[06:22:08] great, got the torrents done, e.g. https://archive.org/download/wikimediacommons-201211/wikimediacommons-201211_archive.torrent [06:22:30] does anybody have suggestions on how to make a torrent containing a hundred other torrents? [06:23:49] for now the best idea I had is to do just that... torrent containing mere torrent files, if one wants to download everything they move them to relevant hotdir [06:57:52] Nemo_bis: nice work! [07:01:43] thanks [07:02:05] 22,267,943,940 KB uploaded so far :) [13:40:23] Who would be able to OK my OAuth consumer registration on mw.o? [13:41:21] oh, I could have checked that myself - it's available via https://www.mediawiki.org/w/index.php?title=Special:ListUsers&group=oauthadmin [13:41:25] valhallasw: cstreipp [13:43:54] p858snake|l: hm, he's not listed in the oauthadmin list, but I'll poke him. Thanks. [14:08:01] anomie, are you able to OK oauth consumer requests on mw.o? you're listed in the oauthadmin group. [14:10:15] valhallasw: Let me check [14:11:41] valhallasw: There you go [14:12:18] Yay! [14:12:23] * valhallasw hands anomie stroopwafels [14:12:44] * anomie enjoys stroopwafels [14:33:08] oh, i want stroopwafel now [15:38:37] Hi. What happened in Russian Wikipedia with categories? [15:39:53] Everybody see arabic letters instead of russian. [15:48:12] hi, who knows what is wrong with categories on the Swedish and Finnish Wikipedia: see e.g. https://sv.wikipedia.org/wiki/Kategori:Grundl%C3%A4ggande_matematik [15:49:19] it's been broken now over 3 hours [15:53:17] Uh [15:54:07] Huh [15:55:13] wt [15:55:14] The only thing that have changed are some DB hosts, and a PHP package upgrade [15:55:24] in the last 3 hours? [15:55:24] (according to SAL) [15:55:35] He said over 3 hours [15:55:39] Arabic in Russian wp ? did you confirm yet? [15:55:42] 3-5 hours, I think [15:55:43] ok 5 hours [15:55:49] putnik: Got a link for example? [15:56:03] svwiki, fiwiki, ruwiki [15:56:19] not only Wikipedias, also other projects [15:56:21] looks Russian to me https://ru.wikipedia.org/wiki/%D0%9A%D0%B0%D1%82%D0%B5%D0%B3%D0%BE%D1%80%D0%B8%D1%8F:%D0%A3%D0%BC%D0%B5%D1%80%D1%88%D0%B8%D0%B5_%D0%B2_1927_%D0%B3%D0%BE%D0%B4%D1%83 [15:56:36] s2 and s6 [15:56:40] look at the letters for the headings [15:56:49] All looks Greek to me xD [15:56:51] oooh, i see it now [15:56:51] wrong alphabet [15:56:54] and [15:56:54] yea [15:56:56] * apergos stabs T13 [15:57:04] * T13 is often not helpful... :p [15:57:15] that particular joke warrants autostabbing [15:57:22] anyways, s2, no work done that I know of [15:58:16] that's two of the three [15:58:20] * Reedy looks for the RT ticket for Alexs upgrade... [16:00:46] remember we installed the new font [16:01:07] also hu-wiki: https://hu.wikipedia.org/wiki/Kateg%C3%B3ria:Koordin%C3%A1tageometria [16:01:34] mutante: but that was earlier this week [16:01:35] imagescaler font stuff can't be related, right [16:01:37] https://rt.wikimedia.org/Ticket/Display.html?id=5912 [16:02:07] That too [16:02:07] hm huwiki is on s7 [16:02:26] I wonder which other wikis have it [16:02:32] pl-wiki also [16:04:28] The issue sort of looks collation related [16:05:16] I'm logged out seeing these, dunno if that makes a difference [16:06:04] is it all categories on those wikis? [16:06:39] I think not [16:07:09] at least in fiwiki, seems like yes [16:07:29] on ruwiki no (but hard to know what's been cached and what hasn't) [16:07:57] oohhh [16:08:11] * apergos previewed a category page for edit and saw the wrng alphabet [16:09:09] hm el wiki appears not to be hit (at least, i tried the preview trick on a random category and it is still ok) [16:10:44] s6 - fr, ja, ru s2 - bg, cs, eo, fi, id .. [16:10:54] try bgwiktionary [16:11:55] oh, we have a bug report: https://bugzilla.wikimedia.org/show_bug.cgi?id=55565 [16:12:01] preview on bg wikt looks ok [16:12:57] apergos: i agree, but it's also on s2, so it doesn't seem to be per db cluster? [16:13:03] hu is s7, [16:13:11] nope [16:14:29] Stryn: ah, that's good [16:14:42] so they are all that are using collation settings [16:15:04] looks at https://gerrit.wikimedia.org/r/#/q/project:operations/mediawiki-config,n,z [16:15:33] Don't suppose upgrading PHP would've upgraded any other packages too? [16:15:42] well there's a pile of related ones [16:15:52] and sees just unmerged changes about collation ... [16:16:13] Yeah, I've not moved any wikis for a while [16:16:51] mw1017 was upgraded in this way, we can look at... oh. if they've all been done we can't compare anything. mm [16:16:56] anyways it was done earlier [16:17:04] (yesterday?) [16:19:42] php5-common php5-dev man-db php5-cli libapache2-mod-php5 php5 php5-mysql php5-intl php5-xmlrpc php5-curl [16:19:48] is it worth trying to run mwscript updateCollation.php on something ? [16:19:59] oon the 8th and then today php-pear [16:20:14] php5-intl hm [16:20:24] Reedy: i see stuff like this "reedy@tin:/a/common$ mwscript updateCollation.php --wiki=fawiki [16:20:28] --previous-collation=uppercase" [16:20:52] apergos: Isn't php5-intl mostly a wrapper? libicu or something? [16:21:13] can we use that to see what happens maybe? [16:22:18] looking at it [16:22:45] libicu hasn't been updated so [16:22:56] and yep looks like bindings for it [16:23:44] 2013-10-10 12:18:32 status installed libicu42 4.2.1-3ubuntu0.10.04.1 [16:23:46] mw1020 [16:24:33] Can you tell from what version? [16:25:17] dpkg.log.4.gz:2013-06-13 21:05:29 status installed libicu48 4.8.1.1-3 [16:25:38] Upgrade: libapache2-mod-php5:amd64 (5.3.10-1ubuntu3.6+wmf1, 5.3.10-1ubuntu3.8+wmf1), php5-curl:amd64 (5.3.10-1ubuntu3.6+wmf1, 5.3.10-1ubuntu3.8+wmf1), php5-xmlrpc:amd64 (5.3.10-1ubuntu3.6+wmf1, 5.3.10-1ubuntu3.8+wmf1), php5-intl:amd64 (5.3.10-1ubuntu3.6+wmf1, 5.3.10-1ubuntu3.8+wmf1), php5-mysql:amd64 (5.3.10-1ubuntu3.6+wmf1, 5.3.10-1ubuntu3.8+wmf1), php5-cli:amd64 (5.3.10-1ubuntu3.6+wmf1, 5.3.10-1ubuntu3.8+wmf1), php5-common:amd64 (5.3.10-1ubunt [16:26:05] you got truncated mutante (pastebin?) [16:26:54] http://paste.debian.net/plain/55306 [16:28:46] mw1017 has both 42 and 48 [16:28:46] maybe they all have/do [16:29:10] looks like it [16:30:24] * jeremyb glares @ RT 5944 [16:33:18] dpkg.log.5.gz:2013-05-02 06:35:44 status installed libicu42 4.2.1-3ubuntu0.10.04.1 I hav this on my snapshot hosts from may so... [16:38:38] apergos: the dependency on libicu48 is from php5-intl [16:38:48] jeremyb: yea, don't you see how listinfo says "list run by ops":p [16:39:46] mutante: well also the URL they gave is exactly the same place they have to go to fulfill their own request [16:39:56] haha, yea [16:40:15] mutante: i wonder if this means we need to port over the listinfo makeover to this list too? [16:40:17] please fill out my name for me [16:40:18] would that make a diff? [16:41:16] mutante: 5943 is also fun. note the autogenerated mail is from OTRS [16:41:23] jeremyb: maybe, you know what would be great, move the "list run by" line from bottom to top, center and fontsize x-large :) [16:41:32] haha [16:43:14] well if libicu48 is the crucial one, it's been on there since June on mw1020, probably most other hosts too [16:43:37] jeremyb: heh, nice @ reopen/untaken [16:44:03] apergos: IIRC that's when we did the upgrades so we could do the better collation support [16:45:04] mutante: i was thinking that subject line looked really familiar... [16:45:05] :P [16:51:21] one thing is that the upgrades seem to be around 4-5 hours ago [16:51:27] guess I could get an exact time frame [16:53:12] right around noon utc all of em [16:55:59] 11:52 akosiaris: upgraded php5 packages from php5_5.3.10-1ubuntu3.6+wmf1 