[00:01:30] who? [00:02:56] jeremyb: who what? [00:03:09] we = new wikis importers actually [00:03:19] user who has the dumps = wrh2/ryan iirc [00:03:26] ok guy... thank you so much :) [00:03:41] raylton: have you seen the pageb on incubator? [00:03:48] Nemo_bis: user who has the dumps i meant [00:04:04] ok [00:04:15] raylton: https://incubator.wikimedia.org/wiki/Incubator:Wikivoyage_import#Ex-Wikitravel_editions [00:05:13] TimStarling: if you have the time, could you take a look at https://gerrit.wikimedia.org/r/#/c/33157/ ? it's the RL module that pulls data from memcached / over HTTP. [00:05:25] Nemo_bis: thank you... i'll se this link... [00:05:57] at the moment I am busy fixing wikidata's complete brokenness, because apparently they are too busy to fix it themselves [00:07:17] TimStarling: OK, no problem. unless you tell me not to i'll bug you again in a day. [01:03:29] ori-l: done [01:03:37] thanks! why not HTTPS? [01:04:11] HTTPS is implemented as a gateway in front of squid that converts HTTPS traffic to HTTP [01:04:22] oh, so it goes outside the cluster [01:04:33] and there's no benefit [01:04:54] right, much much slower and it ends up being a plain HTTP request to an apache anyway [01:05:07] if you use HTTP, it should be able to use localhost [01:05:12] so you don't even have network overhead [01:05:36] ah, right, cool. [01:05:59] re: the JS module initialization code -- yes, came to that conclusion myself and implemented it another way in https://gerrit.wikimedia.org/r/#/c/33594/4/EventLogging.module.php . i should have rebased that on top of the change you just reviewed, sorry. [01:09:29] also: i wonder if it'd be appropriate to patch the memcached client to make the default expiry (when the parameter is unset) not the absolute maximum (30 days) but 29 days + some random fraction of 24 hours. [01:10:15] that way you don't have a whole bunch of keys expire at once a month after restarting memcached or something. [01:12:36] ^ TimStarling [01:12:42] there's no default expiry in the memcached client [01:14:13] if you omit the expiry time, it will be infinite [01:17:29] TimStarling: i misremembered, but it's still somewhat of a default: in includes/objectcache/MemcachedBagOStuff.php, the expiry time is passed through fixExpiry(), which sets $expiry to 2592000 if 2592000 < $expiry < 1000000000. [01:18:28] that's incorrect [01:18:57] it should just return $expiry + time() [01:19:10] there's also a bunch of code that explicitly sets the expiry to be 2592000, so depending on how real an issue this is there may even be good cause for either modifying the calling code or deliberately fudging 2592000 [01:19:44] memcached supports expiry times longer than 30 days, it's just that the relative time feature is broken [01:20:23] It's an attempt to detect whether $expiry is actually an interval or a UNIX timestamp, IIRC [01:20:39] sure, it's broken though [01:20:56] fixExpiry() itself does the same detection [01:21:08] except it puts the pivot point in 2011, not in February 1970 [02:35:52] All my gadgets on enwp are down, but monobook.js is working [02:36:03] I've cleared my cached, used a different browser [02:36:09] And another user is having the same issue [02:37:05] and its cross-skin too [02:37:29] The gadgets tab doesnt even show up in Special:Preferences [02:43:14] Someone must've made a syntax error on an interface page [02:44:00] Jasper_Deng: no one has editted common.js/.css or skin variants in a few days [02:44:53] And the only error I'm getting with Firefox's Error Console is on a personal userscript via monobook.js [02:52:43] https://en.wikipedia.org/wiki/Wikipedia:Village_pump_%28technical%29#Missing_Gadgets_Tab [02:52:46] anyone have ideas? [02:57:25] legoktm: Are you getting any errors in your error console? [02:57:33] I imagine it's fundraising JS. [02:57:36] But that's just a guess. [02:57:51] Brooke: I don't see any [02:58:23] Brooke: http://cl.ly/image/3p392I2B2u2z [02:58:28] But those are individual scripts [02:58:32] yeah probably [02:59:01] popups still works for me, but only