[17:20:23] RoanKattouw: howdy [17:20:36] I saw your review of that debug mode patch [17:20:39] interesting stuff [17:23:26] Yeah I poked Krinkle about it too [17:23:33] He saw it but had to leave [17:23:50] He didn't quite seem to understand the second issue so I'll explain that when he gets back [17:27:12] Speaking of the devil [17:27:23] Krinkle: You're right about the first issue, only=scripts will do i [17:27:27] The second issue is more involved [17:27:39] The issue is not with the scope of the done variable [17:28:00] But with the fact that we're loading multiple files in parallel, and the first one to be addScript()ed need not be the first one to load [17:29:19] The done variable does its job, which is to ensure every file in the module has loaded before it fires the callback, just fine. It's the loading order of the individual files within the module that's the problem [17:29:29] I think it needs to be done multi-stage rocket style [17:30:00] i.e. something like addScript(A, function() { addScript(B, function() { addScript(C, callback); } ); } ); [17:30:07] Hm... so we can either keep it simple and make sure the loader wont addScript untill the previous one is callback'ed [17:30:08] or.. [17:30:26] make implement smarter to not implement it untill it has to, but the request can be made ahead of time. [17:31:06] in the latter concept we dont have to hold back http requests, assuming that is possible. [17:31:36] I don't follow [17:32:04] The first thing you're saying is the same thing I was saying, I think. addScript(B) from A's callback [17:32:23] Oh, I think I get it [17:32:30] Ah, wait [17:32:37] No, this won't work, we're talking about debug mode [17:32:41] Right [17:32:50] Implement has allready happened at that point [17:32:58] In debug mode, you're grabbing raw files so requesting==executing [17:32:58] implement is calling addScript for the given urls. [17:33:03] Yes, indirectly [17:40:39] hi Krinkle [17:40:44] Hey [17:43:34] did you see Roan's short novel on that patch we worked on in Berlin? [17:50:34] TrevorParscal: Yes [17:51:46] Looks like we didn't consider every single edgecase. [17:56:33] I pitched a sprint with you, roan and I in the same place at the same time for about a week - in late july - we shall see how that goes [17:59:34] I guess it could end up being mid-July instead, but whatever [18:00:11] are you coming here? [18:00:30] RoanKattouw: Well, not before July 16 [18:00:34] Ah, OK [18:00:49] I'm probably gonna be off the hook school-wise a week earlier than that [18:00:52] Or so I hope [18:01:02] RoanKattouw: You've got mail. [18:01:29] Ah, cool [18:03:00] Gotta love this week's episode of "domas vs. foundation-l" XD [18:03:03] RoanKattouw: Krinkle: ideally we would bring you both out on the same flight, on Sunday the 17th and drag you both to OSCON, then send you off sometime after that (30th or 31st) [18:03:22] we will see if that can actually happen though [18:03:46] Yeah ideally we give him the same itinerary as we're trying to get for me [18:03:49] I'm hoping we can do an RL2 sprint in the week of July 18th [18:03:57] do we have any good examples of api documentation? [18:04:06] Remember, there's Wikimania right after OSCON, too [18:04:06] hit api.php [18:04:19] RoanKattouw: assuming any of us actually bother to go to Wikimania [18:04:21] (Gotta fly out on the 31st to make it to the hacking days on Aug 2) [18:04:33] Well as you may remember we're doing a talk there [18:04:41] (If they accept it) [18:05:08] right, also it may not really be worth it if travel is indeed so tight that it means we can't do other things [18:05:15] danke shoen. gsoc protege is looking to write up api documentation before coding. [18:05:16] It'll be fine [18:05:24] We'll have a week and a half [18:05:25] also, I'm increasingly uninterested in this Wikimania [18:05:30] OSCON doesn't "really" start until Wednesday [18:05:43] RoanKattouw: yes, that's the day we would show up [18:05:52] or maybe 1 day prior [18:06:07] Yeah maybe the night before [18:06:23] But we'd still have 7 working days (one week + Mon + Tue) [18:06:36] right [18:07:16] And I, for one, am determined to not let WMF's attitude towards X spoil X for me, for pretty much any value of X [18:07:30] basically, if our talk at Wikimania is accepted, I will consider going - but the other talks I'm on (visual editor and testing panel, of which i think Krinkle should be one instead) aren't really going to be very useful [18:08:10] what's up? [18:08:45] it would be fun to do some travel, but maybe not that fun to go to Israel - I'm just a bit more on the fence than I was before [18:08:47] OSCON sounds interesting, hadn't looked at it before. [18:09:22] Krinkle: OSCON is also insanely expensive, unless you're presenting or work for an organization that got some extra free passes [18:09:28] (And guess what) [18:09:32] i seem to have a special talent for destroying usb connectors [18:09:44] managed to horribly mangle this headset's connector and it no worky no more :( [18:10:08] brion: I have your headset that clicks btw [18:10:14] Anyway, the schedule for OSCON looks awesome [18:10:15] you left it at my house at some point [18:10:22] hah [18:10:33] TrevorParscal: bring it in to the office and i'll just use that one :D [18:10:34] brion: I have a similar talent. My earbuds usually are either volume 0.01% after 6 months or my soundcard is up or replacement. [18:10:36] RoanKattouw: Krinkle: we supposedly have like 6 3-day passes to OSCON [18:10:38] i've got another at home actually [18:10:41] Somehow I mess up the connector [18:10:52] brion_: will do - tomorrow of course [18:10:56] thx :D [18:11:18] yeah i managed to get the headset cord caught on my office chair while it was plugged in to my desktop tower [18:11:34] spin that around and *snurk* 45 degree angle is not how usb is designed [18:11:41] brion_: you can always blame a cat [18:11:48] I blame children [18:11:54] i, too, am finding this wikimania to be less interesting. [18:11:55] it works quite well [18:11:56] there's no office cats, alas [18:12:14] oh, yes... you can blame peter then [18:12:18] hehe [18:12:34] blaming peter should be our new "go to" excuse. [18:12:39] "peter broke the site." [18:12:53] "peter won't eat meat" [18:12:57] that does work well [18:13:19] I'm a fan of this [18:13:22] Who's peter ? [18:13:27] new ops guy [18:13:33] aka "whipping boy" [18:13:38] LET THE HAZING COMMENCE [18:13:52] He was at the Berlin conference [18:14:04] as the most senior ops person here who hasn't taked down the site, I think I deserve a little more respect.... [18:14:12] just sayin' [18:14:21] you wiped out the puppet keys [18:14:24] that counts ;) [18:14:31] notpeter: most senior? [18:14:44] en wiki was never down [18:14:45] Ryan_Lane, go root, become peter, kill something.. [18:14:46] ;) [18:14:53] TrevorParscal: who hasn't taken down the site. [18:15:02] TrevorParscal: the former portion heavily depends on the latter portion [18:15:09] indeed [18:15:21] i see... aren't you alone in that category though? [18:15:26] asher [18:15:29] Has Asher yet? [18:15:29] also hasn't [18:15:30] ah.. [18:15:42] and I've got a month on him [18:15:53] notpeter enjoys 120 days of seniority [18:16:00] only a month? [18:16:05] I thought it was much more than that.. [18:16:11] "most senior ops person who hasn't taken down the site" just means "most junior ops member" :) [18:16:19] indeed [18:16:35] TrevorParscal: my first day was... march 7? something like that [18:16:40] haha [18:16:45] brion_: potatoe, potatoe [18:16:57] (those were pronounced differently) [18:17:08] use {{IPA}} [18:17:18] Po-tah-to , po-tae-to ? [18:17:20] I fucking love drinking IPAs!!! [18:17:23] .ogg or it didn't happen [18:17:54] seriously, everyone takes down the site at least once. it's just a matter of time ;) [18:18:35] Even non-ops people do [18:19:05] RoanKattouw: who should take credit for taking the site down when we melted the API servers with 20k req/sec? [18:19:16] As variation on Godwin's law, every shell user will eventually break the site, just a matter of time [18:19:25] Oh, right, there was that [18:19:33] - eventually ;) [18:19:36] Probably you, you wrote it [18:19:40] I failed to catch it in review [18:19:52] I have never taken down the site. [18:19:52] I've had incidents caused by code I actually wrote [18:20:06] You don't have shell access though [18:20:15] If and when we give it to you, the timer starts ticking [18:20:33] it's actually a little