[01:27:21] <legoktm>	 TimStarling: ParserTestTest?
[01:28:00] <TimStarling>	 we do need ParserTestRunnerTest, I have some more concrete ideas for that
[01:28:02] <legoktm>	 (we do have a MediaWikiTestCaseTest)
[01:28:21] <TimStarling>	 things like making sure certain kinds of data doesn't leak beyond teardown
[01:56:10] <legoktm>	 TimStarling: do you think you could take a look at https://gerrit.wikimedia.org/r/#/c/310306/ and see if you think the architecture is a good direction?
[01:59:03] <TimStarling>	 good commit message
[02:05:53] <legoktm>	 oops yeah, that was 7am after an all nighter -.-
[02:08:46] <TimStarling>	 looks a bit slow, this is (hopefully) quite a hot path
[02:09:26] <TimStarling>	 I mean hopefully as in, hopefully the rest of the link construction stuff is not far far slower
[02:11:21] <TimStarling>	 I think I'd be happier if your dispatcher was less generic, like an array of handlers per namespace and an array of handlers for all namespaces
[02:12:45] <legoktm>	 hmm okay. I was also considering having shouldHandle() return self::EXISTS for the super fast ones to be the equivalent of Title::isAlwaysKnown()
[02:14:01] <legoktm>	 are you thinking that the namespace separation would reduce the overhead of all the shouldHandle() calls?
[02:14:06] <TimStarling>	 yes
[02:14:51] <TimStarling>	 or maybe benchmark it to see if this is a rational concern
[02:15:26] <TimStarling>	 but the rule of thumb I usually use is that you pay per function call
[02:15:44] <TimStarling>	 you can basically count the function calls as a performance estimate
[02:17:12] <legoktm>	 alright. does isset() count as a function call?
[02:17:24] <TimStarling>	 no
[02:19:19] <TimStarling>	 so there's 6 subclasses, so 6 shouldHandle() calls for add(), then exists() make 6 calls of shouldHandle() and 6 of exists(), so we're up to 18
[02:22:25] <legoktm>	 ok, I should be able to cut down the shouldHandle() calls
[02:22:53] <legoktm>	 there are also going to be 2-3 more subclasses added by extensions that currently use the TitleIsAlwaysKnown hook
[02:26:25] <TimStarling>	 yeah, all the more reason to make it O(1) in the number of extensions
[02:28:16] <TimStarling>	 that's the difference between bloated and fully-featured, right? whether or not an application has a latency proportional to its number of lines of code
[02:28:48] <legoktm>	 heh
[05:26:41] <AaronSchulz>	 TimStarling: https://gerrit.wikimedia.org/r/#/c/310731/
[05:29:44] <TimStarling>	 what was it previously used for?
[05:32:42] <TimStarling>	 LBFactory::disableBackend()
[05:36:12] <TimStarling>	 which is itself unused in core, btw, now the installer calls MediaWikiServices::disableStorageBackend directly
[05:54:15] <AaronSchulz>	 TimStarling: https://gerrit.wikimedia.org/r/310536 
[07:33:04] <legoktm>	 TimStarling: I'd appreciate your help with figuring out how to benchmark this, and how to interpret the results... I tried using enwp's Australia with templates + InstantCommons and ended up with https://paste.fedoraproject.org/428247/92468114/raw/ (--reset-linkcache is introduced in https://gerrit.wikimedia.org/r/310743)
[07:34:23] <legoktm>	 I also rewrote the patch based on what you suggested and implemented a Title::isAlwaysKnown() like shortcut for the faster existence checks
[07:36:16] <TimStarling>	 suspiciously fast
[07:37:06] <TimStarling>	 I can't really discuss it in detail right now because I have a very short window of overlap with the europeans in #mediawiki_security
[07:37:40] <legoktm>	 ok, tomorrow or monday is fine
[16:41:47] <legoktm>	 ostriches: ping on https://gerrit.wikimedia.org/r/#/c/309519/
[16:43:20] <ostriches>	 oh yeah
[16:46:14] <legoktm>	 thanks :) I also have https://gerrit.wikimedia.org/r/#/q/topic:linker-formatsize to deprecate Linker::formatSize() which is a dumb function
[16:56:59] <legoktm>	 ostriches: did you have a dashboard for the globaltitlefail log?
[17:01:27] <ostriches>	 legoktm: no cuz it doesn't go to logstash
[17:04:54] <legoktm>	 bummer
[18:47:54] <hashar>	 legoktm: around? Do we have a way to resubmit the LocalRenameUserJob jobs ?
[18:48:04] <legoktm>	 hi
[18:48:11] <hashar>	 there is a few being blocked on mw.org rename due to https://phabricator.wikimedia.org/T145596#2640418
[18:48:14] <legoktm>	 fixStuckRenames.php in CentralAuth
[18:48:22] <hashar>	 will try thanks !
[18:51:57] * hashar cherry picks
[18:59:48] <hashar>	 legoktm: so gotta run the script against each wiki that got the stuck job right ?
[19:00:10] <legoktm>	 yeah
[19:00:48] <hashar>	 will do the grunt work :]  thx
[20:08:49] <Krinkle>	 I guess https://phabricator.wikimedia.org/T41473 can be closed now?
[20:08:52] <Krinkle>	 ( TimStarling  )
[20:18:23] <SMalyshev>	 anybody knows how I can get comments that are in file history?
[20:19:09] <Reedy>	 ?