to php5_5.3.10-1ubuntu3.8+wmf1 on apt.wikimedia.org and there's nothing after that even close to relevant [16:56:01] meh [18:31:02] ^d: https://www.mediawiki.org/wiki/API:Query [18:31:07] on mw.o: This version of MobileFrontend requires MediaWiki 1.22, you have 1.22wmf21. You can download a more appropriate version from https://www.mediawiki.org/wiki/Special:ExtensionDistributor/MobileFrontend [18:31:13] https://www.mediawiki.org/wiki/User_talk:Leucosticte [18:32:22] Any idea if this is a bug in MobileFrontend or a bug in ExtensionDistributor? [18:32:33] I imagine it's in MobileFrontend [18:32:41] now it works [18:32:46] Anyone familiar with the category collection code? It appears to be broken, we're looking for the culprit to revert. Report at https://bugzilla.wikimedia.org/show_bug.cgi?id=55565 [18:33:03] Nemo_bis: How'd you do that? [18:33:09] Please ping greg-g if you have more info. [18:33:28] siebrand: have you seen scrollback? [18:33:43] siebrand: collation*. bawolff and tim starling know how that stuff works [18:33:49] (and i have a vague idea myself) [18:33:51] from about 17.38 CEST [18:34:01] yeah, those were the two names we thought of, MatmaRex [18:34:04] MatmaRex: Brian't not here, Tim has 05:00 or soemthing. [18:34:10] MatmaRex: could you take a look into it, maybe? [18:34:13] siebrand: it looks like somebody updated some packages… [18:34:15] MatmaRex: Would you want to do some detective work, please/ [18:34:16] ? [18:34:20] kaldari: reverted just now, from -ops [18:34:24] <^d> kaldari: Hm? [18:34:35] ^d: nevermind, fixed now [18:34:38] <^d> k [18:35:00] MatmaRex: what packages would affect this? [18:35:02] siebrand: has anybody tried running maintenance/generateCollationData.php with cluster packages, replacing the file it generates in production and testing if it solves the issue? [18:35:16] greg-g: php5-intl, anything with "icu" in the name [18:35:18] libicu? [18:35:21] ugh, I see [18:35:33] MatmaRex: I just saw the issue mentioned in #mediawiki-i18n and escalated to the release manager after reading the bug. [18:36:12] siebrand: greg-g: there was another bug similar to this somewhere, let me find it [18:36:22] (that was before that started getting deployed) [18:36:26] MatmaRex: so, reading the scrollback, it doesn't seem like there was any movement after people identified that the packages were indeed upgraded [18:36:40] apergos: Reedy ^^ [18:36:49] greg-g: https://bugzilla.wikimedia.org/show_bug.cgi?id=46036 [18:36:52] 4.8 is known good [18:36:59] newer should work, but it seems like it doesn't [18:37:19] greg-g: also [18:37:24] apergos: greg-g: https://bugzilla.wikimedia.org/show_bug.cgi?id=46036#c1 [18:37:27] When you change the ICU version, you have to rebuild all the sort keys for [18:37:27] wikis that use uca-* collations, with updateCollation.php --force. Lucid has [18:37:27] ICU 4.2, and precise has 4.8, so when the upgrade to precise was in progress, [18:37:27] Faidon built a custom php5-intl compiled against ICU 4.2. [18:37:37] so, do this [18:37:57] it'll take up to a few hours for the largest wikis with this deployed [18:38:05] siebrand: ^ [18:38:10] gotcha [18:38:24] * MatmaRex arrived just in the nick of time to save the day [18:38:25] Reedy: ^^^ the fix to the category collation stuff, apparently ^^^ [18:38:33] MatmaRex: thank you, sir. [18:38:34] ah it was against 4.2? I see [18:38:46] good someone knew all those details cause I sure didn't [18:38:49] MatmaRex: Awesome. Thanks. [18:38:50] thanks [18:38:50] yeah [18:38:57] yay community and all that ;) [18:38:58] ops broke it! [18:39:03] ;) [18:39:18] Is there a way this could have surfaced in a test? [18:39:27] This is a particularly nasty type of issue. [18:39:58] only way I could think would be to upgrade Beta Cluster packages before