[02:45:59] !log LocalisationUpdate completed (1.20wmf6) at Fri Jul 13 02:45:59 UTC 2012 [02:46:15] Logged the message, Master [08:02:29] Hi! test.wikipedia.org is still broken: I can't upload nor access the edit pages, such as: http://test.wikipedia.org/w/index.php?title=Sandbox&action=edit [08:02:34] The error message is: Error: ERR_ZERO_SIZE_OBJECT, errno [No Error] [08:02:41] can anybody do something, please? [08:07:44] so I think we prefer test2.wikipedia these days (someone can correct me if that's wrong) [08:07:53] doh [08:07:53] also that url shows me the edit form [08:07:56] what is test2 ? :-D [08:08:02] go there and see :-D [08:08:33] save of the sandbox worked for me [08:08:36] I mean, test is bound to srv193 and runs MediaWiki from NFS [08:08:38] peter17: [08:08:40] test2 I have no idea [08:08:48] yes, test2 is supposedly set up "correctly" [08:08:52] ohh [08:08:57] that nfs mount thing is a horrible hack [08:09:22] that let us verify our change before syncing it to everyone [08:09:45] ohh so test2 is just a production wiki with specific configuration g [08:09:48] good to knowè [08:10:19] as for test.wikipedia.org , Apache keep segfaulting, I am not sure if we have a ticket for it [08:10:29] no idea [08:10:41] apergos: hum... weird... I can't display the edit field. I made some tests yesterday with jeremyb and it was failing for both of us [08:10:57] odd [08:11:05] peter17: yeah that is a server side issue [08:11:07] but I did just edit the sandbox, check the hsistory ;-) [08:11:31] peter17: related bug https://bugzilla.wikimedia.org/show_bug.cgi?id=38366 [08:11:32] :) [08:11:40] and the bug is https://bugzilla.wikimedia.org/show_bug.cgi?id=38364 [08:12:53] while we are speaking about test.w, who can I ask to delete some files I uploaded there? [08:13:21] you don't have admin rights? I might, lemme see [08:13:50] yep looks like I can delete [08:15:37] great! can you please delete http://test.wikipedia.org/wiki/File:Test02.jpg and http://test.wikipedia.org/wiki/File:Test03.jpg ? [08:15:39] I'm coding on http://commons.wikimedia.org/wiki/Commons:Tools/KIPI_uploader and it uploaded more files than I meant to. [08:17:05] test03 appears to not exist [08:17:18] sorry, http://test.wikipedia.org/wiki/File:Test03.png [08:17:31] nice image [08:17:42] buy not free of rights :) [08:18:35] *poof* gone [08:18:55] thank you very much apergos [08:18:58] sure [08:19:46] hmm I don' t knnow digiKam [08:19:54] I wonder if it handles nef type [08:20:09] has libraw, looks good [08:20:25] it handles a lot of raw formats, i think [08:20:28] that's really great [08:20:42] I have a database of 35k photos [08:21:07] yeah, I have a ridiculous number too (many dups, only one or two of a series is worth anything) [08:21:07] the plugin we are coding will be more efficient than commonist and can take the metadata from that database [08:21:50] guess I'm going to have to check this app out, thanks for mentioning it! [08:22:47] I'm trying to improve some stuff on the plugin, but upload fails on test and I'm not allowed to upload on test2... [08:23:27] why not? [08:23:41] test2: "The action you have requested is limited to users in one of the groups: Autoconfirmed users, Administrators, Confirmed users." [08:23:49] test: same problem as displaying the edit field [08:23:50] you must not be autoconfirmed yet:-/ [08:25:36] here I'll make you a confirmed user [08:26:48] ok try that [08:26:51] great, thanks! [08:27:31] lemme know if the uploading works. [08:28:30] yes, works well, thanks: http://test2.wikipedia.org/wiki/File:Test02.jpg [08:28:51] yay [08:29:19] peter17: the test.wiki issue is tracked with https://bugzilla.wikimedia.org/show_bug.cgi?id=38364 feel free to CC to it [08:29:30] peter17: I have opened an internal ticket to track it with ops :-) [08:29:41] hashar: thanks! [08:29:58] peter17: which is https://rt.wikimedia.org/Ticket/Display.html?id=3264 