[00:09:14] We just started our Image Optimization script on our production file server. Looking pretty good so far :) once we have more numbers, we can try to pimp it to you guys [00:09:15] ;) [00:09:54] initial stats: looked at 22,000 images and optimized them by about 400 megs [00:22:09] Sean_Colombo: Lossless compression or what? [00:23:05] Sean_Colombo: quite often our users reupload pngs etc after losslessly compressing them [00:23:13] I attacked SVN not so long ago ;) [02:10:35] !log LocalisationUpdate completed (1.19) at Wed Mar 28 02:10:34 UTC 2012 [02:10:42] Logged the message, Master [04:20:55] how will it impact server performance if I redirect all users from China Mainland to https page in Common.js? [04:21:19] or a default gadget [04:21:28] that would be kinda pointless because they would hit the http version first [04:21:39] and would result for double calls on all non https calls [04:22:41] p858snake|l: that's only for the first page they visit [04:23:05] and any sunsquent pages if they don't stay on https [04:23:23] why won't they stay on https automatically? [04:23:57] depends, generally yes but you never know [04:27:39] liangent: There are no server/load concerns. [04:27:55] But you shouldn't do something like that without appropriate community consensus. [04:33:00] Joan: there would be increased load on the https proxies if we did that on a wmf project... [04:33:16] although if would be enough to do any harm would be a question for ryan [04:33:59] http://lists.wikimedia.org/pipermail/wikitech-l/2012-March/058808.html [04:36:32] and doing it server side would be much nicer than on wiki also [09:11:10] What's the usual turnaround for "shell" bugs? [09:12:10] Normal priority, that is. [09:17:58] depends on the aligment of the moons and sun. We don't really have a set SLA time frame for dealing with shell requests [09:19:51] :P Okay. Is it a question of days, weeks, months though? [09:25:40] generally weeks at the moment [09:25:47] What is the request for? [09:57:57] p858snake|l: Sorry, got distracted. It's for a maintenance script to be run, to fill some holes in the enwiki_p. Not sure how difficult that is, I have to say. [10:00:27] What script? [10:01:23] https://bugzilla.wikimedia.org/show_bug.cgi?id=34981 [12:48:12] can somebody change wgMainPageTitle for be.wikisource.org? Not enough time to fill a bug :*( [12:48:38] it should be the same as in https://be.wikisource.org/wiki/MediaWiki:Mainpage [14:05:21] !seen binasher [14:12:02] Jarry1250, then the usual waiting time is months or year :) [15:05:29] hallo. i'm helping with an edit-a-thon in the National Library of Israel. will it possible to create a lot of accounts here or will they be blocked after some accounts? [15:06:12] aharoni, for the English WP? [15:06:24] Hebrew or English Wikipedia [15:06:29] there is a userlimit per IP, I think 6 [15:06:32] we can lift that per wiki [15:06:34] ugh [15:06:50] have a sysop onboard they have no limit [15:07:21] when is that edit-a-thon taking place? [15:07:41] hmm, he disconnected.. [15:07:44] ah, there you are [15:07:45] when is that edit-a-thon taking place? [15:07:52] it begins now [15:08:10] ok, swift action then ;) [15:08:43] hrm [15:08:54] how many accounts (roughly?) [15:09:03] and from which IPs ? [15:10:28] Reedy, can you increase the account creation throttle limit for a few hours on hewiki and enwiki, per aharoni? [15:10:30] 192.114.7.2 [15:11:05] if it's not hundreds, I can set up the accounts myself for en.wiki [15:11:08] about 50 accounts... if it's not too much :) [15:11:20] and it will be SUL, right? [15:11:38] yes [15:12:35] what's your accountname aharoni? [15:12:40] amire80 [15:13:06] then do it yourself ;) [15:13:13] let me see [15:14:22] i'm a sysop, so maybe i can open them myself, but we would love to teach the people to open the account myself [15:14:31] aharoni, as a fallback I set accountcreator rights for you, so you can create unlimited accounts through http://en.wikipedia.org/wiki/Special:UserLogin/signup if all else fails [15:15:37] but can the limit be increased or something? or maybe just for this IP? [15:16:11] like Platonides said, sounds like something for Reedy [15:19:59] Reedy was connected to irc some minutes ago, but seems to not have noticed this :/ [15:21:08] Hi [15:21:12] I took over Aharoni [15:25:17] , until Reedy will notice this coversation, can you set some more account