if I load it from my monobook.js file [03:00:27] TimStarling: Any idea why the "Gadgets" tab has disappeared from ? [03:04:09] RoanKattouw: Or you, if you're around. Same question. [03:05:01] Looking [03:05:37] It shows up at Meta-Wiki. [03:05:55] > var_dump(Gadget::loadStructuredList()); [03:05:55] array(0) { [03:05:55] } [03:06:01] no gadgets, so no gadget preferences [03:06:11] Localisation cache bug? [03:06:47] https://en.wikipedia.org/wiki/MediaWiki:Gadgets-definition seems normal. [03:08:51] looks like the memcached value was corrupted somehow [03:10:11] I'll just delete it [03:11:11] !log deleted corrupted memcached value enwiki:gadgets-definition:6 [03:11:18] Logged the message, Master [03:11:24] this happens on beta quite a lot [03:11:30] Gadgets tab returned. [03:11:31] or it did [03:11:37] Thanks, Tim. :-) [03:11:48] thanks Tim [03:11:51] I think there's a bug too for it [03:12:14] yay \o/ [03:12:35] thank you [03:13:26] https://bugzilla.wikimedia.org/show_bug.cgi?id=12128 [03:13:32] > Extension:Gadgets with strange page caching (?) bug on special:preferences [03:14:10] It seems Daniel has a knack for bypassing auto-linking. [03:19:32] (joining late - presumably folks know about gadgets being down?) [03:19:59] lexein: It should be resolved now. [03:20:22] lexein: I saw your comment at en.wiki's WP:VPT. [03:20:25] testing [03:20:40] Individual user gadget prefs should be unaffected. [03:22:22] After ctrl-shift-R in Firefox, still no Twinkle [03:23:08] gadgets tab is back in Preferences [03:23:37] Ok, now Twinkle is back. [03:24:36] Neat! Thanks to all for their diligent work. A beer to whoever fatfingered, too! [03:26:19] Oh, I see that there was no fatfinger, just a corrupted cache [03:33:48] TimStarling while you're fixing things, do you feel like knocking back the #0000FF instensity of the begging bar, before it turns off too many people? [03:34:15] hehe [03:34:37] something more Apple Gangam style, you know silver and less silver? :) [03:37:10] that blue bar was discussed on the staff list about 5 months ago [03:38:19] and nobody managed to convince the fundraising team to give it a rest, despite some strong opinions voiced [03:38:56] pls. convery my thanks to the fundrasing team for not making it blink in all RGB primary maximum values [03:39:05] convey * [03:42:22] mareklug: Doing so [03:42:47] btw, it is the fundraising team that released the Polish version that was evidently machine translated and lacked diacritics, so that the general population started calling it in as a possible phishing site? :) [03:49:40] mareklug: you're referring to http://meta.wikimedia.org/w/index.php?title=MediaWiki:Centralnotice-2012translations-facts-1/pl&direction=prev&oldid=4362582 ? [03:50:34] That seems to be a faithful substring of the real deal I saw on a Polish anti-phishing website, yes [03:51:22] I guess that's what you get for using a wiki for your translations, instead of paying money [03:51:23] Actually, isn't tomaszf on hand in San Francisco? A native speaker? [03:52:25] https://meta.wikimedia.org/w/index.php?title=Special:Translate&taction=proofread&group=page-Fundraising+2012%2FTranslation%2FLanding+Page+and+Banner+messages&language=pl [03:53:23] at first blush, all tht looks fine. [03:53:25] https://meta.wikimedia.org/w/index.php?title=Translations:Fundraising_2012/Translation/Landing_Page_and_Banner_messages/9a/pl&direction=prev&oldid=4550230 [03:53:32] because it was edited [03:53:55] the original was created by Phdasa [03:54:16] if I may say so, highly negligent [03:55:28] not sure what you mean [03:55:55] on user:Phdasa's part, to make that entry [03:57:11] you can't really be negligent if you have no responsibility [03:57:44] also, a minor mystery -- how did the complete mistranslation appear on site (the same content as the English banner). I don't see it in the history (the machine translated version). the full text in this hitosry does not appear until Leinad introduces the proper Polish on the 15th. [03:57:49] you heard the man. +2 in core for everybody. [03:58:00] * ori-l