but you can't access it :D [08:31:03] I have no idea how to debug seg faults :-( [08:33:08] hashar: can you tell me who should be looking at https://bugzilla.wikimedia.org/show_bug.cgi?id=38333 [08:33:19] I keep finding more breaches of our IP blocks [08:34:30] that would be platform engineering [08:34:36] aka Robla team [08:34:50] ie me, tim, reedy, demon, aaron schulz, chriss steip [08:35:32] I keep finding more examples [08:35:38] sDrewth: do you have checkuser ? [08:35:47] I am a checkuser and steward [08:35:52] \O/ [08:36:07] hence why I am here gently pushing [08:36:09] I love stewards people, you guys are doing an awesome job :-] [08:36:17] :-) [08:36:30] is my favourite with our current spambots [08:36:34] are you sure the user IP are indeed in the same range block ? [08:36:42] aha [08:36:57] 108.177.128.0/17 should be 108.177.128.0 to 108.177.255.255 [08:37:27] (just making sure, don't take it personally hehe) [08:37:38] I know my IP blocks, and I did check it via the tool to guarantee [08:38:53] not seeing the backend, means that I cannot poke and test beyond just what I see [08:39:04] indeed :-D [08:39:16] I am not really familiar with the blocking system nowadays [08:39:25] i am not even sure how we handle global blocks hehe [08:39:33] nor me, beyond pressing buttons [08:39:51] but I have access to logs so I can dig in ;:-]] [08:40:36] thaks, it is just weird to have at least two wikis failing to have blocks, and one at least 36 hours block was put in [08:40:56] <- just added a new example to reproduce [08:48:43] sDrewth, hashar: if rangeblocks aren't working I would move that to high or highest priority, that could be a very big problem depending on how widespread it is. [08:49:16] I am not sure whether it is one set not working, or all not working, hence why I am here rattling my cage [08:49:33] I would say better to get that checked ASAP. [08:49:59] Also you might want to mark that as a security sensitive bug so it's not publicly visible [08:52:59] so the range is blocked [08:53:09] Pine: my gut feel is that it is not global failure [08:53:19] but maybe a logged in user is still able to create accounts anyway [08:53:37] shouldn't be [08:53:51] other users would show up in the range check as creating the account [08:54:18] and not on two different wikis [08:54:21] And aren't rangeblocks supposed to be effective against accounts unless they're rangeblock exempt? [08:54:22] enWS and mw [08:56:10] I have no idea how the block system works really : / [08:57:11] okay, if you would do me the favour of at least poking someone about the hole, that would be great [08:57:44] doing /private [08:58:00] thanks :-) [08:58:24] it being Friday here, I didn't want it disappearing over the weekend [09:00:04] hashar, sDrewth: with your permission, I'll mark this bug as high-major since it has security implications and we aren't sure how widespread the effects are. [09:00:34] Pine: sure feel free to raise the priori [09:00:36] ty [09:02:04] Done. :) [09:02:06] never worry about my permission, people play with priorities all the time [09:02:41] I feel that priority is more decorative than anything else [09:02:50] like a light on a toy for Xmas [09:03:03] Well I hope it is enough to get people to pay attention to it before the weekend. [09:03:56] hashar: do you have a way of making this bug hidden due to the potential security implications? I'm not sure that I want spammers to attempt bypassing rangeblocks if they see this bug report. [09:04:18] Pine: assign it to the "Security" product in bugzilla [09:04:31] note that mails notifications for that bug have already been sent so the bug content is already public :-D [09:04:55] Thanks, moving it anyway. [09:05:56] I do not like the looks of this bug at all. [09:32:17] hashar: is there a way to show "profile info" on gerrit? like, if I want to know the name/commit