[16:59:23] * csteipp disappears to interview.. [17:09:32] ori: https://gerrit.wikimedia.org/r/#/c/200821/ [17:11:14] aspiecat: {{done}} [17:11:29] lgtm as well [17:11:53] I saw in backscroll this morning that you found some of the mw-v stat cache WTF problems [17:12:31] bd808: https://gerrit.wikimedia.org/r/#/c/200823/ [17:12:31] HHVM's inotify based cache is awesome except for the fact that it won't work with any remote fs at all [17:12:44] 6MediaWiki-API-Team, 10MediaWiki-extensions-OAuth, 5Patch-For-Review: Message describing OAuth activities is confusing to end user (in context of Wikidata Game) - https://phabricator.wikimedia.org/T69082#1167521 (10Peteforsyth) >>! In T69082#1167473, @Ricordisamoa wrote: >>>! In T69082#738278, @Peteforsyth w... [17:17:03] 6MediaWiki-Core-Team, 10CirrusSearch: Fix highlighting for phrase prefix queries - https://phabricator.wikimedia.org/T93014#1167528 (10Jdouglas) On enwiki, this does indeed appear to be problematic: {F107483} [17:29:34] * legoktm wonders why someone added me as a reviewer to my own patch [17:30:21] * bd808 asks legoktm to review himself for correctness [17:39:46] 6MediaWiki-Core-Team, 10CirrusSearch: Fix highlighting for phrase prefix queries - https://phabricator.wikimedia.org/T93014#1167586 (10Manybubbles) Ok - so this is an error in the highlighter I think. It must not know what to do with these queries. Its actually a reasonably fix - we reproduce it with the uni... [17:41:01] * marktraceur has to go get the car serviced [17:41:05] May be incommunicado. [17:41:33] Get that roadtrip prep done [17:42:06] You don't want to break down between St Louis and the new world [17:46:29] 6MediaWiki-Core-Team, 10CirrusSearch, 10Wikimedia-Site-requests, 7Easy: Remove "this wiki is using a new search" - https://phabricator.wikimedia.org/T77920#1167606 (10Manybubbles) https://wikitech.wikimedia.org/w/index.php?title=Deployments&diff=151438&oldid=151399 [17:46:38] bd808: https://gerrit.wikimedia.org/r/#/c/200818/ [17:51:21] 6MediaWiki-Core-Team, 10CirrusSearch: Fix highlighting for phrase prefix queries - https://phabricator.wikimedia.org/T93014#1167627 (10Jdouglas) Now I'm having trouble reproducing this locally with the patch in https://gerrit.wikimedia.org/r/#/c/197397/: {F107495} [18:03:46] bd808: do you have any thoughts on how difficult https://phabricator.wikimedia.org/T94321 will be? [18:07:41] Huzzah, waiting room wifi [18:07:55] Polar Mazda Treats You Right ™ [18:08:16] Also: Free pepsi and coffee. I should get my car washed like every week here. [18:39:39] * csteipp shakes fist at 5am self [18:42:38] legoktm: I assume you tested Gerrit 200727 already? [18:43:12] anomie: yes, but I didn't have Thanks installed at the time... [18:47:15] 6MediaWiki-Core-Team, 10VisualEditor, 10VisualEditor-Performance, 3VisualEditor 2014/15 Q3 blockers, 5WMF-deploy-2015-04-01_(1.25wmf24): Stash VisualEditor edits using edit stash - https://phabricator.wikimedia.org/T90040#1167818 (10Jdforrester-WMF) 5Open>3Resolved [18:51:16] anomie: your revision patch reminded me of https://gerrit.wikimedia.org/r/#/c/197978/ [18:51:59] * anomie will look at that in a little bit [18:52:10] e.g. queries could silently fail (returning false) afterwards...seems a bit esoteric though, but possible [18:53:45] then again <<$this->mId = $rev_id !