chugs. [03:59:31] TimStarling well, we can then consider user Phdasa irresponsible, no? It would be sort of like me going into a bahasai indonesia banner version and mucking with it (I am not a speaker of Indonesian) [04:00:51] both "irresponsible" and "negligent" imply that there was some duty of care which was breached [04:01:28] in my eyes, yes. That of being a responsible editor. [04:01:58] you're saying this Phdasa is an editor? [04:02:21] but more to the point, that a full machine translated text did appear after all. And it is not the fragment that Phdasa contributed -- it was complete. [04:04:20] * ^demon is confused why this is a problem for #wikimedia-tech. Isn't it something for meta to deal with? [04:06:53] ^demon sorry. [04:14:15] ^demon: #wikimedia-fundraising is more relevant? [04:14:44] <^demon> I'm not sure if it's fundraising either.... [04:14:53] <^demon> If it's translations on meta, it should be dealt with by meta. [04:15:01] * ^demon goes back to dinner [04:16:14] the translate extension does seem to have a review system [04:16:33] so I suppose the policy question is what level of review you require before you launch a message [04:16:58] but that's still not a technical issue, since the messages were copied into CentralNotice manually [04:58:38] !log updated payments cluster to c898ce38e162516e73 [04:58:45] Logged the message, Master [05:24:39] Hey I noticed on Wikiquote, the buttons in the top right don't have "My" prefixes anymore. Did something change? [05:25:06] they're not yours any more :P [05:25:27] i proposed dropping them and the reactions were, on the whole, positive [05:25:30] let me dig up the bug [05:25:36] https://bugzilla.wikimedia.org/show_bug.cgi?id=41672 [05:56:59] Lirodon: i guess we could restore the My with some simple user JS [07:19:42] whee, we've got a JS error in TMH killing all the JS on the live site for logged-in users [07:19:55] TMH? [07:20:37] TimedMediaHandler [07:20:37] oh, timed media handler. [07:20:53] link? what do i need to do to reproduce? [07:21:12] go to enwiki, see the error [07:21:31] no errors in chromium nightly [07:21:39] it uses mw.getConfig() all over the place however this function is not defined [07:21:52] oh, yup. sec. [07:21:56] I see errors in FF and Opera [07:23:43] * RoanKattouw_away sighs [07:24:36] I don't see it [07:24:39] Link? [07:26:06] RoanKattouw, e.g. on https://en.wikipedia.org/wiki/CAT:RFU collapsible content doesn't get collapsed and search suggestions don't work [07:26:29] WFM [07:26:40] (Logged in) [07:26:53] "mw.getConfig is not a function" [07:27:26] and indeed, where is it defined? [07:27:52] I don't think mw.getConfig exists [07:28:20] No wait, it does [07:28:22] what? [07:28:36] Where the hell is that defined [07:28:57] the o nly instance I see in MwEmbedSupport [07:30:30] I see [07:30:43] So there must be some dependency breakage then [07:31:38] * RoanKattouw looks at Firebug net panel and laughs [07:31:51] Firefox is parallelizing my upload.wm.o requests between IPv4 and IPv6. Well done, Firefox [12:46:52] hey guys, somebody really shoukd backport and deploy https://gerrit.wikimedia.org/r/#/c/33417/ [14:18:57] what happened with right click to edit sections? [14:19:03] I can't do it on two different machines [14:20:13] removed? [14:20:42] the function was removed? [14:20:45] that's a thing? [14:21:00] are you sure it's not some script you have that's stopped working? [14:21:09] it's an option in Special:Preferences [14:21:30] not sure. preference cleanup has been talked about for a while and I don't know whether it's a victim [14:21:38] and it's something that's just stopped working today [14:21:41] hmm, so it is [14:22:04] I'm fairly certain I was on my work computer today and the function wasn't working [14:22:17] nope, wfm [14:23:30] not working for me in Chrome either [14:23:47] how the fuck is this an issue with my computer [14:24:59] anything in the error console? [14:25:11] how do I open that up again [14:25:28] ctrl+shift+J in firefox [14:26:02] several warnings [14:26:18] just look at the