOStuff is returned on production when you call MediaWikiServices::getInstance()->getLocalServerObjectCache() ? [18:12:10] Hmmm I suppose I could just try calling it... [18:14:53] Try it in eval.php [18:15:15] > echo get_class( MediaWikiServices::getInstance()->getLocalServerObjectCache() ); [18:15:15] PHP Fatal error: Class 'MediaWikiServices' not found in /srv/mediawiki-staging/php-1.28.0-wmf.22/maintenance/eval.php(78) : eval()'d code on line 1 [18:15:18] Heh, I did [18:15:57] Oh, everything got namespaced. [18:16:04] And eval is in global ns. [18:16:07] Fun [18:16:20] \ [18:16:21] Er, lotsa stuff namespaced. [18:16:56] Eh, cli lies. [18:17:02] hmm eval.php [18:17:03] It uses EmptyBagOStuff [18:17:04] at least e don't have mucha nested namespaces [18:17:10] > echo get_class( MediaWiki\MediaWikiServices::getInstance()->getLocalServerObjectCache() ); [18:17:10] EmptyBagOStuff [18:17:18] (other than wikibase) [18:17:27] I was just trying mwrepl [18:18:39] ostriches: [18:18:40] hphpd> echo get_class( MediaWiki\MediaWikiServices::getInstance()->getLocalServerObjectCache() ); [18:18:42] echo get_class( MediaWiki\MediaWikiServices::getInstance()->getLocalServerObjectCache() ); [18:18:44] APCBagOStuff [18:19:09] it should be APC [18:19:43] Yeah. mwrepl gives better answer :) [18:21:36] I wonder does mwrepl actually run in a web server context like a web request? [18:21:42] thx btw!!!! [20:02:40] ori: the FF x-wm-debug extension's logstash link works again. I think this PR will fix the Chrome version -- https://github.com/wikimedia/ChromeWikimediaDebug/pull/5 [20:03:15] cool, I'll test it now [20:08:04] works [20:19:52] bd808: if I create a Google Group, I could use it to give you publishing rights to the extension. Do you mind if I do that? I think I would need to use a non-WMF Google account, though -- it's not happy with Google Apps e-mails [20:20:43] sure. you can add "casadebender" [20:21:03] they won't let me make a "bd808" user account. too short [20:21:34] lame [20:23:14] invitation sent [20:24:32] ^ bd808 [20:24:43] ori: thanks. looking for it now [20:26:15] accepted [20:26:56] Transferring item WikimediaDebug to group publisher chrome-wikimedia-debug (chrome-wikimedia-debug@googlegroups.com) failed. Please try again later. [20:27:51] I tried again and got: 400. That’s an error. The requested URL was not found on this server. That’s all we know. [20:27:55] not even the right status code [20:28:27] oh, it seems to have worked anyway [21:46:06] hey! anyone feel like lending a hand with https://phabricator.wikimedia.org/T144952? 'specially someone knowledgable about ObjectCache implementations and config? Kinda stuck still, thx in advance!! [21:56:12] legoktm: ^ ? [21:58:18] What exactly is the question> (I'm also at offsite so also kind of busy) [21:58:23] ?* [22:02:08] legoktm: ah sorry I didn't know... mmm it's an inconsistent error in MessageCache, where i18n messages that are really CN banners that are really Mediawiki: namespace pages appear as nonexistent when retrieved as messages on certain hosts at certain times [22:02:40] if u look at the bug, maybe start with the second-to-last comment and work backwards... [22:03:06] The last patch-for-review might be a fix but we don't truly know [22:03:12] thx!!! [22:04:54] even rougher notes here: https://etherpad.wikimedia.org/p/CentralNotice_translation_magick [22:24:21] AndyRussG: well, iirc Krinkle investigated that when RL messages started disappearing, though I don't remember if that was a RL issue or MessageCache [22:27:26] legoktm: hmmm K :) [22:28:41] thx :)