== null ? $rev_id : $dbw->insertId();>> is pretty straightfoward...so either it's that or $rev_id is falsey but not null [19:01:57] legoktm: It shouldn't be too hard. There is a rq_status enum that the new state could be added to. There would need to be UI and workflow changes. The proposal needs a bit more detail on what the stewards really want from the feature. [19:02:04] 6MediaWiki-API-Team, 6Collaboration-Team, 10MediaWiki-extensions-Renameuser, 10SUL-Finalization, and 3 others: Move more log types to new user name when renaming a user (for example thanks logs) - https://phabricator.wikimedia.org/T78575#1167890 (10Legoktm) 5Open>3Resolved [19:03:39] 6MediaWiki-API-Team, 10MediaWiki-extensions-CentralAuth, 10SUL-Finalization: GlobalRenameQueue needs a means (status) to set "on hold" function on a rename request - https://phabricator.wikimedia.org/T94321#1167906 (10bd808) What workflow changes would the "hold" feature add? Are you just looking for a way t... [19:04:45] legoktm: hard to guess how long it would take to implement without a more complete plan for the feature change [19:31:11] aspiecat: I know. My best guess at the moment is that there's some sampled profiling or logging of some sort from the guts of the database classes that's started doing an extra DB query, which would have the side effect of clobbering $dbw->insertId(). [19:38:45] could be, can't think of anything off hand [19:38:59] aspiecat: Re Gerrit change 197978, we don't need to worry about recursive errors after all? [19:40:14] report() errors should be handled by MWExceptionHandler, and "throw" errors are only an issue if rollback() or something is issued in the subclass (like PG does) [19:40:44] otherwise, I don't see how just constructing the exception would recurse [19:42:04] Can wfDebug() be configured to write log entries into the database somehow? [19:45:39] I don't see any way (that would be evil anyway) [19:48:58] bd808: is this a known thing? Could not load package phpspec/prophecy in http://packagist.org: [UnexpectedValueException] Could not parse version constraint ^1.0.2: Invalid version st ring "^1.0.2" [19:49:38] ori: I haven't seen it. Where are your running composer? [19:49:46] mediawiki root [19:49:49] $IP [19:49:52] blerg [19:51:05] ok, i ran composer self-update and now it works [19:51:14] it didn't work with version 1.0.0-alpha9, which is what i had installed [19:51:24] yuck [19:51:57] I wonder which dependency pulls that in? We don't need phpspec by default surely [19:52:07] Oh maybe for dev deps? [19:52:36] * ori shrugs. [19:52:55] aspiecat: So if the thrown DBQueryError isn't caught by something else, we'll get into MWExceptionHandler::handleException(), which might come back around to call ->rollback() on the DB object, which seems like it could decide to have another error to report. In this case that error would be an uncaught exception. [19:53:55] ori: Its lame how loose the composer versioning is. Packages require features that merged days ago on a pretty regular pasis [19:54:00] *basis [19:54:33] legoktm has had problems before trying to run "real" releases too [19:54:59] alpha9 is pretty ancient by now I think in composer years [19:55:39] " 312 commits to master since this release" [19:55:59] nod, no big deal. [20:10:23] anomie: self::rollbackMasterChangesAndLog( $e ) can't handle partitions during multiple transactions atm, it should probably catch any rollback errors and move on (via some rollbackMasterChanges() flag). That would also stop loops. [20:11:37] too bad we could just disconnect and let mysql rollback (which does anyway) ;) [20:11:52] 6MediaWiki-API-Team, 10MediaWiki-extensions-CentralAuth, 10SUL-Finalization, 7Easy: Log promote to global renames in the global rename log - https://phabricator.wikimedia.org/T93235#1168116 (10Legoktm) a:3Legoktm [20:12:25] *it does [20:12:53] * aspiecat should make another patch for that [20:16:44] I guess it wouldn't loop