[00:14:00] this gem is from File.php [00:14:26] if ( ... $thumb->getUrl() != $this->getURL()) ) { [00:14:27] } [00:14:40] hehe [00:15:14] must be a check to use the source on 'small' source images no? [00:15:17] No, I don't mean get the Url, I mean get the URL [00:16:31] well, getURL isn't even defined in this class [00:17:44] are php method names case insensitive? [00:18:12] nope [00:18:13] so [00:18:37] the only possible way that could ever be true would be if getUrl was changing over time ;) [00:18:56] this appears to be the result of a search & replace, there are getURL and getUrl all over the place :( [00:19:32] not that my opinion means dick here, but i prefer getURL over getUrl, since URL is an acronym. [00:19:57] getUrl looks like txtspk to my brain. [00:20:08] IMO the needs of camel-case trump the need to put acronyms in all caps. [00:20:11] g3+E4r1 [00:20:22] people always get confused if they try to keep acronyms all caps. [00:20:26] e.g. [00:20:26] YOUR OPINION IS WRONG. PLEASE GO STAND BY THE STAIRS. [00:20:47] getHTTPSomething [00:20:51] getHttpSomething [00:20:57] latter is better [00:21:14] i shall humbly disagree. [00:21:34] but that is purely a matter of taste. [00:22:57] flipzagging: i read this essay a while back called "always deploy from trunk" or summat. your mail made me think about it. [00:24:07] here: http://www.paulhammond.org/2010/06/trunk/alwaysshiptrunk.pdf [00:24:15] that's the same guy [00:24:20] he's my ex-boss ;) [00:24:21] there are some things i agree with there, but some things i radically disagree with. [00:24:22] oh. [00:24:28] well, then. i'm betting you've read it. [00:24:32] lived it [00:24:55] I feel the same way, I'm not saying hammond has all the answers, believe me. [00:25:07] i think that having a bunch of ifdef statments is an interesting principle, but probably difficult to use with large development teams. we kind of have that with extensions, mind you. [00:25:11] we can enable, disable, etc. [00:25:18] i think that shipping trunk is $INSANE [00:25:34] there are problems, but it's not insane [00:25:39] do you enjoy using flickr? [00:26:02] explore: why do you think it's insane? [00:26:12] i actually don't use it much, tbh. i kind of liked the previous version better than the new one, but mostly because i don't like the way it behaves when i click on an image now. [00:26:21] because you assume that when developing, you write code that can harm data in ways that can't be reverted. [00:26:39] at Flickr, you have to make every change revertible on like day one [00:26:53] well, our current woes with the resource loader and the vector movement are a good example. [00:26:57] so you are thinking about deployment before you even start [00:27:12] that's not a bad principle, and i think it works. [00:27:22] i think it's $INSANE when you have more than, like, 5 people. [00:27:38] why? [00:27:40] you have to be small enough to apply freezes. [00:27:47] no you don't [00:27:54] how is that? [00:27:59] why would you need to freeze? [00:28:11] to at least do smoke tests? [00:28:35] wait. are you telling me that flickr pushes code to production without even smoke testing it? [00:28:43] it's tested [00:28:49] but it goes like this [00:28:54] developer develops [00:28:57] okay, good. i was about to lose many brain cells. [00:29:14] well, make sure you hold on to your brain cells, because flickr has no QA. [00:29:29] urk [00:29:38] anyway [00:29:39] wait! NEITHER DO ! [00:29:41] developer develops [00:29:48] then says "I want to deploy" [00:29:51] our testers are the world! muahahah. [00:29:54] hit deploy button [00:30:10] deploy button creates a mini-deploy environment out of the SVN head at that moment. [00:30:19] runs smoke tests [00:30:44] then developer examines that, particularly for whatever feature they are working on, also does general check that the most-hit pages on flickr aren't exploding. [00:30:57] and then if that's okay hits the "GO" button [00:31:12] so they have a rather robust auto-testing environment. [00:31:17] that's always nice. [00:31:18] not really [00:31:37] maybe i misread. [00:31:38] the smoke tests are very basic. Flickr is in general an untestable mess of PHP code, just like MediaWIiki. [00:31:51] by "runs smoke tests" you mean the developer does them, or they're automated? [00:31:54] both [00:31:58] automated API tests [00:32:03] ah. [00:32:08] and a quick check by the developer. [00:32:23] (i'm curious about this because of some meeting tomorrow about this type of stuff) [00:32:24] and then badaboombadabing deployed to several thousand servers. [00:32:36] flickr should have more tests [00:32:47] i think i got spoiled by working at Elemental. [00:32:50] and last I was aware of it, they were moving in that direction