name/email of someone knowing only one of them and even if the user doesn't have commits or reviews if possible? [09:40:16] Nemo_bis: no idea [09:40:34] I don't think there is a list of available users [09:41:37] Nemo_bis: what I do is that I use the add reviewer field [09:41:43] and enter something like: chadh [09:41:54] which would look up username / email address [09:42:10] you can probably look up on labsconsole Nemo_bis [09:42:18] https://labsconsole.wikimedia.org/wiki/Special:ListUsers that might help too [09:42:25] p858snake|l: good idea :) [09:42:36] Nemo_bis: the Gerrit users are in the Labs LDAP [09:42:49] yes but then I can't do much on the wiki [09:43:18] only know if a user exist, or maybe use the email feature (and only if I'm registered) [09:44:42] hashar: I suppose there's no way to access the LDAP though :) [09:45:24] Nemo_bis: probably not. [09:45:59] hashar: in gerrit there are user groups, right? and there isn't a list of members of groups? [09:48:00] Nemo_bis: let me check [09:49:34] Nemo_bis: can't find any [09:49:43] but that should be any members of the project-bastion group [09:51:50] sigh https://code.google.com/p/gerrit/issues/detail?id=873 [09:53:01] * Nemo_bis confused by https://code.google.com/p/gerrit/issues/detail?id=321 [09:53:10] isn't all that information inculded in git already [09:54:22] nem [09:54:29] Nemo_bis: what are you looking for any way? [09:54:41] I have access to the LDAP if you need a one time search [09:54:42] anyway, what I need is a way to easily link a starting point to contact/ping devs about issues on TWN: https://translatewiki.net/wiki/Template:Support [09:54:53] aghhhhhh [09:54:54] so [09:55:30] I guess the main point would be #mediawiki-i18n [09:55:42] on gerrit we could potentially create an i18n-reviewer group [09:56:44] I end up looking for names, user names or emails on the extensions' pages or histories, in bugzilla, on mediawikiwiki users list, on bugzilla, with gerrit blabla and none of this really works [09:57:17] lot of extensions already have a group of project owner/creator. Example: extension-Translate contains Raimond [09:57:27] another way is to find out the list of commuter for a given project [09:57:34] that will give you names and email :-D [09:57:42] hashar: the channel is useless, we already know how to reach people idling there; I don't understand the idea about group [09:57:44] though the commit name is not the mediawiki user name [09:57:58] well I am not sure what you are really looking for in LDAP anyway [09:58:15] and is there a way to query the list of projects or project owners? [09:58:17] if you want a point of contact foreach extension, that information is not really in Gerrit [09:58:35] the list of extensions hosted on WMF Gerrit is at https://gerrit.wikimedia.org/mediawiki-extensions.txt [09:58:38] there's not even a way to get the list of committers for a directory or file... [09:58:42] IIC [09:58:46] *IIRC [09:58:50] * Nemo_bis gotta go afk [09:59:00] going to lunch :-D [09:59:07] see you this afternoon Nemo_bis ! [10:13:54] out for lunch [13:26:06] so, hasharDeadmau5, in the end (sadly) the easiest way to find a dev is to use bugzilla DB, which is the most complete resource; autocompletion, for instance, can give most emails from (nick)names and viceversa [13:26:27] Nemo_bis: might be the best indeed [13:27:02] we have some components having default owner / cc [13:27:08] so I guess that is the best way to notify people [13:28:26] hasharDeadmau5: yes bt there's no way to automately link to anything, sigh [13:28:42] what do you mean? [13:29:01] hasharDeadmau5: I mean that on TWN we have hundreds of requests for clarification of interface messages [13:29:19] of course I can't spend ten min per message looking for the name of the dev etc. to poke him [13:30:42] so currently I'm linking the gerrit "profile" from the thread hoping it finds