[05:49:58] Where near the WMF office is still open where one can get reasonable food? [10:11:16] Krinkle: Around? [10:12:58] Yes [10:13:44] Coming two weeks I'll be spending most of the day without school, not much activity around now. [10:13:49] Whats up ? [10:14:38] Oh cool, same here [10:14:45] Well I wanted to get started on the RL 2.0 stuff [10:14:55] Yeah. [10:15:28] At this point that basically consists of speccing out some of the new 2.0 things we talked about, as well as poking at some of the bugs currently filed [10:15:55] Including things like 'what happens if we put scripts back in the head' [10:16:18] Have you read / thought about the split' approach ? [10:17:09] I've briefly read about it [10:17:26] Is that the one where we identify things that need to be in the head as such, and put them there, while keeping other stuff at the bottom? [10:17:46] RoanKattouw: Yes, lemme pull it up. [10:17:53] So things like collapsibletabs would be in the head while things that e.g. add a click handler to something (twinkle?) go at the bottom [10:18:24] yeah, the 'Steve Sauders' ideal. [10:18:35] https://bugzilla.wikimedia.org/show_bug.cgi?id=27488#c15 [10:19:40] key being "how many of this code is actually being executing before or at the dom-ready event" [10:20:05] Right [10:20:08] Actually, I think this should be easy [10:20:16] What if we move the startup script to the [10:20:23] that for sure [10:20:36] That'll do a doc.write for mw+jquery, which will then be loaded immediately (also before the rest of the paeg) [10:20:59] We then have the loader load top-queue modules immediately in the same fashion [10:21:05] (but /with/ timestamp goodness) [10:21:22] http://eiximenis.wikimedia.org/RL2 [10:21:44] Ah yes [10:21:55] yep, two queues [10:21:56] We should definitely clear the trivial bugs section, for one thing [10:22:08] I'm just responding to your cacheability concern [10:22:21] This could totally work, without cacheability concerns [10:22:21] Let's not do [done] things at etherpad RL2, what's done can be scrapped. It's too cluttered already :D [10:22:28] awesome [10:22:32] Yeah just delete what's done [10:22:48] The only downside is that jq+mw is a significant amount of code to load+parse+execute in the head [10:24:30] RoanKattouw: True, but it was that way < 1.17 as well, [10:24:39] I know [10:24:52] But there's always room for improvemenet [10:25:00] Plus, pre-1.17 we weren't loading jQ on every page [10:25:50] Hm... sure ? [10:25:57] in wmf atleast it was. [10:26:52] in 1.16wmf4 yes [10:26:57] Because I hacked it into CommonSettings.php [10:27:23] yeah [10:27:29] Using a BeforePageDisplay hook that called $wgOut->includeJQuery() [10:28:09] Where there complains about jquery breakage in 1.16wmf4 (where it didn't do blacklisting). And bytheway, now I read back IE6 is supported both in MW and jQ... so it's < IE5.5 we're blacklisting. [10:28:09] And that was a looong time after the 1.16wmf4 deployment. It was a few weeks after I'd introduced $wgJQueryOnEveryPage I think [10:28:21] Right [10:28:29] Ah, no, no complaints that I know of [10:29:14] So perhaps we can just drop that as a wontfix. I mean, Firefox 1.5 and IE 5.5... perhaps we can get some statistics on that. [10:29:15] Krinkle what will be the next Toolserver tool for me to write about ?? [10:29:35] GerardM-: Recent Anonymous Activity is going to be localized [10:29:40] GerardM-: Krinkle's time is mine now, mwahahahah [10:29:47] http://toolserver.org/~krinkle/recentAnonymousActivity.php [10:29:58] Krinkle: I'll grab stats, hang on [10:30:06] hehe [10:30:56] IE5.5 is 0.06% :oO [10:31:09] public ? [10:31:26] Firefix is 0.03% [10:31:28] Yes [10:31:32] http://stats.wikimedia.org/wikimedia/squids/SquidReportClients.htm [10:31:42] k [10:31:46] So you're right, these clients are waaay below the threshold where we stop caring [10:32:04] As long as things don't horrendously break, it's fine [10:33:37] RoanKattouw I prepare the blog post for when Krinkle is ready [10:34:03] Oooh, just looked at the 2.0 list and remembered ESI [10:37:07] I almost forgot about that one, need to read up though. I'm not sure why it was so cool (since timestamps aren't in the raw dom but in the startup module) [10:37:40] ESI