creators? [15:25:38] TomerA, nah, just a sysop like you who lurks this channel [15:25:46] oh, more account creators [15:25:47] sure [15:25:52] you can do that yourself too [15:25:52] :) [15:26:03] How? [15:26:16] http://en.wikipedia.org/wiki/Special:UserRights [15:26:27] Ahh, I know how to create accounts through my account. But I think there's still a limit [15:26:40] And I'm not a sysop on he.wiki [15:26:58] as long as you are on en.wiki you can create the accounts there, and they will be SUL [15:27:03] so they can log in on he [15:27:18] Ok. [15:27:31] But isn't there a limit? [15:27:46] yeah, but you bypass that by having the accountcreator flag on your account [15:27:51] which you have since 10 minutes [15:28:15] it's settable by sysops [15:28:39] maybe hashar is connected, he would also be able to change it [15:28:46] How can I have it if I haven't given my username yet [15:28:56] (btw, also make sure nobody gets blocked, or you'll have autoblock for everyone there) [15:29:10] what's your accountname aharoni? [15:29:10] amire80 [15:29:30] Aharoni has left this computer [15:29:34] I'm a different person :) [15:29:43] With a different account [15:29:46] ah [15:29:54] what's yours? [15:31:07] Ohai [15:31:12] Platonides: I am here [15:31:21] [[User:Deror avi]] , [[user:esh77]], [[user:תומר א.]] [15:32:05] hashar, see scrollback :) [15:32:14] is it possible to get account creation throtling suspended for an hour or so on en.wiki and he.wiki for 192.114.7.2 ? They're having an edit-a-ton [15:32:37] yeah [15:32:37] and a severe lack of planning ;) [15:32:43] ahhh and Reedy is there [15:32:44] Indeed. [15:32:44] can just steal the config that's there overly set [15:32:50] Severe lack of planning [15:33:04] just when I was like "oh the heck do we add an exception to the account creation throttler" [15:33:38] Until when? [15:33:44] 4 hours [15:33:50] hashar: search for creation in commonsettings ;) [15:34:18] And there's another address we need to add. [15:34:40] !log reedy synchronized wmf-config/CommonSettings.php 'Bump account creation limit per request on -tech' [15:34:41] Logged the message, Master [15:34:50] !log srv221 is full [15:34:52] Logged the message, Master [15:35:13] 192.114.7.39 [15:36:06] hashar: Btw, thanks for the nice, expanded commit message. My git-fu is weak and I didn't know review would just recycle my local commit message. :) [15:36:14] !log reedy synchronized wmf-config/CommonSettings.php 'Bump account creation limit per request on -tech' [15:36:16] Logged the message, Master [15:37:09] Coren: git commit --amend lets you change both the commit message and the content of the commit [15:37:33] RoanKattouw: Noted. Danke. [15:37:56] Coren: yeah you can rewrite history :-D [15:38:36] Coren: specially while the commit has not be merged. It is still possible after a merge but unwanted cause that will cause too much mess in the repository (all descendants sha1 get recomputed) [15:38:57] Coren: the process is usually: git-review -d 1234; git commit --amend; git-review -f [15:39:20] hashar: You can't amend commits post-merge, Gerrit doesn't let you [15:39:33] Coren: you could even edit some files, add them (with git add) then git commit --amend and git-review. That is how you generates a new patchset [15:39:44] So, are we good? [15:39:59] RoanKattouw: even when allowed to push anything ? :D [15:40:02] So, that was my first review and I went "wait, no editor for a message? Oh. That's going to be terse." [15:40:13] hashar: Yes, force push is disabled [15:40:18] TomerA: yup looks like Reedy allowed you to create ton of accounts [15:40:34] Even at github you have to use git push -f to rewrite the history of things that have already been published [15:40:34] TomerA: !!og reedy synchronized wmf-config/CommonSettings.php 'Bump account creation limit per request on -tech' [15:41:00] RoanKattouw: oaah force push. I had a mail exchange about that with Chad cause gerrit kept forbidding me a push [15:41:06] Thanks [15:41:13] RoanKattouw: ends up I did not check the "force" checkbox :D [15:41:57] TomerA, if I were you, I would ask anyone creating a new account to make a note on their userpage or talk that they are in a hackathon. At least for en.wiki, and have someone idle in #wikipedia-en and #wikipedia-he in case something goes wrong (I forsee blocking/autoblock mayham) [15:42:04] Will any of you guys be here in the next 60 minutes or so? [15:43:29] , most people