though since rollbackMasterChangesAndLog/handleException don't catch themselves [20:38:03] 6MediaWiki-API-Team, 10MediaWiki-extensions-OAuth, 5Patch-For-Review: Message describing OAuth activities is confusing to end user (in context of Wikidata Game) - https://phabricator.wikimedia.org/T69082#1168179 (10Peteforsyth) >>! In T69082#1164512, @MarkTraceur wrote: > Patch incoming, thanks! Hi M... [21:08:56] Krinkle: what do you think would be the best way to fix the delay in the rendering of $wgLogo? The image is currently set via a background-image: rule in ResourceLoaderSkinModule.php, which is loaded via an external load.php request, causing it to render quite late. The style would have to be inlined to fix that, I think. [21:09:25] ori: Is it not in the top module? [21:09:46] background image requests are async [21:10:20] Embedding might work, but not for external urls, and probably too big [21:10:53] ori: When cached though, it should appear immediately. [21:11:00] And yet it doens't [21:11:08] ori: The html order might be relevant [21:11:16] It being on the bottom and all [21:11:27] I don't think so [21:12:13] https://en.wikipedia.org/wiki/Special:BlankPage [21:12:38] still late [21:12:49] ori: Chrome timeline says it made the request after wikimedia-button was requested and loaded [21:13:01] network -> images [21:13:08] it's not being triggered until very late [21:13:23] because the wikimedia button has its src set in the html [21:13:54] 6MediaWiki-Core-Team, 6Security, 5Patch-For-Review: Custom JavaScript may yield privilege escalation - https://phabricator.wikimedia.org/T85855#1168298 (10csteipp) [21:13:55] why isn't the logo just an image tag? [21:13:55] yeah, but the stylesheet is loaded ealier, and the relevant html element is encoutered before the foote,r too. [21:14:47] 6MediaWiki-Core-Team, 6Security: xml_parse doesn't expand internal entities sometimes - https://phabricator.wikimedia.org/T88310#1168307 (10csteipp) [21:15:17] 6MediaWiki-Core-Team, 6Security: Billion Laughs attack in SVG and XMP Metadata - https://phabricator.wikimedia.org/T85848#1168313 (10csteipp) [21:15:30] 6MediaWiki-Core-Team, 10MediaWiki-API, 6Security, 7HHVM, 7Vuln-XSS: Reflected XSS in api.php using wddx formatting - https://phabricator.wikimedia.org/T85851#1168316 (10csteipp) [21:15:53] ori: because of https://phabricator.wikimedia.org/T58257 [21:15:56] 6MediaWiki-Core-Team, 6Security, 5Patch-For-Review, 7Vuln-XSS: XSS in language converter when used with Html class's tricky escaping - https://phabricator.wikimedia.org/T73394#1168329 (10csteipp) [21:16:00] ori: I ripped out the media queries and SVG in case that makes it wait for viewport rendering [21:16:01] https://test2.wikipedia.org/wiki/Special:BlankPage [21:16:07] No difference [21:16:15] (low res png of course, but same delay) [21:16:36] 6MediaWiki-Core-Team, 6Security: Stored XSS in SVG via embedded SVG - https://phabricator.wikimedia.org/T85850#1168338 (10csteipp) [21:16:46] 6MediaWiki-Core-Team, 6Security, 7Vuln-XSS: Fix SVG blacklist for animate - https://phabricator.wikimedia.org/T86711#1168341 (10csteipp) [21:18:13] csteipp: I don't see the announcement yet? [21:18:45] I can't do everything at once :) [21:19:01] :P [21:22:17] Krinkle: https://chrome.google.com/webstore/detail/perfmap/hgpnhiajcdppfbogcpfdgcceepgkhdmk?hl=en&gl=GB is useful, btw. It uses the Resource Timing API to highlight elements that rendered late [21:26:55] Krinkle: http://www.ravelrumba.com/blog/a-look-at-how-browsers-download-and-render-css-background-images/ suggests that images referenced in a stylesheet are blocked until *all* stylesheets