allows embedding of the startup module through Squid/Varnish [10:38:12] I imagine we're getting a lot of requests for the startup module, both from people whose 5-minute cache has run out and from new visitors [10:38:28] ah, so no need to make something a seperate request "just" because it must not be cached in the static html. [10:38:40] Varnish caches the startup module and only requests it from the backend every 5 minutes, but it'd be much better if it could just embed it in HTML pages [10:38:40] yeah, that's nice since it's just a small module [10:38:58] sure thing [10:39:13] So right now startup is in 5 min cache ? [10:39:16] I imagine this'll save a significant number of requests, although Varnish doesn't break a sweat with the current load [10:39:18] Yeah [10:39:24] It's max-age=300, s-maxage=300 [10:39:41] Oh and [10:39:50] Our 'move shit to head' plan will have another positive effect [10:40:05] Right now things like the history module are loaded in the head, for the reasons we discussed [10:40:17] (Or their CSS is at least) [10:40:22] yep [10:40:32] But that means they use unversioned URLs, which also go into the 5-min cache [10:40:36] RoanKattouw: Ah, it's in max-age header ofcourse thats why I didn't see it. [10:40:47] on that though, I see my css modules in thead have a max-age of 0, any reason for that ? [10:40:52] head* [10:41:15] So people are re-requesting the history CSS all the time, and Varnish pulls it from the backend every 5 mins, but none of that is needed after the restructuring we came up with [10:41:22] both the user module and the core/extensions module request for css only are max-age 0 [10:41:24] Huh, which URLs? [10:42:03] http://i.imgur.com/1MwXR.png [10:42:13] http://bits.wikimedia.org/commons.wikimedia.org/load.php?debug=false&lang=en&modules=mediawiki.legacy.commonPrint%7Cmediawiki.legacy.shared%7Cskins.vector&only=styles&skin=vector [10:42:28] That max-age=0 is in the /request/ header, not the /response/ header [10:42:36] It's Chrome's way of saying 'gimme a fresh version plz' [10:42:50] (And it gets a 304, lol) [10:43:27] weird [10:43:39] Cache-Control:public, max-age=300, s-maxage=300 [10:43:39] indeed [10:44:00] Same for Vector CSS too [10:44:15] So with ESI, we can give these a longer max-age and put a timestamp in the url ? [10:44:22] Gotta be careful there, though, CSS that needs to work even without JS should stay up there [10:44:34] Not even with ESI [10:44:44] Just by moving the startup module to the head and using doc.write appropriately [10:44:55] yes. [10:46:01] I'm a little confused two as for these short max-ages, I didn't know that . expeected the to be much longer. Are they all just 300 ? [10:46:24] ah, the dynamic ones are 2592000. [10:46:47] Yeah the short ones are all 5 mins and the long ones are all 30 days [10:46:53] This is configurable in LocalSettings though [10:47:18] Hmm I have another idea [10:47:22] I wonder whether this works: [10:47:54] ... ... [10:48:27] If that works we can version the skin CSS when JS is enabled and fall back on unversioned refresh-every-5-minutes behavior if JS is not available [10:48:54] noscript in head.. never tried. [10:49:19] seems to be popular google search (suggested via) [10:49:32] http://www.google.nl/search?q=noscript+tag+in+head [10:52:12] http://webreflection.blogspot.com/2007/09/noscript-tag-behaviour-and-head.html [10:52:13] perfect [10:52:55] "Can anyone explain me why noscript can't be after a script inside head and why noscript can't contain in every part of page an external resource?" [10:52:56] too bad. [10:54:25] It's not valid HTML [10:54:30] Read the comments [10:54:39] yes, reading them now [10:54:41] Someone even says that the spec is silent on it, and that this is a bug in the validator [10:54:57] we need to make sure it's supported though. [10:55:09] External CSS tags apparently don't work inside