something, and in that case in a couple clicks one can send an email [13:32:13] ohhhhh [13:32:15] the message [13:32:23] well there is no really message owners [13:32:38] what would be possible is to find out who first introduced a message in git [13:33:06] git has a feature to find out which commit introduced a specific line of code [13:33:19] hasharDeadmau5: the ideal would be to have a search which given a directory/file gives you the most active committers there [13:33:28] yes, I know, but that's going to be too complex [13:33:32] so we could do something like: git "'messagename'\s=>" languages/messages/ [13:33:35] that would give the commitf [13:33:38] from there find the author [13:33:49] and then find out if he's still around [13:33:53] ahh [13:33:54] hmm [13:34:01] git let you do that let me craft a command for ya [13:34:10] for extension, the only hope is to contact someone who is active on them [13:34:22] for the core it's more difficult becase nobody "owns" anything [13:34:55] and the most difficult are big projects like visual editor, LQT, AFT etc. who have big teams and nobody to contact for such small things :) [13:38:56] ok so that [13:38:59] hasharDeadmau5: what we need, ideally, is a system which allows even almost clueless translators to find the dev and poke him about their doubts instead of relying on siebrand and Niklas and few others, so that there's a "social pressure" for good interface, and more in general a code review with more feedback from the community [13:39:00] what we could do [13:39:11] * Nemo_bis is eager to hear [13:39:12] is develop a small script that would find out who introduced a message for the first time [13:39:21] based on MessagesEn.php [13:39:30] that is not really that hard to do [13:39:40] extensions too [13:40:21] if it can be a web tool which also gives you the email of the committer, that wold be wonderful :) [13:44:24] well just do a git blame [13:44:32] that would give you the last author that modified the line [13:44:35] should be good enough [13:45:00] hasharDeadmau5: more relevant gerrit FR https://code.google.com/p/gerrit/issues/detail?id=866 https://code.google.com/p/gerrit/issues/detail?id=298 [13:45:17] what line? [13:48:27] Nemo_bis: git blame --show-email languages/messages/MessagesEn.php [13:48:34] that would give you the data you are looking for I think [13:48:48] you can also use git blame --porcelain languages/messages/MessagesEn.php [13:48:58] that will produce a machine friendly output [13:50:01] https://code.google.com/p/gerrit/issues/detail?id=298 : finding change that affect a file can be done by using: git log | grep Change-Id [13:50:12] that will give you the list of change id which in turn you can look for in gerrit [13:51:51] Nemo_bis: and full text search is more or less being implemented by using a Lucene backend [13:53:30] ah, that's good news [13:54:14] while I'm not that fond of the idea that I have to learn git actually :p [14:01:07] you will eventually have to :-D [14:24:20] my connection is too precarious anyway for a git clone these days [14:32:44] hasharDeadmau5: how would I git-blame a string on *all* files of my local copy? both core and extensions [14:40:04] Nemo_bis: find . -name '*php' -exec git blame {} \; > hugeblame [14:40:07] something like that [14:41:13] hm [14:42:13] hasharDeadmau5: or just grep everything and then blame on the specific file? [14:42:45] you want the opposite I guess :-) [14:45:04] hasharDeadmau5: why? :( I mean, if I'm too lazy to manually discover what extension a message belongs to [14:45:23] well just ask real question [14:45:34] "grep everything and then blame on the specific file" does not make any sense :D [14:46:33] sorry :/ [15:00:55] Nemo_bis: maybe mail to wikitech-l and expose what you are trying to build [15:01:19] Nemo_bis: apparently it is about figuring out who introduced any new message [15:01:30] or at least who modified it recently [15:02:15] so for each files ( messages/MessagesEn.php , extensions i18n files), you would git blame the file then pass a script on it to pair a message with an author/email [15:06:30] maybe Nikerabbit will know better what I want ^ [15:07:06] (better than me I mean) [15:17:07] !log reedy synchronized wmf-config/CommonSettings.php 'Setting wgDBerrorLogInUTC = true' [15:17:16] Logged the message, Master [15:19:49] !log reedy synchronized php-1.20wmf6/includes/DefaultSettings.php [15:19:57] Logged the message, Master [15:20:27] !log reedy synchronized php-1.20wmf7/includes/DefaultSettings.php [15:20:35] Logged the message, Master [15:20:55] !log reedy synchronized php-1.20wmf6/includes/GlobalFunctions.php [15:21:03] Logged the message, Master [15:21:19] !log reedy synchronized php-1.20wmf7/includes/GlobalFunctions.php [15:21:27] Logged the message, Master [15:21:27] hasharDeadmau5: ^ live everywhere [15:21:50] !!! [15:21:50] Stop using so many exclamation marks ! [15:21:52] thanks :) [15:23:26] I find wm-bot too condescending [15:24:44] Nemo_bis: get accustomed to that – after the maschines will have taken over that will be the norm ;) [15:28:15] !log reedy synchronized wmf-config/InitialiseSettings.php 'Disable aftv5 on en_labswikimedia due to out of date schema' [15:28:23] Logged the message, Master [15:28:43] DaBPunkt: hmm maybe Erewhon people were right then [15:34:08] hm, why doesn't bugzill have a keyword for system message related stuff [15:37:30] hello dungodung [15:37:39] hi Nemo_bis :) [15:39:50] hasharDeCrecy: looks like we've got a few still coming through in CEST... [15:39:59] grmblbll [15:40:07] and IDT [15:40:19] hmm, job runner could be still running a process on old code [15:40:34] might be [15:40:41] though they are short living [15:40:58] indeed, 20 minutes seems adequate [15:41:00] did I typo.. [15:41:21] I wish we had a proper logging infrastructure [15:41:29] to drop syslog / udp2log and whatever generate the fatalmonitor [15:42:27] log2redis( :warn, :style => full, "My message" ) [15:43:04] grrmbmbl [15:43:13] having lot of Error parsing memcached response [15:58:17] Reedy the out of date schema for AFT in beta labs is a problem I'd like to fix. did we miss a check-in somewhere? [16:01:09] There's no patch for adding that field [16:01:29] there are a couple of "wmf" update patches, but running those make no difference as they seem to have already been applied [16:02:04] Reedy: thanks I'll see if I can get that patch in. [16:02:36] There might be other missing fields, but I'd rather not spend ages comparing a schema on a testwiki [16:02:50] if there's no useful data, we can just nuke the tables and recreate from the main sql file [16:17:00] Reedy: about i18n tag, should it be used for all problems in English messages wording (which can also be typos, grammar etc. not affecting i18n)? [16:17:08] (on bugzilla) [16:17:19] it can be [16:17:30] Because depending on what's wrong with it, it can mean the translations might be wrong etc [16:17:58] also makes it easier for siebrand et al to find it it's not logged under MediaWiki -> Internationalization [17:21:04] !log Running copyFileBackend.php for commons (shards 1-4) (actually started yesterday) [17:21:13] Logged the message, Master [17:32:00] !log reedy synchronized wmf-config/ [17:32:08] Logged the message, Master [17:56:54] !log reedy synchronized php-1.20wmf6/ [17:57:03] Logged the message, Master [17:57:37] !log reedy synchronized php-1.20wmf7/ [17:57:45] Logged the message, Master [17:58:26] !log reedy synchronized wmf-config/ [17:58:34] Logged the message, Master [18:08:07] !log reedy synchronized php-1.20wmf6/includes/GlobalFunctions.php [18:08:16] Logged the message, Master [18:08:45] !log reedy synchronized php-1.20wmf7/includes/GlobalFunctions.php [18:08:53] Logged the message, Master [22:23:07] gn8 folks