have loaded [21:29:56] * AaronS hands bd808 https://gerrit.wikimedia.org/r/#/c/200813/ [21:40:52] ori: all stylesheets the browser knows about [21:41:12] ori: So I guess that's a good reason to ensure our bottom queue doesn't start too early [21:41:38] Although I don't want to intentionally delay it either [21:45:16] ori: interesting post from ravel rumba! [21:46:52] ori: It doesn't talk about stylesheets loaded by a script in the footer though [21:46:57] we only have stylesheets in the top otherwise [21:46:59] so that should be fine [21:51:15] Krinkle: I think I have a fix. Are you going to be around in 15? [21:51:23] minutes, yes [21:59:19] 6MediaWiki-API-Team, 10MediaWiki-extensions-OAuth, 5Patch-For-Review: Message describing OAuth activities is confusing to end user (in context of Wikidata Game) - https://phabricator.wikimedia.org/T69082#1168493 (10Krinkle) I suppose it would also be desirable to allow tools to request authorisation with OAu... [22:05:10] Krinkle: ooof. Do you know why enwiki is setting the background-image for high DPI displays using Common.css rather than $wgLogoHD ? [22:05:35] ori: That variable does not exist afaik. last I checked it was a patch in Gerrit with quality issues. [22:06:23] Krinkle: no, it's in core: https://gerrit.wikimedia.org/r/#/c/195195/ [22:06:40] k [22:11:44] ori: Sounds like we can use that then :) [22:17:39] 6MediaWiki-API-Team, 10MediaWiki-extensions-OAuth, 5Patch-For-Review: Message describing OAuth activities is confusing to end user (in context of Wikidata Game) - https://phabricator.wikimedia.org/T69082#1168572 (10Anomie) >>! In T69082#1168493, @Krinkle wrote: > I suppose it would also be desirable to allow... [22:24:10] 6MediaWiki-API-Team, 10MediaWiki-extensions-OAuth: Only show grants that the user has permission to use, to avoid confusion - https://phabricator.wikimedia.org/T94478#1168626 (10Anomie) Something else that came to mind: would you show a grant like "Delete pages, revisions, and log entries" (see [[https://www.m... [22:33:13] 6MediaWiki-API-Team, 6CA-team, 10MediaWiki-API, 10MediaWiki-User-login-and-signup, 10Wikipedia-App-Android-App: App says I'm logged in, but edits are saved from IP - https://phabricator.wikimedia.org/T75086#1168645 (10Mattflaschen) Are you testing in the production environment? Maybe it has something to... [22:38:13] 6MediaWiki-API-Team, 10MediaWiki-extensions-CentralAuth, 10SUL-Finalization: GlobalRenameQueue needs a means (status) to set "on hold" function on a rename request - https://phabricator.wikimedia.org/T94321#1168652 (10Billinghurst) Pretty much the case that there would be no need to change the workflow. I th... [22:51:20] 6MediaWiki-API-Team, 6CA-team, 10MediaWiki-API, 10MediaWiki-User-login-and-signup, and 2 others: App says I'm logged in, but edits are saved from IP - https://phabricator.wikimedia.org/T75086#1168697 (10Dbrant) I have a patch that implements assert=user, but it's still only theoretical, since I can't repro... [22:52:55] 6MediaWiki-API-Team, 6CA-team, 10MediaWiki-API, 10MediaWiki-User-login-and-signup, and 2 others: App says I'm logged in, but edits are saved from IP - https://phabricator.wikimedia.org/T75086#1168700 (10bearND) I think an important step is missing: 5. Wait for x hours/days. Not sure how long, but it'll tak... [22:53:22] anomie: is the Bug56269 log still useful? [22:54:43] 6MediaWiki-API-Team, 6CA-team, 10MediaWiki-API, 10MediaWiki-User-login-and-signup, and 2 others: App says I'm logged in, but edits are saved from IP - https://phabricator.wikimedia.org/T75086#1168731 (10bearND) Never mind. Got your updates after I wrote my comment.