from he.wiki are here anyway [15:43:45] and I think there won't be much edit in en.wiki anyway [15:44:07] TomerA: I am around here for thenext 45 minutes or so at least [15:44:15] can't guarantee I will stick there later though :-( [15:44:45] hashar, thanks. I'll beep you if there be any problem. [15:45:12] I have to log off to get back to the guys. [18:49:33] !log catrope synchronizing Wikimedia installation... : Bugfixes for ArticleFeedbackv5, ArticleFeedback and ClickTracking [18:49:35] Logged the message, Master [18:57:42] sync done. [19:56:03] is anyone around who can help me spelunk api.php as related to search? [19:56:27] Jeff_Green: What's up? [19:56:43] Jeff_Green: Shouldn't this go to #mediawiki ? [19:56:54] !log Running a patched version of l10nupdate that rebuilds the localization cache [19:56:55] hoo: good question. I don't know. [19:56:55] Logged the message, Mr. Obvious [19:57:31] RoanKattouw: I'm working on a QA script for search/lucene, and I could use some help figuring out the format of a subrequest between api.php and the lucene cluster [19:57:54] specifically how is a request like the following routed and formatted: [19:58:10] http://es.wikipedia.org/w/api.php?action=opensearch&search=d&namespace=0&suggest= [19:58:37] i *believe* since suggest= is defined it's a prefix query [19:59:13] but I haven't managed to find the spot in mw code where the backend URI for such queries are formatted [20:05:11] Wait, can you comment in gerrit without having to review? [20:05:17] Yes [20:05:20] Give a review score of 0 [20:05:30] kk [20:09:25] Since Saudi Arabia enfoces censorship on certain Wikipedia articles, I was wondering if it's acceptable to have a bot that scans the Arabic edition looking for blocked pages. [20:09:54] OsamaK: That's a local community question, not a technical question, right? [20:10:12] RoanKattouw, not really [20:10:23] RoanKattouw, is is generally acceptable to have bots scanning all articles? [20:10:42] OsamaK: What exactly do you mean with scanning? [20:11:10] hoo, it would load them to check. [20:11:27] OsamaK: Check what? Whether they're accessible? [20:12:04] hoo, yes. a "this page is blocked" message is shown if it's blocked. [20:12:56] OsamaK: Well, I guess if you follow the usual rules for bots it should be fine [20:13:15] (set a meaningfull user agent and limit the requests per time) [20:13:27] If you're hitting page views only, you'll probably be fine [20:13:47] When in doubt, don't run parallel requests [20:14:43] i see [20:15:46] btw, if anyone is interested in seeing what's currently known to be blocked: thttps://en.wikipedia.org/wiki/Wikipedia:List_of_articles_censored_in_Saudi_Arabia [20:16:31] Theory of evolution :P [20:17:29] hoo, yeah, that's really really stupid of them. [20:18:07] I'm interested in seeing whether there are political articles that are blocked. [20:19:27] 172,570 pages... not to much [20:19:45] really? :) [20:20:02] https://ar.wikipedia.org/wiki/%D8%AE%D8%A7%D8%B5:%D8%A5%D8%AD%D8%B5%D8%A7%D8%A1%D8%A7%D8%AA ... how do they block, btw? [20:20:16] If you use the API it works fine, no? [20:20:56] No, they only block the 'main' URL. all alternative urls work. [20:21:16] for example, http://ar.wikipedia.org/w/index.php?title=evolution should work [20:21:52] Of course, it's totally meaningless in HTTPS. [20:22:10] heh [20:22:12] Man muscle memory; that's at least the 7th time today that I 'dig commit' :-) [20:22:41] it's just a way for them to say what they want us to, but they cannot (currently) really control it. [20:22:56] * want us to read [20:25:34] OsamaK: From my machine it would need about 33h to get all articles (assumin .7 seconds per article) [20:27:25] yeah, it isn't urgent.. [20:29:56] Hm With the dabase wrapper functions that take a conditional, an array will /and/ all the conditionals. How does one /or/ conditionals? [20:30:51] Coren: $dbr->makeList( LIST_OR, array( ... ) ); [20:31:08] Wait, swap that [20:31:14] makeList( array( ... ), LIST_OR ) [20:31:26] Found the function doc. Thanks. [20:35:52] Coren: I added patchset 4 for your convenience :) [20:36:06] saper: Oh. How... convinient. [20:36:21] * Coren tries to figure out how to merge that into his own tree now. [20:37:30] Coren: are you your changeset branch? [20:37:38] just "git fetch" [20:38:09] saper: Hm. It either worked quietly, or it failed quietly. [20:38:40] and then merge [20:38:54] what does "git merge --no-commit