[20:19:15] <Reedy>	 From where?
[20:20:02] <SMalyshev>	 Reedy: if I have File object
[20:22:33] <Reedy>	 $file->getDescription( File::FOR_THIS_USER, $user )
[20:22:42] <Reedy>	 That's vaguely what's used on an image page
[20:23:06] <Reedy>	 		// Don't show deleted descriptions
[20:23:06] <Reedy>	 		if ( $file->isDeleted( File::DELETED_COMMENT ) ) {
[20:23:06] <Reedy>	 			$row .= '<td><span class="history-deleted">' .
[20:23:07] <Reedy>	 				$this->msg( 'rev-deleted-comment' )->escaped() . '</span></td>';
[20:23:07] <Reedy>	 		} else {
[20:23:09] <Reedy>	 			$row .= '<td dir="' . $wgContLang->getDir() . '">' .
[20:23:11] <Reedy>	 				Linker::formatComment( $description, $this->title ) . '</td>';
[20:23:13] <Reedy>	 		}
[20:23:20] <Reedy>	 SMalyshev: Look at ImageHistoryList.php
[20:24:10] <SMalyshev>	 Reedy: aha, thanks, this looks like what I need!
[21:21:30] <legoktm>	 Krinkle: closed
[21:38:33] <ostriches>	 legoktm: https://phabricator.wikimedia.org/P3984 went from 101 -> 94 entries (since last week)
[21:38:44] <legoktm>	 yay!
[21:39:17] <legoktm>	 do you want to start filing bugs for those? some of them look pretty easy
[21:40:04] <ostriches>	 Lots of them are probably unmaintained core crap :p
[21:41:04] <legoktm>	 or easy stuff like https://gerrit.wikimedia.org/r/311028
[21:41:08] * legoktm glares at MaxSem
[21:43:08] * MaxSem is not ashamed!
[21:46:39] <legoktm>	 TemplateSandbox fixes: https://gerrit.wikimedia.org/r/311029
[21:57:46] <legoktm>	 >  16 htmlspecialchars/Message->__toString/Message->toString/Message->parseText/MessageCache->parse
[21:58:05] <legoktm>	 something calling htmlspecialchars on a Message object...we need one more layer of the stack trace for that
[22:00:13] <legoktm>	 The GetPreferences hook needs context
[22:00:22] <legoktm>	 ConfirmEdit: https://gerrit.wikimedia.org/r/311034
[22:04:07] <legoktm>	 ostriches: is your EditPage wfMessage patch not merged yet?
[22:04:47] <ostriches>	 No not yet
[22:04:59] <ostriches>	 https://gerrit.wikimedia.org/r/#/c/309044/
[22:14:57] <ostriches>	 legoktm: Prolly shouldn't last minute swat these today... ;-)
[22:15:14] <legoktm>	 heh
[22:15:29] <ostriches>	 Much as I would like to see that log empty out some more.
[22:15:31] <legoktm>	 https://gerrit.wikimedia.org/r/#/c/311042/ fixes 2 extensions!
[22:17:31] <ostriches>	       1 wfCollectionSuggestAction/Message->parse/Message->toString/Message->parseText/MessageCache->parse
[22:17:34] <ostriches>	 Ewwww, Collection
[22:18:51] <ostriches>	 Ewww, it uses AjaxResponse * $wgAjaxExportList
[22:18:54] * ostriches runs and hides
[22:19:35] <MaxSem>	 ostriches, it also has $wgCollectionStyleVersion
[22:20:05] <ostriches>	 I miss $wgStyleVersion
[22:21:07] <ostriches>	 That $wgCollectionStyleVersion is completely useless.
[22:21:20] <ostriches>	 Basically appends a ?N to a single image.
[22:21:29] <MaxSem>	 it's used in oly one place, feel free to remove
[22:21:29] <legoktm>	 ostriches: Krenair has a patch for that, but it's stuck on some other things :(
[22:23:54] <ostriches>	 https://gerrit.wikimedia.org/r/311047
[22:25:23] <MaxSem>	 congrats, pending patches to that repo just got even less mergeable :}
[22:38:12] <legoktm>	 one of the AF warnings: https://gerrit.wikimedia.org/r/311049
[22:42:48] <legoktm>	 ostriches: we could backport https://gerrit.wikimedia.org/r/311052
[23:13:59] <TimStarling>	 is there any way to check out the subversion repository?
[23:14:03] <TimStarling>	 phabricator-ssh-exec: This repository ("rSVN") is not available over SSH.
[23:14:15] <TimStarling>	 says svn checkout svn+ssh://vcs@git-ssh.wikimedia.org/diffusion/SVN/ subversion
[23:14:39] <paladox>	 TimStarling that is better if you ask #wikimedia-devtools
[23:14:47] <paladox>	 since that channel is for phabricator
[23:15:06] <paladox>	 It may be because ssh instead setup for svn+ssh
[23:15:10] <paladox>	 try the http one
[23:16:21] <TimStarling>	 what HTTP one?
[23:17:22] <paladox>	 It should have offered you the http one
[23:17:30] <paladox>	 Oh maybe it is hidden
[23:18:23] <paladox>	 Oh they need to add the http links to https://phabricator.wikimedia.org/diffusion/SVN/manage/uris/
[23:19:43] <paladox>	 oh sorry TimStarling http access does not work for svn, guess you want to ask twentyafterfour who would know if something needs a config change for this too work