[00:03:16] adam_miller, it's supposed to [00:03:34] just not the simplest to test (and as said previously, rewrites etc) [00:04:01] no? [00:04:14] just clear your tables and make a request [00:04:23] then look at your tables again [00:04:27] yeah, just faffing with urls [00:04:31] rather than having a gui to poke ;) [00:04:49] brb, toilet. I'll check then [00:04:53] faffing? [00:05:55] fiddling [00:05:56] playing [00:05:59] this sound so wrong [00:06:34] haha [00:07:23] ok so it's 8pm here so i'm gonna say i've had enough faffing around with this thing for today [00:07:38] i'll send an email with my status and what i still need help from you on yet [00:07:47] THEN imma go eat a hamburger [00:07:50] with an egg on it [00:07:59] and an avocado [00:08:05] and maybe bacon [00:08:20] yeah. [00:08:30] i'm gonna seriously faff that thing up [00:08:47] lol [00:16:38] Ashlee: We had that in the topic before the channel rename, with a link to the logs even. Lemme restore that [00:20:24] There [17:15:50] http://eiximenis.wikimedia.org/SentenceLevelEditing31August2010 [17:18:03] adam_miller: ping! [17:18:08] x2002? [17:18:16] no, actually, we have to move it to later today. [17:18:29] oh? [17:18:30] we'll figure out a thing during the features meeting call in. [17:18:37] what's this ping about [17:19:10] just to let you know that we aren't having the scrum. [17:19:32] which was scheduled for now. [17:19:43] adam_miller: we will have the 11am status meeting instead today [17:20:10] k [18:01:48] http://eiximenis.wikimedia.org/FeaturesTeam20100831 [18:04:07] adam_miller: You can call into x2002 now [18:06:00] how do I get an account on rt.wikimedia.org ? [18:06:41] flipzagging: Ping mark [18:09:10] she said semantic! [18:11:21] Reedy: do we get to point and giggle or something? [18:11:35] !hss [18:11:35]

http://bit.ly/2sYkfl [18:11:40] danese: #mediawiki humor [18:11:45] danese, ^ [18:11:50] x 2 [18:53:00] RoanKattouw: can you explain this bug to me? 24906 [18:53:50] OK so you know those tabs at the top that say Article / Talk (or User page / Talk or whatever, depending on namespace)? [18:53:58] yes [18:54:08] They currently use Vector-specific messages that are otherwise equal to the core messages the rest of the skins use [18:54:16] ok [18:54:41] Except for one namespace I think, where they differ (or core doesn't have one so it uses the ns name while Vector has an explicit message; I forget) [18:54:47] So what we wanna do is consolidate this [18:54:54] And use the core messages instead of the Vector messages [18:55:09] This way people won't have to duplicate all their local customizations [18:57:57] so is this changing something in an i18n file? [18:58:30] Mostly just updating callers [18:58:37] You can eliminate the obsolete messages in the i18n if you want [18:58:45] in Vector.php? [18:58:51] But the work we really care about is making those msgs obsolete in the first place [18:58:53] I think so yes [18:59:27] (brb lunch) [19:20:44] AryehGregor: Sorry, I got distracted earlier. I really don't know the reason that the mobile site is in Ruby. [19:20:57] It doesn't make any sense to me from a sysadmin POV or a developer POV. [19:21:08] Ashlee, why are you responding here? [19:21:19] Anyway, I'm guessing it's in Ruby because hcatlin likes Ruby. [19:21:22] I thought it'd be less intrusive than in -tech. [19:21:38] Just like it's probably on github instead of our SVN because hcatlin prefers git to SVN. [19:21:46] That doesn't make sense. [19:22:01] Well, it makes sense, but it isn't logical, y'know? [19:22:07] We don't clamp down enough on this kind of thing. Everything dev-related should be PHP and in our SVN. [19:22:14] Right. [19:22:21] Unless it's C++ or something, okay. [19:22:25] External dependencies are idiotic. [19:22:31] Or PHP otherwise just doesn't work for the purpose. [19:22:38] As is building a new project in a language that nobody except