gerrit" say? [20:39:19] fatal: 'gerrit' does not point to a commit [20:39:32] do you know what's the name of your remote? [20:39:44] ("git remote" lists them all) [20:39:55] I had followed the instructions of git/workflow to begin the amend. [20:40:09] gerrit and origin [20:40:10] using git-review? [20:40:20] ok [20:40:36] so "git fetch gerrit" is silent? [20:41:00] Ah, not, "git fetch gerrit" wasn't. "git fetch" was. :-) [20:43:18] But that didn't do anything close to what I was expecting. [20:43:21] !log LocalisationUpdate completed (1.19) at Wed Mar 28 20:43:20 UTC 2012 [20:43:23] Logged the message, Master [20:43:31] I'm gessing I started my fixes wrong. [20:43:41] * Coren stranges git. [20:43:47] strangles* [20:45:27] Ah, wait that worked [20:46:42] Coren: git fetch? [20:46:56] Coren: what does "git log HEAD ^HEAD^" say? [20:47:10] I think I just suceeded. [20:47:30] Looks like. [20:50:18] saper: Can you make sure I didn't manage to undo your fixes? [20:51:44] Getting used to gerrit will take some doing. Even the terminology confuses my old-school habits. :-) [20:53:02] GAH! [20:53:24] I managed to rollback my own changes [20:53:51] * Coren cries [21:01:18] Coren: how? [21:02:01] Coren: can you try "git checkout fcc05dee9b93e080b13fc4b0d5b83a1c75d34362" ? [21:02:09] saper: I wish I knew. I think the fetch I did trampled the changes I had. [21:02:27] fetch does not mungle your working tre [21:02:56] Well, my changes are gone, and the thing I've did a commit then review to seem to be a null change. [21:03:17] Or was I, like, switched from tree to tree? [21:03:44] Hm. I'm not (no branch) [21:03:49] I'm on [21:04:30] yes [21:04:37] you always jump from tree to tree [21:04:51] "git reflog" gives you (local) history of your jumps [21:05:24] so you never really loose your "old" commits - they are purged sometimes by some git garbage collection magic [21:12:48] Editors are whining about edit counters not updating -_-. They want an ETA and preferably one that has everything normal by March 31 (Disambiguation challenge?). See http://enwp.org/WP:VPT#Toolserver_replication_lag [21:14:36] Dispenser: Then they should send their whines to the toolserver ops, not the WMF ops ;) [21:15:06] Send us a dump first [21:15:31] ? [21:16:20] Dispenser: what do you need ? [21:16:31] nothing [21:19:49] Well a scape goat, I made the blunder of giving out my internal estimates and its been (incorrectly) published in the Signpost and people are asking me questions like I know what's going on with the ALTER TABLE [21:20:21] what is the difference between http://wikipedia.org and https://wikipedia.org (http and https) [21:20:33] ssl [21:20:39] Dispenser: Sorry, realized you're one of the toolserver techs after reading that enwiki drama [21:22:13] Dispenser: hrm, not sure what i can do ... [21:22:16] Except they can write better and quicker in their secondary language than I can in my native [21:22:46] It's a database thing [21:22:53] And AFAIK Asher is out [21:23:14] yeah, very not reachable [21:23:26] see if you can find domas ? [21:25:37] ssL? [21:25:54] Guest44527: Encryption [21:26:01] Guest92901: have a look at http://en.wikipedia.org/wiki/HTTP_Secure [21:26:35] Guest92901: https basically mean the communication between your computer and our server is encrypted so nobody should be able to discover what you read on wikipedia :) [21:26:52] Guest92901: that also protected your password when it is sent from your computer to our servers [21:32:02] Coren: https://bugzilla.wikimedia.org/show_bug.cgi?id=20267#c6 <-- somehow I like this idea of re-using ipb_auto [21:34:20] nighty~ [21:38:33] whatsup [21:38:49] lesliecarr: ? [21:39:10] if database help is needed by the toolserver guys [21:39:35] uh oh [21:42:37] !log awjrichards synchronized php/extensions/MobileFrontend/javascripts/banner.js 'r114576' [21:42:39] Logged the message, Master [21:42:48] !log awjrichards synchronized php/extensions/MobileFrontend/javascripts/application.js 'r114576' [21:42:49] Logged the message, Master [21:43:04] !log awjrichards synchronized php/extensions/MobileFrontend/templates/ApplicationTemplate.php 'r114576' [21:43:05] Logged the message, Master [21:43:14] !log awjrichards synchronized php/extensions/MobileFrontend/MobileFrontend.body.php 'r114576' [21:43:16] Logged the message, Master [21:43:25] !log awjrichards synchronized php/extensions/MobileFrontend/javascripts/banner.min.js 