sad how easy it is to break the site [18:20:36] hence why i do not want shell access, and will refuse it if thrust upon me. [18:20:41] we should make this more difficult at some point :) [18:20:49] Worse than taking down the site was when we had that "bug" in PrefSwitch that sent the user through a login screen and then included their password in plain text in the returnto URL (because we used $_REQUEST instead of $_GET) [18:21:18] Yeah that was also interesting [18:21:41] not sure there were any accounts that got hijacked form that, would be hard to know, but that was bad [18:22:42] I think Roan is even guilty of a sql injection [18:23:31] when i was at EA they kept trying to give me the SYSTEM password for the oracle instance that handled peoplesoft. [18:23:35] I kept refusing. [18:23:51] "If I don't know the password, I can't log in and break anything, so nothing can be my fault." [18:24:04] did bobby drop tables pay you a visit Roan? http://xkcd.com/327/ [18:24:07] unspoken: "also, you can't call me at 3 am and have me try to fix something" [18:24:50] oh, little bobby tables [18:25:09] Bryan: You've said that a few times yet I have no recollection of that [18:25:28] I seriously mean that. I don't mean to try to save face, I've made plenty of other embarrassing mistakes :) [18:25:29] oh, oakland. you lie when you say "sunny side of the bay" [18:25:30] Bryan: now you must prove it with a revision number! [18:25:51] Roan, everyone in here has done something hideously stupid on a production site. [18:25:57] jorm: oakland is full of lies [18:26:05] Of course [18:26:13] stacey and her dad are out on a bike ride. whupz. [18:26:33] I've introduced PHP syntax errors to eswiki and mediawikiwiki, overloaded the site with LU, caused fatal errors on every page view when LU was disabled... [18:26:50] And probably more [18:27:30] *jorm thinks. [18:27:37] Hrm. I've taken down Sun Microsystems before. [18:28:03] http://www.gaijin.com/2010/03/taking-out-sun-microsystems-by-accident/ [18:28:08] TrevorParscal: Is there a way I can tell whether the user submitted an e-mail for a given rating? Is that association stored anywhere? [18:28:27] oakland is full of awesome. [18:28:51] well, there's that email checkbox [18:28:52] RoanKattouw: https://bugzilla.wikimedia.org/show_bug.cgi?id=14526 [18:29:09] http://www.mediawiki.org/wiki/Special:Code/MediaWiki/31260 [18:30:01] Ah, there we go [18:30:04] not right now it's not. it's full of rain. [18:30:10] So you fixed it, that's why you remember [18:30:25] lol, so I broke the site in '08 [18:30:34] I thought it was later than that [18:30:42] jorm: sunny side of the bay != has protective anti-rain shielding [18:30:45] presumably [18:31:36] i beat you all [18:31:36] TrevorParscal: Sure but how do I tell in the DB? Do we store it there? [18:31:39] i broke the site in 2002! [18:32:03] *jorm is going to strive to never be in a position to break the site. [18:32:09] *Krinkle catches up on the channel log [18:32:57] surprisingly little of value to catch up on. [18:33:49] RoanKattouw: helpimprove-email will be int eh afp_value_text [18:33:52] but the problem is [18:34:07] hmm [18:34:12] no, that works [18:34:18] yeah, that's where it's at [18:34:23] it's in a piped list [18:34:38] sorry, that's probably not very easy to get at in a single query [18:34:44] Well, i wouldn't wanna have missed the bobby drop tables commic [18:34:50] it should always be the lat one on the list though.... [18:35:46] TrevorParscal: RoanKattouw Any idea why both the success and error callback fire in IE6 and IE7 ? [18:35:54] This might be affecting more than just our test suite here [18:36:06] Krinkle: you have both asked and answered that question [18:36:11] haha [18:36:19] IE6 and IE7 are evil? [18:36:26] It's been making TestSwarm look like a christmas tree since day 1. [18:36:30] you are talking about RL? [18:36:33] Yes. [18:36:38] Krinkle: The bobby tables comic is classic. I always assumed everyone knew it, but surprisingly many people in my WM10 audience seemed to be seeing it for the first time [18:36:42] mw.loader.using( 'is.awesome', ok, error) are both called. [18:36:47] somehow [18:36:52] probably because we did something wrong with the onreadystatechange thing [18:37:11] Hm.. [18:37:15] I don't think that's likely, but it's possible [18:37:19] Looks like it's fixed. [18:37:29] Does it happen in IE8-in-IE7-compat-mode? [18:37:32] it's tabIndex that's erroring now. [18:37:36] now mediawiki.js [18:37:38] my bad,. [18:37:42] not* [18:37:43] In that case you can set a breakpoint in the error callback and see a stack trace [18:37:53] if we don't resolve the onreadystatechange thing correctly, RL might get into a state where it thinks there was a loading failure and call the error handler [18:38:43] it was a bug in the test suite [18:38:44] http://www.mediawiki.org/wiki/Special:Code/MediaWiki/88705 [18:38:57] the 'delete awesome' was doing something funky. [18:39:20] however tabIndex seems malfunctioning in IE6+IE7 still [18:39:43] of course it is [18:39:43] http://toolserver.org/~krinkle/testswarm/job/63/ [18:40:00] It's only checking the attribute, nothing else. [18:41:54] TrevorParscal: Thanks, got it [18:42:21] I need to be looking for LIKE '%helpimprove-email%' instead of ='helpimprove-email' because it's a pipe-separated list of things, that was the catch [18:42:28] somewhere in the IE code.... http://pastebin.com/3LjeQ9i7 [18:42:46] RoanKattouw: yeah, that will do it [18:42:47] haha [18:43:06] and because we hide, but not remove the checkbox, it's always checked in subsequent saves, so it's persitent [18:43:10] i verified this [18:43:13] Krinkle: http://xkcd.com/221/ [18:45:43] TrevorParscal: Seriously tho, http://toolserver.org/~krinkle/testswarm/?state=runresults&run_id=315&client_id=706 #jquery.tabIndex_in_IE7 [18:45:57] it's the Obfuscator [18:46:11] ok [18:46:35] The MS Obfuscator for Trident, that Trevor captured quite well in one of his drawings :P [18:47:35] Oh right [18:47:48] Trevor, is that drawing up on the web yet? :) [18:49:54] And then there were Wikimedia VMs to populate the test swarm. Awesomeness is coming to town, (Thanks Ryan_Lane) [18:50:41] http://www.trevorparscal.com/stuff/msiejs/ [18:51:07] Krinkle: Sweet [18:51:15] http://toolserver.org/~krinkle/testswarm/user/Wikimedia%20Foundation [18:51:29] http://toolserver.org/~krinkle/testswarm/ all browsers have atleast one version covered now. [18:51:33] Oooooh [18:51:36] \o/ [18:51:41] Now I remember where I know that thing from [18:51:43] and I killed all the selenium VMs :) [18:51:51] You drew it on one of the South side 3rd floor whiteboards in Jan 2010 [18:52:01] south siiiiiide [18:52:02] yeah [18:52:10] i diagramified it in berlin [18:52:16] Or wait, North side [18:52:22] yes, north [18:52:23] Anyway, the "old" tech side [18:52:28] Minna St side [18:52:32] it was between my and michael's desks [18:52:48] Oh right, that one [18:53:33] We used it for our table of browsers and features too [19:02:17] http://fluidproject.org/blog/2008/01/09/getting-setting-and-removing-tabindex-values-with-javascript/ [19:02:50] werid [19:02:54] Why is it not working. [19:03:18] oh cool somebody made a tag cloud on codereview when i wasn't paying attention :D http://www.mediawiki.org/wiki/Special:Code/MediaWiki/tag [19:03:40] oh, yeah I saw that... [19:03:54] m name is in blue... [19:04:16] oh, meeting! [19:10:38] Krinkle-away: Already over [19:11:25] Rather than the todo tag, we could just decide what to call the "needsmore" status and sort that Add path filtering to https://bugzilla.wikimedia.org/show_bug.cgi?id=28274 [19:11:32] cp fail [19:12:42] i'm not sure how much need there is for a resolution for "it's ok as it is, no action is required, but oh by the way it might be nice to make it better someday" as this will apply to everything currently marked 'ok' :) [19:13:38] haha [19:13:55] but also, it's applicable where stuff isn't omg this is broken... [19:20:28] so, what are we gonna do with broken extensions? [19:20:42] which are not really used [19:20:45] like the most on http://www.mediawiki.org/w/index.php?title=Special:Code/MediaWiki/status/fixme&path=/trunk/extensions&offset=77974 [19:21:01] ArticleEmblems, PageObjectModel, TemplateAdventures, etc. [19:23:12] if they're not deployed, we sorta don't care [19:23:34] if it's security issues we should do something about it [23:40:55] robla: pdhanda: ^demon|away -- phonecall? [23:41:09] oh...yeah, we're on x2003 now