one person is programming in. [19:22:50] <^demon> I don't mind that its in Ruby even. I just wish it was in svn. [19:23:01] And then there's the matter of different server configuration. [19:23:17] I just don't understand how this happened (so recently, as opposed to in 2003 or something). [19:23:43] Math support being obscure and written in OCaml makes sense. It was volunteer work and it was forever ago. The mobile site is recent and was commissioned. [19:23:46] So I'm just a bit lost. [19:24:04] Yes, it would be more understandable if it were a volunteer thing. [19:24:36] Well, in a typical situation like this, a volunteer project written in a completely different language requiring different server configurations and such would just be outright rejected for WMF. [19:24:46] There must be a backstory here. [19:25:47] I imagine it's "nobody cared, so hcatlin did whatever he felt like". [19:25:53] Just guessing. [19:26:09] Perhaps. [19:54:11] Ashlee, AryehGregor, from what I recall, that's pretty much what happened. They want to rewrite it in PHP this year, iirc. [19:54:30] tomaszf will know. [19:54:41] I'm lost. [19:55:13] guillom: context ? [19:55:32] tomaszf: The reason the mobile site is written in Ruby. [19:56:30] ahhh there we go. i didn't have 90% of the chat :) [19:57:26] it was written in ruby as the dev who built our iphone/android app was most familiar with it and could create a working version very quickly [19:57:40] hampton made it very modular and could easily add subtract things as needed [19:57:44] so it worked quite well [19:57:44] bu [19:57:46] t [19:57:55] Was there some sort of time constraint? [19:58:06] bleeding edge ruby dev is a real pain for ops [19:58:23] as you constantly need the latest packages to keep up with that dev world [19:58:25] It's also a real pain for development, surely. [19:58:32] and we don't want to get into that game [19:58:52] Ashlee: not really. ive talked to plenty of ruby devs that love to be that bleeding edge [19:59:08] so its not an issue [19:59:14] tomaszf: I mean 99% of WMF projects aren't in Ruby. [19:59:33] and 99% of them arent in php either [19:59:37] If something breaks, who besides Hampton can or will fix it? [20:00:00] basic ops issues are well documened on wiki tech with our standard ops template [20:00:16] anything else is escalated to hampton and the volunteer devs that work on it [20:00:42] Is it going to be rewritten? [20:01:07] we run a fairly varied stack overall .. from php, java, python, perl.. etc .. so there is not intrinsically wrong with another language .. but ruby doesn't seem to be the best fit it [20:01:19] yup, were looking to port it [20:01:31] Almost everything dev-related is written in PHP. The number of exceptions is very small. [20:01:36] I don't think already having a mess is a very good reason to add to it. [20:01:42] Ops stuff seems to use more Python, but that's mostly separate. [20:01:47] i'd prefer php so that as new proj[21:12:09] Not sure, ask siebrand [21:20:22] alirght Reedy...how do i get my current users's rating? [21:20:41] i'm sending userrating=1 but doesn't seem like i get anything different back [21:20:44] why is copy paste being shit [21:21:15] Hmm, that's an annoying point [21:21:22] hmmm [21:21:23] ? [21:21:41] i've tied it into revid, but that's only valid if the page is still that rev [21:21:44] just need the newest [21:21:45] fail [21:22:11] so...it's just not working yet? [21:22:28] you can test the output if you put in some revid you know you have rated against [21:22:59] pretty sure i haven't changed this article i'm testing against ever [21:23:43] no, i mean a rev you rated [21:24:22] that's what i'm saying, i rated rev 1505, and am sending 1505 in my request for the results [21:24:37] http://192.168.0.190/w/api.php?action=query&list=articleassessment&aapageid=1&aarevid=37&aauserrating [21:24:47] oh [21:25:01] i was sending