error tab [14:26:29] no errors [14:26:34] huh [14:26:44] I turned off and turned the option back on and now it's working fine [14:26:57] welp [14:27:14] at least it's not like whatever the fuck happened with my YouTube UI recently [15:04:37] The issue that Tim-away fixed last night seems to be back [15:04:49] Regarding the gadgets disappearing [15:17:52] legoktm: If you edit the page that'll fix it again [15:18:08] MaxSem: ^^ That gadgets memcached bug is appearing in production now [15:18:17] Reedy: Er edit which page? [15:18:50] Reedy, WFM [15:19:05] Ditto on enwiki [15:26:20] err, or was it on a different wiki? [15:29:58] legoktm: ? [16:56:50] Reedy: chairwiki 12 ? [16:57:18] oh it's not closed [16:58:13] hashar: if you have a minute, could you +2 https://gerrit.wikimedia.org/r/#/c/33721/ (if it looks OK?) that should fix login on beta [16:58:25] chrismcmahon: I am on it [16:58:37] figuring out how the wgObjectCaches works out [16:59:10] we have soooo many different stuff: [16:59:11] CACHE_MEMCACHED => array( 'factory' => 'ObjectCache::newMemcached' ), [16:59:17] versus 'memcached-pecl' => array( 'class' => 'MemcachedPeclBagOStuff' ), [16:59:17] ;) [17:03:32] chrismcmahon: Matthias talked about that to me this afternoon [17:03:36] I was clueless :-( [17:04:03] Like I said, the best way to improve beta is to use it. [17:05:27] hashar: both newMemcached & memcached-pecl should do the same (= access memcached), but using different wrappers [17:06:17] ObjectCache::newMemcached will result in an (old) MemcachedPhpBagOStuff instance, whereas MemcachedPeclBagOStuff is a wrapper over PECL's memcached implementation [17:07:11] both would do the job (reach for memc instead of redis) [17:07:19] oh nice [17:07:25] we have sooo many way to handle stuff :( [17:07:47] indeed :) [17:08:47] If someone had told me 12 yrs ago that one of the most biggest and most important sites on the Internets runs on PHP I'da responded something along the lines of ".. Yeah, and cows fly." [17:08:56] are we going to move off memc completely? [17:09:11] <^demon> If we are it's news to me. [17:09:22] jubo2, and not only it;) [17:09:44] mlitn: I changed a few your code with patchset 3 https://gerrit.wikimedia.org/r/#/c/33721/ [17:09:52] mlitn: aka I got rid of the multi write system [17:10:05] and just override on $cluster == 'wmflabs' [17:10:14] $cluster will be phased out with an ongoing rewrite [17:10:29] MaxSem: yup memcached is going to disappear as far as i know in favor of reddis. [17:10:32] i just noticed, is cleaner this way [17:10:48] MaxSem: which let us scale out easily and has a disk backed cache [17:11:27] mlitn: now we need someone else to deploy it [17:11:34] though I could [17:11:41] hmm let me review it a second time [17:11:53] hashar, so far we have an assumption that if cache has an entry, it will be retrieved very quickly. disk-backed cache will be different in this regard [17:12:23] I think the idea is that whenever a server dies, it can repopulate its cache from the disk or something similar [17:12:27] i have no idea really [17:13:55] mlitn: can't merge, there is another change in :( [17:13:58] Reedy: ping pong [17:14:07] ? [17:14:08] !g I4e2e3eca2d14cc720a0b3b0ed65d1c892305175a [17:14:08] https://gerrit.wikimedia.org/r/#q,I4e2e3eca2d14cc720a0b3b0ed65d1c892305175a,n,z [17:14:18] that change has been merged but is apparently not deployed [17:14:29] it is not in the master branch on fenari [17:14:36] ohh [17:14:37] It doesn't need deploying... [17:14:40] that is just refresh-dblist [17:14:45] will pull it in [17:15:05] lol https://en.wikipedia.org/wiki/File:Mammal_Diversity_2011.png [17:15:41] humans are represented by Brezhnev and Nixon... [17:19:23] ^demon: Reedy: what do you think about merging multiversion in mediawiki-config repository ? [17:19:38] Indifferent [17:19:52] <^demon> Don't really care. [17:19:53] It seems a bit weird at least [17:19:53] ^demon: Reedy: I am not sure there is any point in maintaining it in a different git repo and that will definitely help anomie [17:20:09] are we using