@import url(...) Yeah we need to test this cross-browser before using it [10:55:48] I can build a quick test page at ts and run through browsershots.org. [10:56:41] Please do that [10:56:52] I have to go and get my picture taken soonish [10:57:01] k [10:57:31] If you have any time left, I guess you can start picking off some of the easy bugs on the Etherpad [10:57:41] yeah [10:57:44] Then when I get back we'll start working on moving stuff around [10:57:52] eta ? [10:58:02] Probably half an hour, but no guarantees [10:58:25] These pictures are for a Canadian Citizenship Certificate application and their photo requirements are insane [10:59:04] wow [10:59:21] I thought you were moving to SF [10:59:28] "It must be you in the photo" [10:59:48] I am, still :) [10:59:56] k [11:00:02] Well, nothings garunteed till you get a work visa ;) [11:00:12] I'm not applying for Canadian citizenship as such, I technically have it since birth [11:00:19] I just need a document to prove that [11:00:32] Reedy Well, nothings garunteed till you get a work visa [11:00:36] And that is very true [11:01:39] Do you get to start using "eh" in every sentence? [11:01:51] http://www.cic.gc.ca/english/pdf/kits/citizen/cit0021e.pdf [11:02:16] Reedy: Only if I repeat it in French afterwards ;0 [11:02:20] * ;) [11:02:38] Alright, going now [11:02:43] glhf [11:02:43] http://krinkle-tools.grizzdesign.nl/noscript-head/ [11:02:51] RoanKattouw_away: works in Safari 5 atleast [11:03:22] Reedy: Can you test that in your current browser with JS on, and JS off ? [11:03:34] Excellent, WFM in FF4 [11:03:46] Krinkle http://ultimategerardm.blogspot.com/2011/04/what-toolserver-tool-will-be-next.html [11:05:01] What's it supposed to look like if it works? [11:05:14] Blue with JS on and green with JS off [11:05:16] As it says [11:05:19] :P [11:05:25] If it's black, you have a problem [11:05:28] (or we do) [11:05:40] It doesn't say anything [11:05:44] that makes sense [11:05:54] ? There should be some text on the page [11:05:57] There is [11:05:59] Blue JS | Green Noscript-CSS | Black neither [11:06:00] But it doesn't make english [11:06:06] haha [11:06:07] Where's JS turn off for Chrome [11:06:12] OK, really leaving now [11:06:36] Reedy: Good question [11:06:44] Google has many OLD answers [11:06:47] Testing in 74 browsers with js off: http://browsershots.org/http://krinkle-tools.grizzdesign.nl/noscript-head/ [11:07:03] (may take a while, busy factories) [11:07:37] Found it [11:07:58] Goes green with JS off [11:08:08] Chrome 12.0.752.0 dev [11:08:36] Want me to test it in IE8? [11:08:42] sure [11:08:55] Got FF 3.6 too [11:09:27] Works on FF 3.6 [11:09:55] Please put on ether pad [11:10:09] Chrome doesn't display it the same as IE of FF [11:10:10] Link? [11:11:05] Krinkle_, etherpad link? [11:11:10] Works in IE [11:11:16] http://eiximenis.wikimedia.org/RL2 [11:11:21] 3rd or 4th section [11:11:31] ===

[14:37:08] and Opera freaked out? [14:37:13] But CSS inside noscript (*any* CSS, including inline doesn't work inside a noscript in Opera 11.0X [14:42:48] I was saying, we can wait until people upgrade to 11.10 and 11.0X becomes disused, or we can just use ESI to solve the same problem in a nicer way [14:44:28] RoanKattouw, looking at enwiki, 11.01 came out end of jan, so most of them must update regularily [14:44:38] 11.10 build 2077 came out yesterday [14:44:39] Right [14:44:55] I'd guess that means it won't be long [14:44:58] Yeah [14:45:03] Still, it'll probably be a few months [14:45:09] And using ESI is even better [14:45:22] The purpose of the noscript fix is to request the skin CSS only when it changes instead of every 5 mins [14:45:40] The purpose of ESI is to have Varnish embed the skin CSS into the page, no separate request needed [14:45:57] I'm not sure whether ESI-ing a larger resource like vector CSS is a good idea thogh [14:46:31] I'd best go find food [14:46:36] apergos, have you eaten today? ;) [14:46:37] :) [14:46:40] Lunch, I'm guessing? [14:46:52] It's 4pm so you can't possibly be talking about dinner [14:47:08] damn right [14:47:23] Out of milk, best go get some [14:48:41] TrevorParscal: http://eiximenis.wikimedia.org/RL2 see section "Scripts position in the DOM" [14:48:48] That's what Krinkle and I have been poking at today [14:48:53] I'll be coding this up soon [14:49:21] I need to fix the CR path fragmentation bug today or tomorrow [14:49:41] And redo my damned asp .net mvc tests [15:01:11] RAGE [15:01:19] That's like the 4th lockup today [15:01:50] dude [15:01:57] new OS install time? [15:02:15] I installed a newer kernel