userrating, not aauserrating [21:25:51] now it's working [21:26:05] are you going to send back the rev id when the user last rated? [21:26:11] or a number of revisions since the user rated? [21:26:27] i need that for the messaging we put in for stale ratings [21:29:41] Reedy: he has to remove the English message (MessagesEn.php) and remove the keys from messages.inc. If the messages are optional or ignored, the keys also have to be removed from messageTypes.inc. [21:31:05] I'd setup the stale to put a flag on the output... Do you just want he revid? (Which I've included anyway, but that's slightly kooky still [21:32:01] are rev id's sequential for each individual article? can i do some arithmetic to get the number of revisions since they last rated? [21:32:31] nope [21:32:55] can filter by page id, and count id's since rev x in sql [21:39:04] Heh [21:39:07] order by bug ;D [21:45:50] adam_miller, http://mediawiki.pastebin.com/abAQ7JSV [21:45:53] How's that suit you? [21:46:05] stale will appear with pageid/revid if applicable [21:46:58] stale may need a bit more tweaking.. [21:47:09] i think that's what i'm getting now, isnt it? [21:47:45] that's with the revid from http://192.168.0.190/w/api.php?action=query&list=articleassessment&aapageid=1&aauserrating&format=jsonfm [21:48:04] uncommitted [21:51:14] mobile doesn't seem to be using nothing that a custom skin couldn't do [21:51:35] *Reedy ponders [21:53:14] getting the rating for the article, left joining in the user will only give the user rating IF they've rated it, though.. We're not joining on revision, just page and rating [21:53:15] Sooo [21:53:26] At worse, we might be not picking up the newest rating [21:53:27] *Reedy tests [22:04:40] adam_miller, nope, looks about right now [22:21:11] adam_miller, have you done anything (generally) with regards to generating some uid for users (anons in this case), we can save in a cookie and store for the rating stuff? [22:21:22] not yet [22:21:25] i can [22:28:55] just gotta decide what size field, and let submit take it [22:29:03] and i suppose, list.. [22:30:07] Need to write postcards later and find some stamps [22:57:27] TrevorParscal, RoanKattouw the manifest will also be cool for MediaWiki:Common.js updates .. so people don't report admin javascript errors days latter since the manifest can update rev_id / time for those resources [22:58:40] what's the manifest? [23:00:46] mdale: Quick revalidation of Common.js / skin JS was always an objective [23:01:02] Currently Common.js + skinname.js is in its own custom module called 'sitejs' [23:01:21] [23:50:07] To force an indexed array with no values, do you have to do something like array( 1 => 0 )? [23:51:59] Saying that, for this, doing the title wouldn't be a bad thing for "maintainability" [23:53:06] An indexed array with no values? You mean {} as opposed to [] ? [23:53:39] *Reedy shrugs [23:54:02] You can totally do that though [23:54:09] array() becomes [] in JSON [23:54:22] array( 'foo', 'bar', 'baz' ) becomes ['foo', 'bar', 'baz'] in JSON [23:54:25] As above though, might aswell use the key as the title, so you can see what they are, rather than just artbitary numbers? [23:54:55] Right so array (123 => 'foo', 456 => 'bar', 789 => 'baz') becomes {'123':'foo', '456':'bar', '789':baz'} [23:55:02] And to get {}, you'd json_encode( new stdClass ) [23:55:10] It's staying as php ;) [23:55:14] (or FormatJson::encode() in a MW context) [23:55:23] Oh it's just staying in PHP? [23:55:28] Ya [23:55:29] Then just create an empty array [23:55:40] PHP doesn't distinguish between indexed arrays and 'hashmaps' [23:56:25] Both are the same structure, and the JSON encoder uses array_keys( $arr ) == range( 0, count( $arr ) - 1 ) to determine if something is a real array or just a map [23:59:54] if you use array( 1 ) vs array( 1=> 'Foo' ), and use array_key_exists, using a key of 1, only returns true for the second