'r114576' [21:43:27] Logged the message, Master [21:43:54] !log awjrichards synchronized php/extensions/MobileFrontend/javascripts/application.min.js 'r114576' [21:43:56] Logged the message, Master [21:44:44] !log awjrichards synchronized wmf-config/CommonSettings.php 'Bumping mobile frontend resrouce version' [21:44:46] Logged the message, Master [21:46:31] Coren: https://bugzilla.wikimedia.org/show_bug.cgi?id=20267#c10 [22:35:45] apergos: hey, can you help with getting Toolserver a fresh en-wp dump? http://lists.wikimedia.org/pipermail/toolserver-l/2012-March/004866.html Or maybe you know what's going on here and my request is moot [22:38:35] notpeter: do you happen to know whether Asher is working this week? looks like there's an open request for him ^^^ [22:41:10] sumanah: don't think he's back this week [22:42:33] sumanah: he's at some conference with a hardware vendor in east asia. I think he'll be rather busy with that [22:42:45] oh, is that where mutante is as well? [22:42:56] ok, how should http://lists.wikimedia.org/pipermail/toolserver-l/2012-March/004866.html be followed up on? [22:43:02] DaBPunkt is waiting for it, I gather [22:44:01] hrm... I would say wait for asher to get back, or send him an email specifically about it [22:44:27] or someone else with some significant mysql chops [22:44:34] DaBPunkt has, I think... is there anyone else who could help DaBPunkt out? [22:45:07] domas_ ? perhaps ben or jeff. [22:45:17] it looks like all of Toolserver is suffering here, including a lot of tools that en.wp depends on [22:45:24] edit counters are the least of it [22:49:12] sumanah: Heard of the Dab challenge project? (http://enwp.org/WP:DPL) JaGa figures out which links have been eliminated and awards points in his monthly contest [22:49:24] they can use faker! [22:49:31] faker fixes all replags [22:49:41] or full flash [22:49:44] full flash fixes all replags too [22:50:09] (It is still not a good idea to name a enwp-project like me…) [22:51:27] (I've got a tool named dabfix) [22:52:33] uh oh sha1 column [22:52:34] Ok. I call defeat. I managed to lose my fixes and confuse myself. Anyone available for a git/gerrit tutor session for 5-10 mins? [22:53:05] No, I had not heard of that project, Dispenser [22:54:50] I need to head off, but DaBPunkt it sounds like maplebed and Jeff Green might be people you could nicely nag to get you a fresh en-wp dump [22:54:53] good luck [23:00:56] maplebed: JeffQuassel: One of you have a moment for me and the toolserver? [23:02:10] eh? [23:02:21] I am not Jeff Green. [23:02:59] DaBPunkt: ^ [23:03:00] JeffQuassel: oh, sorry :) [23:03:06] np [23:05:11] DaBPunkt: Can't we just kill ALTER TABLE, let the TS catch up and then import the schema update from a dump? [23:06:10] DaBPunkt: can you tell me which server the toolserver is replicating from? [23:08:32] maplebed: db36, why? [23:09:08] Dispenser: no, because the ALTER-table adds a field [23:09:31] last I heard the TS db was behind because its master was still running the alter table. [23:09:39] but mw-1.19 isn't deployed yet [23:10:39] maplebed: yes, but we also need a fresh dump because our database is corrupted; and a fresh dump you contain the altered-table already, what is s bonus here [23:10:54] you → would [23:11:18] I'm also searching for docs on the process by which a toolserver dump is generated. haven't found them yet. Do you know where they are? [23:13:14] I get the feel it dump and import the table would be faster than this ALTER-table [23:13:27] maplebed: no. Normaly a slave is depooled, the slave is stopped, the master-position is noticed, the dump is made, the slave is started again (and later re-pooled) and then the dump will be copied to the toolserver [23:13:32] I get the feeling dumping and importing* [23:14:06] Coren: can I help you with recovering your changes? [23:14:22] can you post me your "git reflog" to some paste service please? [23:15:14] DaBPunkt: it's the parts around 'the dump is made' and 'copied to the toolserver' that I want more detail for - are any columns left out? are all tables dumped? what format is used for the dump? etc. [23:15:53] trainwreck [23:16:47] maplebed: we get a complete dump for normal ( you can filter the job-table if you like, but it is not worth it IMHO); and the format is the normal mysqldump-format [23:19:51] heh [23:22:40] DaBPunkt: lemme see what supporting docs I can find. [23:27:10] ok [23:59:52] gn8 folks