version yesterday [15:02:15] alrighty, all ts-int pending stuff committed, back to RL2 [15:02:21] Now downloading the latest stable kernel, gonna try that out [15:06:46] Reedy: yes, I'm ahead of the game, I ate a small lunch [15:15:59] heh [15:24:34] Alright, compiling my own kernel now [15:24:43] Let's see if my machine is more stable on 2.6.38.2 [15:27:19] goodluck [15:49:44] weee, 25??C and *very* sunny today [15:50:01] I guess Summer is here early. [15:53:18] we had some sunny days a week or so back [15:53:20] then they went away [15:53:22] back again today [15:58:06] it's gorgeous up on the hill today here. [17:09:54] Hi sumanah, is the GSoC website issue resolved? [17:10:02] hi akshayagarwal [17:10:14] are you talking about the problem jorm had in saying he wanted to mentor you? [17:10:20] yes [17:10:26] I believe it is resolved, yes [17:10:40] hey folks. Who is in the WMF SF offices today? I'm looking for kaldari [17:10:49] do i need to do anything on that, sumanah? [17:11:09] sumanah, you might want to review http://www.google-melange.com/gsoc/proposal/review/google/gsoc2011/tmcpro/1 :P [17:11:24] jorm: as far as I know, you don't need to do anything [17:11:54] Reedy: ok [17:12:09] thanks sumanah [17:39:48] alolita been around? [19:54:13] folks, did anythin just get deployed that you know of? [19:55:56] http://bit.ly/wikisal [20:06:20] was there to be an upload wizard deployment? [20:06:41] Yes, later tonight [20:06:43] hmm [20:06:44] Or, for me, tmorrow [20:06:49] apergos: Bad time? [20:06:52] ok, please keep ryan in the loop on it [20:07:01] if things go wacky you'll need to pull back. [20:07:01] so, what would be a good icon image for "Project Invite?" [20:07:11] invite what? [20:07:14] red tape [20:07:15] :) [20:07:18] Well TBH I'm not in the mood to get up early and her "no, go away, bad time" [20:07:23] *hear [20:07:34] it would be more like [20:07:40] "woops that's not working out" [20:07:54] What I'm saying is [20:08:23] well, we just had load issues, but it was likely because of doing regular requests and the rsyncs at peak load [20:08:28] Should you guys decide today (our time) that it's a bad idea to deploy tomorrow (morning, around 10am EET), let me know [20:08:35] ah. no, I didn't mean that [20:08:36] I meant: [20:08:43] Hmm I should probably just quote UTC times :P [20:08:54] if you deploy, (which I guess is just announce "hey here's this thing") [20:09:03] No, we have a bugfix deployment too [20:09:06] and then we get a bunch of people using it and it turns out there are performance issues [20:09:15] I want to make sure someone is watchign it that can tell you [20:09:21] I guess we can push the bugfixes, publicize, then if it breaks, unpublicize [20:09:28] so you can turn it off or whatever, limit it. [20:09:36] Well Ryan won't be around, I'm guessing you will be [20:09:52] oh in the morning. [20:09:56] Deployment is scheduled for 10am Greek time [20:10:00] =midnight PDT [20:10:00] mmm [20:10:16] ok I guess I'll make it a point to be here [20:10:22] are you guys expecting a major increase in load? [20:10:36] (I'm not) [20:10:37] Not expecting, really, no [20:10:40] The only thing is [20:10:55] If we announce and encourage UploadWizard, we'll probably see more stashed files [20:11:06] maybe people hitting it like "oh hey, let me try this out" [20:11:09] Yeah [20:11:18] I think we are stable enough for that [20:11:21] So for a stashed file there's a creation then a move [20:11:24] right [20:11:28] As opposed to creating it in the right place [20:11:33] right [20:11:36] so more writes [20:11:37] so we'll see if that is enough of a burden to matter, or not [20:11:40] Plus, stashed files are thumbnailed, and those thumbs never get cleaned up or moved AFAIK [20:11:42] got one. [20:11:43] in theory it shouldln't be so much [20:11:49] but, who the **&^ knows right? [20:11:56] RoanKattouw_away: at some point we should be cleaning those [20:12:01] Yes [20:12:05] Cause there will also be abandoned files [20:12:10] yeah [20:12:12] yeah that thing where they're never cleaned up is less than wonderful [20:12:14] Right now this is a limited problem, but it'll get more serious [20:12:24] yes. very much so [20:12:24] *jorm less than wonderful [20:12:28] There are abandoned files from 2008 lying around and we don't really care [20:12:33] performance decreases with an increase in the number of files [20:12:34] Cause abadonment is rare now [20:12:36] damn. unintentional burn. [20:12:36] but that's not what I'm worried about in thwe short term [20:12:54] so, if we don't clean up unused crap, it's a waste [20:13:07] Also, more thumbnailing for recently uploaded files [20:13:14] *RoanKattouw_away goes back to watching TV, commercials are over [20:13:19] the scalers seem very stable right now [20:13:24] so, I think it's ok [20:13:55] except for that little blip a few minutes ago, yeah [20:13:56] and as more rsyncs finish the scalers will have less load... [20:14:11] but performance will decrease :-D [20:14:16] true [20:14:24] dddd fffff ssss... [20:14:28] hurry up nelson :-D [20:14:28] heh [20:14:30] indeed [20:14:51] ok, where was I. doing something unrelated to work... right. [20:15:04] yeah. I'm going back to preparing for the talk :) [20:15:16] enjoy [20:15:22] (but the site's all yers) [20:15:25] heh [20:15:35] well, I'm going to be kind of occupied for a few hours tonight [20:15:42] so hopefully someone else will be watching too [20:15:45] RobH: ^^ [20:15:56] notpeter: ^^ [20:16:26] :-D [20:16:26] ? [20:16:31] tah! [20:16:31] what did you guys do to fix it this last time? [20:16:44] fix what? [20:16:47] the image scalers? [20:16:59] apergos suspended the rsyncs [20:17:11] I'm going to start them up when load decreases [20:17:21] it's on ms5 in a screen session [20:17:25] suspended via ctrl-z [20:28:17] heh. how awesome to be able to suspend wikipedia with ctrl-z [20:28:30] i want to be able to do that to human beings. [20:29:49] well you can't suspend a login shell [20:29:54] so there's your trouble :-P [20:30:02] "Hey, let me tell you about my level 17 paladin!" [ctrl-z] [suspended] [20:30:10] HAH! [20:30:16] you don't want to suspend them [20:30:18] be real [20:30:24] you want to kill -9 them [20:30:33] so, last night i sat on the porch and wrote a long blog post about kurt cobain while smoking several cigars. [20:30:43] how many cigars can you smoke at once??!! [20:31:03] i came in and gave stacey a kiss and she said, "kissing you is like licking a dirty ashtray. kissing me must be like licking a clean ashtray." [20:31:16] well, not in parallel. [20:32:17] ugh [20:32:20] that just might be tmi [20:33:04] i thought it was funny. [20:36:52] neilk_: Hey you got any core changes to deploy tonight, or just Ext:UW? [20:37:04] nothing core [20:37:07] a few config changes [20:37:10] OK [20:37:13] Cool [20:37:23] I have a core API change for next week probably [20:37:36] will email you soon about it [20:37:39] If could write those config changes up and get them in my mailbox before midnight? [20:38:00] sure [20:39:27] And I'm glad that core change is an API change and not some change in filerepo stuff I don't know much about [20:43:09] neilk_: + if ( message.indexOf('~~~~') == -1 ) { [20:43:20] Are you aware that ~~~ also has certain magic signature behavior? [20:43:27] yes [20:43:30] I didn't write that [20:43:40] I suggested both. I HAVE CHAT TRANSCRIPTS [20:43:40] + $j( '#contentSub' ).html('Please let us know what you think of Upload Wizard!'); [20:43:47] haha [20:43:52] That one has hardcoded English [20:44:05] Kaldari said he was working on i18nizing it. [20:44:12] + data: 'action=edit&title='+encodeURIComponent(mw.UploadWizard.config['feedbackPage'])+'§ion=new&summary='+encodeURIComponent(subject)+'&text='+encodeURIComponent(message)+'&format=json&token='+encodeURIComponent(token), [20:44:28] Did you know that jQuery.ajax (which this is a parameter to) also takes data: { 'action': 'edit', .... } ? [20:44:40] I didn't write that. [20:44:53] haha, nah as I said it I thought so [20:44:58] Kaldari said he wanted to use mw.Api which makes this yet simpler, but just wrote it the old fashioned way. [20:45:04] I said "Did you know" and I went hmm he probably does [20:45:10] Also, this was in there last time, but: [20:45:11] + var $input = $j( '' ); [20:45:28] ok I *did* write this one. [20:45:34] your point? [20:45:49] That kinda stuff scares me. I think all of these values are sane, but there's a few more