multi version anywhere else? [17:20:34] <^demon> As I explained yesterday, it can be done [17:20:37] I thought we could have some classes to abstract out our config such as the DBList [17:20:39] <^demon> I don't personally care enough to do it though. [17:20:51] or wikiversion.dat and get use of them in both repositories [17:21:22] I would create an orphan branch in mediawiki-config, get the change from the multi version repo then merge. [17:21:24] ohh [17:21:28] need to look at the link [17:21:32] isn't mwscript also part of it? command line scripts mixed with config is scary [17:23:57] mwscript is in multiversion I think [17:25:26] that's what im talking about [17:33:06] well will think about it during the weekend [17:33:10] and drop an email later on [17:40:23] I am out of there, have a good weekend! [17:40:44] have a good weekend, hashar [19:53:48] [[Tech]]; Jerome Charles Potts; /* Server response times */ ; https://meta.wikimedia.org/w/index.php?diff=4576904&oldid=4456489&rcid=3700042 [20:18:00] lol [22:07:46] Reedy: are you sure https://bugzilla.wikimedia.org/show_bug.cgi?id=42152 is the bug you wanted to close? [22:08:01] No [22:08:02] ah you reopened it already ^^' [22:08:06] Thats why I already reopened it [22:08:07] Yeah [22:08:15] ;) [22:09:48] Nemo_bis: It would seem my manual run has seemingly stopped on enwikinews for some reason [22:10:51] Actually [22:10:59] I'll blame AaronSchulz and FlaggedRevs/ValidationStatistics [22:13:22] I bet that's why that SQL error appears... [22:13:31] Reedy: so they don't like each other? [22:15:27] LQT is always guilty; if not installed, then it must be FlaggedRevs [22:15:39] I'm disabling ValidationStatistics for enwikinews [22:17:26] saper: are you sure it doesn't infect neighbouring wikis too? [22:19:36] [[Extension:STD]] [22:23:08] TODO: work out wth is up with enwikinews [22:26:29] I'm curious, Reedy - what are the symptoms? [22:26:46] updateSpecialPages.php hangs [22:29:24] and it is ValidationStatistics... Just how I tried to disable it won't work [22:31:31] it's flaggedrevs self-protection [22:31:44] a wiki with FR and without those pages is quite useless ;) [22:32:41] does the mw.getConfig fix warrant a deployment? [22:32:52] TMH? [22:33:38] MaxSem: What's the fix [22:33:43] https://gerrit.wikimedia.org/r/33819 [22:33:56] lol [22:33:58] I'd think so [22:33:59] It's weird though, because mw.getConfig is actually defined [22:34:08] I mean it's stupid, but it's defined, it "should" work [22:34:19] But yeah that should be deployed [22:34:35] Reedy: I just deployed the tablesorter fix 10-15 mins ago, you wanna take this one? [22:34:45] I can do when I come back in a few [22:34:48] Though, MaxSem is capable... [22:34:55] okay [22:35:57] Also need to go get my laptop from my car boot [22:37:32] out for bed, have a nice weekend! [23:25:25] AaronSchulz: about? [23:27:25] * AaronSchulz is nowhere [23:28:30] Any idea why FlaggedRevsStats::update() seems to hang/take for ever? [23:29:25] I'm contemplating putting in a global for it and disabling it's running in special page cache updates [23:32:46] https://gerrit.wikimedia.org/r/gitweb?p=mediawiki/extensions/FlaggedRevs.git;a=commitdiff;h=5c2f93997fce03191a8223cdb7a9ec46891d3a12 [23:32:51] https://gerrit.wikimedia.org/r/gitweb?p=mediawiki/extensions/FlaggedRevs.git;a=commitdiff;h=4f41dcada3bee44816fef66d5af3d30ad6e6df44 [23:33:03] I'm guessing one of those are at fault [23:37:48] AaronSchulz: https://gerrit.wikimedia.org/r/33845 [23:38:11] Oh, that isn't going to work, is it... [23:38:33] Due to the extensionsfunctions usage [23:39:44] global it is [23:48:22] can't you just put something in wmf4? ;) [23:49:07] Are you likely to fix it before wmf5? :p [23:50:24] $reviewDataUser = self::getEditReviewTimes( $dbCache, 'users' ); is slow [23:51:10] and we call it twice [23:51:12] just for the luls [23:51:38] it's called once [23:52:02] well, with anons [23:54:40] + if ( $special === "ValidationStatistics" ) { continue; } [23:54:43] is my current live hack [23:54:49] but it don't want to keep it around long