production, apergos ? [18:40:16] beta cluster: testing is good, even for ops :P [18:40:16] and have someone remember to check it (but automated testing is better) [18:40:35] siebrand: we could add some tests to verify that the first letter of, say, "Epigrafia" is indeed "E", etc [18:40:43] right, 1) make this a browser/whatever test, 2) upgrade beta cluster first with time for tests to run [18:40:48] siebrand: i was under impression that we have such tests already, though [18:40:49] right [18:41:09] ah [18:41:13] we have tests, of course [18:41:23] did they fail? [18:41:26] but they are naturally run with the newer version for both strngs :D [18:41:28] so they worked [18:41:31] or was a human faster in the reporting? [18:41:35] the data in the database is bad [18:41:35] hah [18:41:46] chrismcmahon: ^^^ [18:42:02] so yeah, a browser test sounds like the only way here [18:42:07] chrismcmahon: see bit where MatmaRex talks about how the tests aren't really testing reality [18:42:32] ugh [18:43:13] s/bad/incompatible/ [18:43:48] Reedy's busy doing other fire fighting right now, btw [18:44:56] pin all the stuff! [18:46:29] no kidding [18:46:31] ;) [18:46:51] Reedy: you understand the issue/solution from backscroll? [18:47:18] Sounds like I somewhat guessed the cause earlier too [18:47:20] ;) [18:47:43] yeah [18:47:50] hey, I was one of the affected user by the LabsDB issue, and when I try to log into Wikipedia it asked me to change my password. I tried to change it, but it was "aborted by an Extension". [18:48:02] csteipp: ^^ [18:49:05] tyteen4a03: That should only happen if you're setting it to the same password as one that was leaked... I'm assuming your setting it to something new? [18:49:48] csteipp, I forgot to read and ended up typing my old password for one time, then I try to change it to a new password and it rejected it [18:51:01] tyteen4a03: Just to check, you're not logged in, right? If it successfully reset, you would be logged in. [18:51:15] csteipp, I am not [18:51:23] greg-g: I'm not sure how that issue might play out on the beta cluster, if at all. [18:51:45] Cool. So it should still have your old password. Did you try going to the login page and logging in again? [18:51:52] chrismcmahon: is category collation not enabled on any of our non-enlish wikis in beta? [18:51:57] it should be, no? [18:52:00] Or did you just click back and refill the form? [18:52:00] csteipp, yes, then it still rejected it [18:52:06] csteipp, I'll try again [18:52:15] tyteen4a03: Rejected the password change? Or the login? [18:52:34] csteipp, password change [18:52:40] csteipp, tried again, still aborted by extension [18:52:52] greg-g: today is the first time I have heard the term 'category collation' in my life. Also, we're not actively using non-English wikis on beta as of right now, although we have open bugs for that. [18:52:52] make sure the 'original password' is your old one? [18:53:04] chrismcmahon: they exist though, yes? [18:53:19] I know we aren't actively running tests against them, but they are there waiting [18:53:21] greg-g, the original password field was automatically filled in - you want me to fill it in manually? [18:53:26] greg-g: take that back, I think somewhere we use dewiki, but that might have been a one-off [18:53:35] tyteen4a03: no, just making sure you didn't accidentally overwrite it [18:53:45] greg-g, ok\ [18:53:46] Oh, tyteen4a03, did you type your old password into the form? Or was it prefilled? Some password managers will mess things up by prefilling your newly set password [18:53:57] yeah, that oo [18:53:58] greg-g: if they don't they could be made to [18:53:59] +t [18:54:04] csteipp, the old password field was prefilled, the new password(s) are not [18:54:29] tyteen4a03: Can you try it again, and type in your old