[00:25:10] hi [00:25:31] does anybody knows what is going on here? http://meta.wikimedia.org/wiki/User_talk:Teles#Auto-confirmed_users_not_working_in_ta.wikibooks.org [02:26:25] !log LocalisationUpdate completed (1.20wmf6) at Fri Jun 29 02:26:25 UTC 2012 [02:26:34] Logged the message, Master [02:51:13] !log LocalisationUpdate completed (1.20wmf5) at Fri Jun 29 02:51:13 UTC 2012 [02:51:19] Logged the message, Master [07:37:24] hello [07:56:04] !log hashar synchronized wmf-config/CommonSettings.php 'send header from CS.php only for non CLI scripts {{gerrit|13435}}' [07:56:10] Logged the message, Master [10:17:02] Grr, someone seems to be hogging to Your.org's rsync connection :( [10:18:30] Hydriz: give them a break, it's not even 6:30am yet. ;) [10:18:35] how do you know? [10:18:46] @ERROR: max connections (16) reached -- try again later [10:18:47] rsync error: error starting client-server protocol (code 5) at main.c(1534) [Receiver=3.0.9] [10:19:02] so, that could be 16 somebodies [10:19:04] Breaking all my archiving scripts for Wikimedia datasets [10:19:24] sigh, and its the only rsync server that has the incremental dumps [10:19:55] and this issue seems to have occurred ~2 hours ago [10:23:19] you could just implement exponential backoff [12:38:13] !log dzahn synchronized php/cache/interwiki.cdb 'Updating interwiki cache' [12:38:18] Logged the message, Master [13:32:55] !log hashar synchronized php/cache/interwiki.cdb 'Updating interwiki cache' [13:33:02] Logged the message, Master [13:46:23] !log hashar synchronized php-1.20wmf6/cache/interwiki.cdb 'Updating interwiki cache for 1.20wmf6' [13:46:29] Logged the message, Master [13:46:54] !log fixed interwiki on http://wikisource.org/ main page by hacking a script in production and refreshing cache [13:47:00] Logged the message, Master [17:26:09] Reedy: (and others). Dare we deploy on July 4, or reschedule it for when the US staff isn't taking a day off? [17:28:46] actually, maybe we just do all of the deploy on Monday [17:29:03] * robla kinda likes that idea [17:35:29] well, this was a great conversation. Thanks everyone! ;-) [17:35:37] <^demon> Go team! [19:12:01] could someone remind me who were the people who did the actual svn to git conversion for mediawiki? [19:12:52] github has recently released a way to work on git repositories using an svn client, so I'm looking a bit into the conversion again for pywikipedia [19:15:36] valhallasw: pretty sure you want ^demon and hashar, but neither of them seem to be around just now http://www.mediawiki.org/wiki/Git/Conversion [19:18:06] chrismcmahon: thanks. I'm looking specifically for how the conversion (with splitting etc) was done; that page doesn't seem to have much on that [20:45:07] Can please somebody check if the clock on sodium.wikimedia.org is correct? I have a strange mail here. [21:05:41] hashar: ping; you're one of the people who were involved with the actual conversion of the svn repo to git, right? I'm looking into conversion for pywikipedia at the moment; people still don't want to switch to git, but we might be able to get something with the new github-svn stuff working [21:06:03] valhalla1w: we have a tool to do that [21:06:21] valhalla1w: the first step you want is build a list of authors in your svn repo [21:06:30] is this just svn2git, or a wmf tool? if the last, is it available somewhere? [21:06:33] and match each author with an email :-) [21:06:41] svn2git I think not sure [21:06:52] let me check [21:06:59] * valhalla1w is secretly hoping most of that work has been done in the mediawiki conversion [21:07:10] but I have to check that [21:07:16] (tip, if you want yourself to use git, you can do: git svn clone http://svn.repo/pywikipedia ) [21:07:23] then commit with git svn dcommit :-] [21:07:32] yes, I know :-) [21:07:37] or rather [21:07:40] I know it doesn't work [21:07:49] for the emails, I think we ended up doing user@users.mediawiki.org :-( [21:08:08] for most pwb committers, we can use @users.sourceforge.net ;-) *brrr* [21:08:39] but as noted, I have to look into that [21:09:18] ahh chad is awesome [21:09:35] so Chad (nickname ^demon) is the one that did the conversion [21:09:54] the software used are in a git repo https://gerrit.wikimedia.org/r/p/operations/software.git [21:10:09] or if you have a labs account ssh://gerrit.wikimedia.org:29418/operations/software.git [21:11:22] valhalla1w: gitweb url https://gerrit.wikimedia.org/r/gitweb?p=operations/software.git [21:11:27] thanks [21:11:37] I think that might be a different svn2git than the one I found [21:11:52] considering it's pretty much not ruby ;-) [21:12:03] I think Platonides did some change to svn2git too [21:12:17] he is in europe, usually connected in the evening [21:12:20] (eu time) [21:13:20] valhalla1w: definitely contact Chad for more informations http://www.mediawiki.org/wiki/User:%5Edemon [21:13:48] valhalla1w: and the project page is at http://www.mediawiki.org/wiki/Git/Conversion [21:14:13] lot of stuff on that last page though :( [21:14:13] hashar: OK, thanks [21:14:22] yeah, and not too much on the actual conversion [21:15:33] valhalla1w: not that much [21:15:50] but Chad trained me to do updating conversions in like half an hour :-] [21:15:59] the scripts are in https://gerrit.wikimedia.org/r/gitweb?p=operations/software.git;a=tree;f=svn2git/scripts;h=8ab8aa4a76949d32b8ff28aad1402c17a5ff8166;hb=HEAD [21:16:06] aka /svn2git/scripts [21:16:09] might give you hints [21:16:12] yep, I found those [21:16:18] I'm going to play around with them [21:16:39] I am pretty sure pywikipedia will be easier to migrate [21:16:53] there is probably less branching, weird tags and weird reverts [21:16:57] indeed [21:17:15] did you know I wrote some code for pywikipedia ? :-] [21:17:19] on the other hand, we have the brilliant idea of a rewrite, which is not really a branch, but it has a lot of the same code [21:17:21] (only minor bugs fixes though) [21:17:37] :-) [21:17:58] if you rewrite most of the code, maybe just start a new project [21:18:03] and some translations [21:18:15] ohh [21:18:15] http://pywikipediabot.sourceforge.net/ [21:18:18] (via translatewiki, of course :-)) [21:18:36] I think I wrote the HTML for that page hehe [21:18:42] that is broken of course [21:18:54] hehehe [21:19:06] I don't think the skins-1.5 dir exists anymore, indeed [21:19:18] that's probably skins-1.20-ish by now ;-) [21:19:42] oh, it's bits-based [21:20:02] but also: /w/skins-1.20wmf5/vector/, so yeah - that's old :-) [21:20:43] I still have access to shell.sf.net !!! ;-] [21:20:50] I must still be a member of the project great [21:21:00] I have no idea where to points links too though [21:21:48] hashar: I think the best is to forward it to pywikipedia.org and to finally do something with that [21:22:00] (which is hosted on the toolserver under a multi-maintainer-project) [21:22:39] I think there were some plans to do that at the berlin hackathon [21:22:44] but there also was beer [21:22:50] so hmm my .bash_profile is from november 23 2003 on sf.net [21:22:53] old story [21:23:02] beer is important indeed [21:23:39] -rw-r--r-- 1 hashar apache 3847 Mar 29 2009 index.php [21:24:17] that's not too bad [21:27:02] will use bits [21:27:04] http://bits.wikimedia.org/skins/monobook/main.css [21:27:57] not to mention those whisky sours [21:29:59] oh man [21:30:16] I am proud to have resisted drinking ton of them :-] [21:30:22] I'd like to send out a note to the village pumps of *.wikipedia.org about the change in the date of the 1.20wmf6 deploy...rumor has it there's a tool for that(?) [21:30:37] robla: I am not aware of any sorry :( [21:30:50] valhalla1w: the sf page is updated to use some fresh skin http://pywikipediabot.sourceforge.net/ [21:31:04] cool! [21:31:05] thanks [21:31:23] valhalla1w: some history : http://pywikipediabot.sourceforge.net/changes.html ; ) [21:31:51] a_engels 2003/11/05 [21:31:52] French translation by Hashar [21:32:04] that is basically how i started coding for the wikipedia project [21:32:08] i.e. by doing translations [21:32:16] dah I love computers, they have huge memories [21:32:38] hehe [21:32:44] valhalla1w: do you know if hooft is still active in the project? [21:32:48] you just have to know where to find it [21:32:51] not that I know of [21:33:17] it's mainly xqt [21:33:29] one day I will have to write a self biography [21:33:45] would be a lot of fun trying to find out what I did other the last 30 years or so [21:34:18] valhalla1w: if you need any update on that page feel free to ping me either here (during european business hours) [21:34:25] valhalla1w: or mail : hashar at free dot fr [21:34:58] hashar: thanks. If I set up a new password, I