cases and IIRC there is one in last week's review notes where one of the values wasn't sanitized [20:45:52] Or escaped [20:45:52] the comment nearby notes that IE6 will not accept it done jQuery style. [20:46:08] I know [20:46:20] ok I can just do escaping, that's reasonable [20:46:20] But there's this function that will escape stuff for HTML [20:46:31] I actually don't remember the name because I 'grew up' with jQuery [20:46:41] (cue "children *sigh*") [20:47:01] well, this is not an HTML escape, technically it's attr escaping [20:47:05] which is yet different [20:47:49] Right [20:47:56] maybe you are thinking of encodeURIComponent [20:48:00] I believe there's a function for it somewhere [20:48:02] No, that's for URLs [20:48:04] I know [20:48:11] Let me look in, of all places, ProofreadPage [20:48:30] escapeQuotesHTML() [20:48:40] ok great... is that just global? [20:48:47] Common.js ? [20:48:57] Wait [20:48:58] No [20:49:26] Oh hah [20:49:28] It's in wikibits [20:51:40] didn't Trevor deprecate wikibits and copy everything somewhere else? [20:58:17] neilk_: No, afaik not yet [20:58:28] it's deprecated but in atleast 1.18 still in the global scope [20:58:38] mw.html.escape is the succesor by the way [20:58:45] great [20:59:02] thanks Krinkle [20:59:08] but. [20:59:36] I dont know in what context this is used, it may be easier to do element building with jQuery, can you roughly show how it's used ? [20:59:59] IE6 won't accept changing input type "after the fact" [21:00:31] been known to go awry with name & id too IIRC [21:00:35] ie. $( '') is not recommended, instead $("", { attributeName: value } );, jQuery will handle it internally [21:01:01] + // XXX help? [21:01:05] passing an object as second argument afaik will not do it 'after the fact' [21:01:56] Krinkle: ok, I didn't know about that style [21:02:05] *neilk_ looks it up [21:02:44] It's new in 1.4 [21:03:09] + . 's.innerHTML = "

";' [21:03:14] More lack of escaping [21:04:00] Although it's not as easily fixable in this case. I guess you could use the .src property, then you don't have to escape for attr, but you'll still have to escape $wgExtAssPath for JS (!) [21:04:10] escaping the $wgExtensionAssetsPath [21:04:13] seriously? [21:04:58] RoanKattouw: no offense, but I am not convinced these comments are helping me write overdue software that Erik is pinging me hour by hour about. [21:05:09] haha [21:05:11] I'm sorry [21:05:14] I'm done, too [21:05:17] That was all [21:05:42] And before reviewing I did think "hmm fixing bugs is more important than cleaning this stuff up ATM" [21:06:05] RoanKattouw: no offense taken and I should be encouraging you to do this kind of careful checking. Sorry I am a bit miffed about other things ATM and taking it out on you. [21:06:11] neilk_: That style is actually fairl similar to mw's Html class now that I think about it [21:06:15] (with the attribtues etc.) [21:06:19] No worries [21:07:03] wait, mw has an html generation class? [21:07:09] .. [21:07:25] that seems redundant [21:07:31] I dont know if it's very recent, I'm not that long around here. [21:07:45] neilk_: How so ? [21:07:50] It's more recent than the rest of mediawiki.js [21:07:52] !class Html [21:07:52] --elephant-- See http://svn.wikimedia.org/doc/classHtml.html [21:08:37] class, I didn't mean mw.html JS [21:08:40] I was thinking that jQuery covers html generation already [21:09:12] oh you mean mediawiki in PHP [21:09:15] yes [21:09:17] yes of course [21:09:29] doesn't take away the fact that there is a mw.html constructor as well [21:09:33] I've heard that some people do program MediaWiki using primarily PHP. ;) [21:09:42] which, as you say, looks redundant, with one important difference [21:10:07] jQuery returns (and creates) elments, whereas mw.html JS (just like the PHP class) only makes html [21:10:15] so things like images and scripts are not loaded directly [21:11:39] true [21:13:07] any of you guys know what this might be about: [21:13:09] /usr/local/apache/common/docroot/commons/w/thumb.php?w=17&f=WMA_button2b.png [21:13:10] ? [21:13:49] If that were a URL, it'd make sense [21:13:59] Get a 17px thumb for File:WMA_button2b.png [21:14:10] But that's foo.wikipedia.org/w/thumb.php?... [21:14:21]