password? [18:54:40] chrismcmahon: right right, so, basically, we can use beta cluster to catch this kind of issue, by testing for correct category collation. I don't know all the details, but it appears MatmaRex does, and that there is a test somewhere that kind of tests it [18:54:46] csteipp, instead of letting it prefill itself? Ok [18:55:11] csteipp, nope, still doesn't work [18:55:12] greg-g: I missed the part about a test somewhere, I don't think there is one. [18:55:35] tyteen4a03: That is strange.. which wiki is this? [18:55:35] greg-g: chrismcmahon: there are no browser tests, the cateogry collation framework has some phpunit tests [18:55:42] csteipp, English Wikipedia [18:55:43] chrismcmahon: see around :40 above [18:56:07] which among other things check if the first letters are generated correctly (this is surprisingly non-trivial for languages other than english) [18:57:06] MatmaRex greg-g presumably those first letters are in a database somewhere? [18:57:16] things exploded because the real first-letters were generated from the data from the database (which was generated with older ICU version) by the newer ICU version [18:57:31] the tests worked since they used the same version for both things [18:57:55] chrismcmahon: it's complicated. basically, they're not [18:58:44] chrismcmahon: MW has a hardcoded list of "basic" characters (or character groups) which are considered to be separate letters for all languages MW supports [18:59:06] chrismcmahon: and hardcoded lists of similar characters/groups to be added and/or removed for each specific language [18:59:09] why can't i disable collapsing of the sidebar groups anymore? [18:59:33] tyteen4a03: What's your wiki username? [18:59:40] csteipp, tyteen4a03 [18:59:45] jackmcbarn: which groups? [19:00:00] chrismcmahon: the first list is serialized/first-letters.ser (generated by maintenance/language/generateCollationData.php), the second lists are in Collation::$tailoringFirstLetters [19:00:07] !master includes/Collation.php [19:00:07] https://git.wikimedia.org/blob/mediawiki%2Fcore.git/master/includes%2fCollation.php [19:00:08] there used to be a preference to disable the autocollapse completely, for all of them [19:00:11] now it's gone [19:00:16] MatmaRex: OK, understood. [19:00:18] i remember seeing a gerrit commit about it a while ago [19:00:27] and since a deploy just happened.... [19:00:29] tl;dr: it's a huge mess :D [19:00:43] jackmcbarn: yes, that preference was removed, very few people used it [19:00:56] * jackmcbarn grumbles [19:01:20] jackmcbarn: just expand them all and live with it :)? the state is remembered between pages [19:01:34] can you link me to the commit that removed it, so i can rebuild it with user javascript? [19:01:46] i know, but that's in a cookie, and not my account preferences [19:02:18] i can, sure, but i don't think it's going to be helpful [19:02:28] checking the pref just made the code that adds collapsing not load :) [19:02:29] one sec [19:03:53] jackmcbarn: that was removed entirely from the Vector extension in https://gerrit.wikimedia.org/r/#/c/83590/ and added back in MediaWiki core in https://gerrit.wikimedia.org/r/#/c/83591/ (without the preference and some old experimental code that was not used anyway) [19:06:21] tyteen4a03: Yeah, the only time that error will be returned on enwiki is if your new password is the same as another password you already have set. Can you try setting it to something totally different? You can change your password again after you login, but I just want to make sure that's not the issue. [19:08:13] MatmaRex: thanks [19:09:17] MatmaRex: SOOOOOOOOOOOOOOOOOOOOOO [19:10:56] Do I just need to updateCollation.php with --force for all customised wikis? [19:11:14] Reedy: afaik yes [19:11:22] tim would know for sure [19:11:25] Let's pick a small wiki to test then [19:11:28] just try it on a smallish one and see [19:11:28] :D [19:12:26] Which is small? [19:12:28] csteipp, moment [19:13:00] csteipp, yup, that worked [19:13:13] MatmaRex: plwikivoyage? [19:13:25] Reedy: should be good, it's pretty tiny [19:13:36] does it have uca collation? [19:13:44] uca-pl [19:13:44] yeah [19:13:58] Selecting next 10000 rows... processing...1516 done. [19:13:58] 1516 rows processed [19:13:59] lol [19:16:07] if only i could read polish [19:17:26] greg-g: MatmaRex does, hence choosing pl :P [19:17:55] did it work? [19:17:56] want me to check it? [19:18:00] please! [19:18:17] :) [19:19:11] looking at https://pl.wikivoyage.org/wiki/Kategoria:Ukryte_kategorie , it doesn't seem to have helped :/ [19:19:18] or category purging doesn't work [19:20:50] Reedy: ^ [19:20:55] blugh [19:21:28] (these two subcats should be under "U" and "Z", respectively) [19:26:13] MatmaRex: also broken here, right? https://pl.wikivoyage.org/wiki/Kategoria:Chorwacja [19:26:39] greg-g: yeah, polish has no fancy rules for first letters, common sense applies :) [19:28:15] * greg-g nods [19:28:17] just making sure [19:28:29] MatmaRex: so, next ideas? :) [19:29:31] switch back to the working package version and call in cavalry (ie, tim)? i am not an expert on how this works, just slightly less clueless than most people :) [19:29:43] running the script should definitely fix the issue afaik [19:29:55] hmm, let me test something [19:30:25] if a cateogry was added to a page after package upgrade, it should sort correctly in that cateogry [19:30:28] tyteen4a03: glad to hear. So the password you were originally trying to use matched a password hash that was potentially leaked.. so I would strongly encourage you to pick a new password, and use that for your wikipedia accounts :) [19:31:03] apparently it doesn;t D: https://pl.wikivoyage.org/wiki/Kategoria:Chorwacja [19:31:31] greg-g: Reedy: ^ [19:31:35] this sounds bad [19:31:41] ugh [19:31:45] it also sounds like, lemme find the bug [19:32:02] apergos: possible to downgrade those icu-related packages? [19:32:18] (it sounds like run `php maintenance/generateCollationData.php` and replace the first-letters.ser file in production with the one it generates) [19:32:21] Reedy: ^ [19:32:35] I'm so not here (10:30 pm) [19:32:49] (and also run updateCollation.php later) [19:32:56] but, this would really suck [19:33:08] apergos: sorry, ignore [19:33:09] because the first-letters file is supposed to be compatible with everything [19:33:19] no worries, I will catch up on it all tomorrow [19:33:24] or so i was told by tim and bawolff, i think [19:33:41] mutante: where are you physically now? still awake? [19:34:25] Reedy: can you try that? run generateCollationData.php on the cluster (or with exact same packages and settings), commit new file in wmf branch, deploy [19:34:30] MatmaRex: That doesn't exist [19:34:38] what? [19:34:47] !master maintenance/language/generateCollationData.php [19:34:47] https://git.wikimedia.org/blob/mediawiki%2Fcore.git/master/maintenance%2flanguage%2fgenerateCollationData.php [19:34:50] maintenance/generateCollationData.php [19:34:55] maintenance/language [19:34:59] * Reedy kicks MatmaRex [19:35:33] reedy@tin:/a/common/php-1.22wmf21$ mwscript maintenance/language/generateCollationData.php mediawikiwiki [19:35:34] Unable to find allkeys.txt. Download it and specify its location with --data-dir=. [19:35:34] Unable to find ucd.all.grouped.xml. Download it, unzip, and specify its location with --data-dir=. [19:35:34] You are using ICU 4.2.1, intended for Unicode 5.1. Appropriate file(s) should be available at: [19:35:34] * http://www.unicode.org/Public/UCA/5.1.0/allkeys.txt [19:35:35] * http://www.unicode.org/Public/5.1.0/ucdxml/ucd.all.grouped.zip [19:35:50] yeah, it needs those [19:35:57] i am not promising this will work, btw :) [19:36:04] Except tin can't access the interwebs [19:36:23] (i spent a lot of time making that error message useful a few motnhs ago. a lot.) [19:40:32] first-letters-root.ser gets a lot smaller MatmaRex [19:40:49] -rw-r--r-- 1 mwdeploy mwdeploy 258685 Oct 10 16:40 first-letters-root.ser [19:40:55] -rw-rw-rw- 1 mwdeploy mwdeploy 188294 Oct 10 19:40 first-letters-root.ser [19:40:58] hmmm Reedy, actually [19:41:07] why does this say ICU 4.2.1 [19:41:23] per bug https://bugzilla.wikimedia.org/show_bug.cgi?id=46036 we were using 4.8 or something [19:41:53] ii libicu42 4.2.1-3ubuntu0.10.04.1 International Components for Unicode [19:41:53] ii libicu48 4.8.1.1-3 International Components for Unicode [19:41:58] ii php5-intl 5.3.10-1ubuntu3.8+wmf1 internationalisation module for php5 [19:42:03] I guess php5-intl isn't... [19:42:53] I guess we're not using paravoids backported one now.. [19:43:26] Reedy: *don't ask me about why it that file got smaller, but the bulk of that file is characters from exotic scripts, so most likely removing 90% of it wouldn't have noticeable consequences) [19:44:40] Reedy: do you know which were the old and new icu package versions? changed in the minor shouldn't change the format of the sortkeys afaik [19:44:49] (aka, shouldn't break this for us) [19:44:56] I think mutante might have posted it in the backscroll [19:45:11] this looks like something accidentally downgraded wmf's icu, lol [19:45:33] http://paste.debian.net/plain/55306 [19:48:31] * MatmaRex has no idea how linux package managers work, i don't even have a linux on my primayr computer [19:48:53] I was just trying to see if php5-intl would tell us what icu version it's iusing [19:50:16] Looks like all the eu ops are /awawy [20:02:49] Reedy: assuming nothing is happening now, can you sum it up on https://bugzilla.wikimedia.org/show_bug.cgi?id=55565 ? were packages downgraded during upgrading or what? [20:03:03] I'm not sure [20:03:03] since the script output and the pastebin above points to 4.2 being used [20:03:12] php5-intl definitely was [20:03:14] which shouldn't be true [20:05:52] Dcoetzee: yo [20:06:13] oops wrong chan [20:24:27] MatmaRex: DonRumata has questions about the collation/php upgrade thing [20:25:25] Pay attention on https://bugzilla.wikimedia.org/show_bug.cgi?id=55565 [20:26:00] DonRumata: there's quite a lot of detail in channel logs... you can see logs linked from /topic [20:27:41] ok. [20:37:49] jeremyb: DonRumata: sure, i am not an expert though, really [20:37:57] (and i'm off in 15 minutes) [20:38:20] MatmaRex: i certainly ain't [20:38:35] but also it's hard to get DonRumata to actually ask anything [20:38:43] so maybe no answers to give [20:38:45] ! [20:38:45] an exclamation point [20:38:52] thanks wm-bot! [20:38:52] Hey jeremyb, you are welcome! [20:39:53] I think that bug 55565 isn't related with collation at all. Wrong section headers are present in all categories on mediawiki.org. F.e. https://www.mediawiki.org/wiki/Category:Manual [20:40:36] DonRumata: mw.org is using an uca collation. [20:41:24] 'mediawikiwiki' => 'uca-default', [20:41:53] and the bug only manifests itself on wikis with uca collations [20:44:07] oh, wow, now i get to see bug in action :) [20:46:30] 'ruwiki' => 'uca-ru', [20:55:03] (guys, last call for question if i can help with anything, i'm off to sleep in a few minutes :) ) [20:58:14] MatmaRex: slap wel [20:58:27] He's Polish, and that's Dutch. [20:59:09] twkozlowski: i only speak none of those languages [20:59:14] let me think... [20:59:24] (also, i couldn't remember where he was from) [20:59:41] can't think of it [21:00:13] 'night. [21:00:14] ah, dobranoc [21:00:21] :D [21:00:27] do you also use dobra vece? [21:00:36]