hope I should be able to access it... [21:35:19] I don't even remember my password to log in sf.net :/ [21:35:26] it's the same as for the shell account [21:35:30] or did you use pka? [21:35:31] but most probably should be able to add a shell account for ya [21:35:45] I used an old ssh key that luckily I still have the private key for [21:35:47] \O/ [21:35:48] I'm one of the project admins in any case ;-) [21:36:00] ohh so you should be able to grant self shell access [21:36:08] I have logged into the shell once [21:36:11] like... 5 years ago [21:36:24] that was really smart from sf [21:36:32] bloody hell, gerrit. [21:36:43] I so loved source forge, I am just wondering why it failed at keeping popular [21:36:55] does anyone know what i have to do to convince gerrit i want to delete a branch? [21:36:55] dschoon: what is wrong ? [21:37:03] outages, bureaucracy [21:37:12] dschoon: you need to be able to force push I think [21:37:18] is that a priv? [21:37:20] I think github is so popular mainly because you don't need to get anything approved [21:37:33] and do something like git push -f gerrit :refs/youwanttodelete [21:37:45] right, that's what i'd do were i working with a sane git server. [21:37:47] (aka pushing a null pointer over the refs of your branch) [21:38:01] (though usually i git branch -dr and then git push --prune --force) [21:38:04] force push is a privileged indeed [21:38:10] gerrit tells me to fuck off. [21:38:11] sigh. [21:38:14] it is disabled cause that let people push a totally different history [21:38:19] yeah. [21:38:38] that's my goal, as apparently "delete repo" is not a supported gerrit command. [21:38:51] ha, I can also login with an ancient private key [21:39:04] let's see if the shell server is as slow as I remember [21:39:13] dschoon: I guess you have an issue with one of the analytics project ? [21:39:19] yeah. [21:39:24] housekeeping at the moment. [21:39:34] you could just enable Analytics to force push by checking the 'force push' checkbox at https://gerrit.wikimedia.org/r/#/admin/projects/analytics,access [21:39:48] that is definitely confusing but made as a safeguard [21:40:20] as for deleting repo, I am not sure what the issue exactly. I think it is just there is nothing to clean the gerrit database of all references to the deleted repo [21:40:34] maybe chad will end up coding support for that [21:41:19] that's what i was going to suggest. [21:41:24] let's see if i have the power to do that. [21:42:23] my understanding from Chad is that repo deletion is oft discussed, and probably will be a core feature in the coming year, but isn't yet. Lots of db cruft to clean [21:42:24] it's magic. gerrit saved my changes. [21:42:30] dschoon: ahh i found the doc for gerrit 2.3 http://gerrit-documentation.googlecode.com/svn/Documentation/2.3/access-control.html#category_push [21:42:47] dschoon: Direct Push, Force option : Allows an existing branch to be deleted. [21:42:56] ja. [21:43:06] thank you, hashar! you have solved my problem. [21:43:13] you are welcome :-] [21:43:34] i think i can even go back to my original solution of nuking all history [21:43:36] it is definitely puzzling but you eventually get used to know all the possibilities of gerrit [21:43:54] did you know you can write Gerrit rules using the prolog language ? :-] [21:44:16] personally, i'm looking forward to becoming intimately familiar with its flammability [21:44:22] ahah [21:44:51] or perhaps the physics of its graceful path off a bridge. [21:44:58] dschoon needs to finish laying the tinder for that [21:45:07] tis true. [21:45:16] I can imagine Gerrit becoming sentient. [21:45:18] trust me, sekrit plans are at work! [21:45:48] can't we make gerrit write mediawiki for us? ;-) [21:45:55] need this to get done so that we can kick off the wikitech-l discussion: http://www.mediawiki.org/wiki/User:Dsc/GerritShouldDieInAFire [21:46:45] valhalla1w: that is basically what is being done :-] [21:47:06] valhalla1w: nowadays I merely click the [Submit change] button and barely write any new features in mw [21:47:29] valhalla1w: I think my code in core has been mostly bug fixes [21:47:40] :-) [21:48:11] dschoon: as long as we avoid writing our own tool and the selected tool fit our workflowsI will be fine with whatever solution you find out :-] [21:48:46] something I would love is a way to do pair programming over internet [21:48:55] and a one click button to run tests remotely [21:49:07] I haven't had the free time to dig into these things, but I really love what I've seen so far of http://getbarkeep.org/ [21:49:21] hashar: tests that run while you type / when you save would be even better [21:49:51] valhalla1w: you can do it on save really easily if you use vim :-] [21:50:11] valhalla1w: you can even run php -l as you type to highlight lines having an error [21:50:20] hehe [21:50:36] dschoon: if they are in SF, maybe they can come make a demo in the office? [21:50:57] hashar that'd be great if so [21:51:49] dschoon: not only getbarkeep, phabricator too [21:52:01] yeah, there are quite a few to look over [21:54:21] hashar: hm. is it possible to get a copy of the entire svn repos without shell access? [21:54:33] i.e. without going through the svn server [21:55:12] valhalla1w: I have no idea [21:55:42] dschoon: another option is considering making Gerrit better :-D [21:55:51] let me guess: I should ask chad? ;=) [21:56:04] valhalla1w: ohhh the WMF hosted one ? [21:56:07] yeah [21:56:12] valhalla1w: we can definitely do that, ask Chad indeed [21:56:30] or to use github. there are lots of options. [21:56:32] I am not confident in getting data outside of the cluster without knowing exactly if it is safe to have it out [21:56:34] some of them might even be good. [21:56:56] hashar: of course. no problem. [21:57:08] dschoon: well the mobile team is using github with some success apparently. I think their idea was to get a lot of volunteers to be interested in [21:57:33] as I said during berlin, since each teams as different needs, maybe we will end up using different tools [21:57:44] but that might end up confusing [21:58:14] in terms of a head-to-head comparison on confusion, i think gerrit can take all comers. [21:58:15] valhalla1w: but that is definitely possible. if there is any private stuff, we can filter it out [21:58:39] perhaps even with one core tied behind its back [21:58:47] dschoon: maybe. If you need any inputs, let me know. I will be happy to help as time allow [21:59:18] totally :) [21:59:49] i'm being glib now, but i think we all agree this is an important task, given how central source control and code review are to an engineering org [22:00:01] so there will be plenty of discussion/flaming, and lots of research to do [22:02:28] also consider CodePlex from microsoft :-))) http://en.wikipedia.org/wiki/CodePlex [22:03:06] launchpad is a nice tool too [22:03:19] well enough mumbling. I am out for good now [22:03:28] I don't even remember why I connected in the first place [22:03:37] end up spending like 3 hours fixing stuff ;D [22:03:50] valhalla1w: see you later :-) I am on jabber too if needed : hashar at jabber.org [22:03:52] cya! [22:13:30] so apparently there is a command called 'svnsync'... which does exactly what I want [22:13:34] (just really, really slowly) [22:14:59] Hi guys! A question for the CSS-guru's: is there a css class on .en.wiki that is automagically visible to those with admin bit but not to others? [22:20:06] arcandam: You're talking about limiting CSS to admisn? [22:20:08] admins [22:20:16] yes [22:20:20] There's a Sysop.css page, I think. [22:20:42] https://en.wikipedia.org/wiki/MediaWiki:Group-sysop.css [22:20:48] Not sure why that can't be a redirect... [22:21:08] js/css pages behave differently [22:21:22] thanks a lot [22:21:22] The fact that it eats redirects is probably a bug. [22:21:25] No problem. [22:21:45] I just checked that link and yes, I love you! :-) [22:21:55] many do [22:21:55] * arcandam hands Brooke a couple of stroopwafels [22:22:24] yeah I have a bit of competition [22:25:57] Yum. :-) [22:39:04] * Logan_ eats Brooke. [22:39:50] oh boy [22:40:15] the love of my life has been eaten by a ferocious wild Logan! [22:40:24] what can I do? [22:40:31] Join in. [22:40:37] * arcandam flees and marries someone else [22:40:42] bye bye [22:40:55] What have I done? [22:41:04] * Logan_ is a monster. [22:50:47] Logan_: You're the worst. [22:51:01] :v [22:57:00] ouch Brooke speaking from his belly [22:57:43] Brooke is a whore [22:57:48] Prove it. [22:58:58] gn8 folks [23:03:35] Brooke is all talk and no action [23:03:46] check whois for proof of that :-P [23:04:38] Talk is cheap. [23:05:05] hey, I didn't say it :-P