of making things testable. [00:32:57] jorm: how so? [00:33:43] BTW, I'm not going to hold up Flickr as a paragon of Doing It Right [00:33:44] well, it was NOT okay for you to check in code that didn't have as close to 100% test coverage as you could get. and there were *constant* automated builds, and they always ran the tests. [00:33:58] and it was just culturally NOT okay for you to break the build. at all. [00:34:02] hrm, now actually I disagree with that [00:34:19] people always optimize the metric [00:34:25] and if there was one thing wrong - like, even a simple jsp miscompilation - you got made fun of. [00:34:29] it's really easy to have 100% test coverage that is meaningless [00:34:53] well, yeah, but then, we were about 6 or 7 people. so we all knew what each was doing. [00:34:54] how long did it take tests to run? [00:35:05] a full build was about an hour. [00:35:10] sigh [00:35:16] now that is really depressing [00:35:20] that was build+test. [00:35:22] yeah, it is. [00:35:47] which is why 'shelling out an extra grand for a faster processor' ALWAYS made sense. [00:36:02] I'm assuming that the reason the tests were slow was because you required a database or other I/O [00:36:15] beyond that, we had 3 full time QA guys, a release engineer, and a tech writer. [00:36:22] it was nearly 100% database, yeah. [00:36:26] yeah [00:36:36] now this is why you can't start testing without rethinking [00:36:41] build + no tests was only about 5 minutes. [00:36:54] but still. shaving 5 minutes to 4 is a huge win over the course of a week, even. [00:36:58] if your code isn't isolated from the database (i.e. you can't mock out the db) then you are DOOMED. [00:37:41] if you can mock the db then you can run thousands of tests in like seconds, usually. [00:37:52] and developers stop breaking the build. [00:37:55] i used to get snapshots of the db for personal use. [00:38:06] we had our "dogfood" server and that's what i would steal. [00:38:22] right, that's what you can do as a stopgap issue [00:38:35] because the automated scripts might dump 100 rows into some tables, but what i needed was more like 100,000s [00:38:46] Flickr has major problems there. It cannot be run on a developer's machine, period [00:38:51] "how is this table going to work with 100 rows? easy. now how about 800,000?" [00:39:26] hm [00:39:32] but that's still not all that realistic [00:39:39] yeah. i'd actually take multiple snapshots and load them from time to time. [00:39:39] unless you had only one db server in prod [00:39:45] no, it was totally realistic. [00:40:10] for what the product did. i might have to figure out a way to display hundreds and hundreds of thousands of rows to the user. [00:40:25] elemental was a security snoop-a-tron app. distributed. [00:40:27] well it might be, but one machine running db + code is very different from a world where they are separate. [00:41:04] for enterprises. and you might want to walk through all the hosts that a particular machine had touched over the course of a week, and that could get really big really fast. [00:41:08] anyway I don't know your world, I've worked on sharded crap for the past 5 years, almost exclusively. [00:41:24] web two point oh. [00:41:35] oh, how the DB was structured was irrelevant to me. i just needed a way to pull it up. [00:41:47] so i took snapshots of single db systems that i could run locally. [00:42:02] but the bigger snapshots would grind my wee desktop to a halt when being run actively. [00:45:32] ugh: "Thought: The present is an infinitesimal moment smeared into existence my memory and the intellectual projection of the future. If "present" was a musical note, and you had no capacity to remember or project an anticipated rhythm, a symphony would sound like a discordant series of sudden unremembered tones. Thus to be present, one would need to take themselves out of the equation, or be .now., with no future or past." [00:47:11] You know what's awesome [00:47:16] twitter is more reliable that at&t [00:47:26] for contacting people who are two blocks away from me [00:47:30] that doesn't surprise me. [00:50:43] there must be some way to do texting over the interwib [00:50:57] I seem to remember mailing to number@att.com did it, once. [00:51:05] it still does. [00:51:14] but I'd rather have threaded convos. [00:51:35] emailing to mynumber@cingularme.com still works for me. [00:54:26] except, one has to know the carrier to do email-to-text [00:54:28] fail [00:59:34] hrn. they are going to bury that whale. [05:53:55] @txt.att.net works. [05:54:02] I have cron text me every day. [17:11:25] adam_miller, or jorm_ , the prototype for Article Feedback is http://prototype.wikimedia.org/articleassess/ , right? [17:11:35] yes. [17:11:42] sorta [17:11:43] http://prototype.wikimedia.org/articleassess-d/Main_Page [17:11:51] ah [17:11:52] thanks [17:11:57] i believe the d in that one stands for deployment [17:12:11] oh. so we have a new one. [17:12:13] RoanKattouw could tell you better what the difference between the two is [17:12:19] Adam is right [17:12:22] but i believe one is running mediawiki trunk [17:12:25] No [17:12:27] offhand, the first one doesn't work. [17:12:28] -d runs on 1.16wmf4 [17:12:29] Either way [17:12:40] test.wikipedia.org runs AA too [17:12:50] And en.wikpedia.org is gonna be dark-launched like right now [17:13:39] I'm getting an error when submitting the survey [17:14:01] "An error has occurred, please try again later" [17:14:14] probably the thing i was seeing yesterday [17:14:30] where you dont have a session initialized or something....RoanKattouw did you get that problem fixed? [17:14:47] I didn't rate the article before providing feedback, could it be the cause? [17:15:02] Yes that got fixed [17:15:15] Had to add one call to wfSetupSession(), that was it [17:15:25] jorm: What is the name of the category again? [17:15:49] http://en.wikipedia.org/wiki/Category:Article_Feedback_Pilot [17:16:27] Thanks [17:17:46] the Article feedback tool on testwiki seems to be missing message strings [17:18:16] Like which? [17:21:40] like, all of them; also, the sidebar items don't expand [17:21:40] meh [17:23:51] OK, I have now dark-launched to enwiki [17:24:02] Such that AA is only enabled if you're logged in as me [17:24:11] I can add you guys to the list if you like, just give me your enwiki username [17:24:37] http://img69.imageshack.us/img69/1332/nomessagestrings.png and http://img94.imageshack.us/img94/7030/testwikisidebar.png [17:24:56] first one: no message strings for article feedback; second one: can't expand menu items in the sidebar [17:25:18] btw, note what happens if you uselang=fr [17:25:24] trevor and i are looking at it. [17:25:42] jorm_: You mean that the survey doesn't translate? [17:25:46] guillom: Looks like JS breakage on test [17:26:22] so I guess I shouldn't worry about it [17:26:25] no, in the thing. [17:26:28] the tool. [17:27:38] guillom: Did you check your JS error console? [17:28:01] I have a JS error console? :) [17:28:11] Your browser does. [17:28:37] jorm_: You mean the 's? [17:28:42] yes. [17:29:22] guillom: Is 0 singular in French? [17:29:52] err, I don't know, why? [17:30:11] Neutre : 4.0 4.0 (7 évaluation) [17:30:19] Missing an s [17:30:45] Apparently expanding {{PLURAL:$2|évaluation|évaluations}} expands to the former [17:30:55] While in most languages zero results in the plural case [17:31:54] jorm_: I guess there are nbsp characters (not the entity but the actual character) in the fr messages and we're using .tetx() [17:32:05] .html() would quick-fix that [17:33:46] Ashlee, do I have to install firebug or something? [17:34:16] guillom: Tools --> Error Console [17:34:20] In Firefox. [17:34:28] ahah! thanks [17:35:32] Erreur : $j.ArticleAssessment is undefined [17:36:16] ^^ [17:36:17] Erreur : $j.cookie is not a function Fichier Source : http://test.wikipedia.org/w/extensions/UsabilityInitiative/Vector/Vector.combined.min.js?283s Ligne : 12 [17:36:25] whatever that means [17:36:33] It means weird crap is going on [17:37:05] RoanKattouw, if it's only testwiki it doesn't matter; you can add me to the enwiki dark launch if you want me to check if it happens there too (user:guillom) [17:37:30] testwiki works fine for me, try hard refresh [17:37:37] Will add you to enwiki dark; any other takeres? [17:37:43] *takers [17:37:57] been there, done that [17:38:04] (the ctrl-shift-r) [17:39:27] k. trevor fixed the issue. [17:39:43] me, clearly. [17:39:48] roan, i have a fix. sec. [17:39:52] OK [17:40:01] guillom: Done [17:40:05] jorm, are you talking about mine or another one? :) [17:40:11] another one. [17:40:19] the one. [17:40:30] Gimme commit links when committed [17:40:31] I don't encounter the message strings problem on enwiki [17:40:53] trevor looked at it; it was getting double encoded on the server side. [17:40:57] or summat. [17:41:49] Readable: 4.0 (1 ratings) [17:41:52] ^bug or feature? [17:41:54] RoanKattouw: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/73546 [17:42:04] Thanks [17:42:09] Any more coming? [17:42:24] guillom: Bug, not easy to fix without resource loader though [17:42:29] nope, that's it. [17:42:30] ok [17:43:03] the survey works for me as well [17:43:56] jorm: parsemag does not parse ''revised'' [17:44:14] is it normal that I don't get the survey link once I've filled it once? [17:44:25] (I don't get it any more on another page) [17:44:53] i honestly don't know the difference between them. it wasn't really my change, it was trevor's from my computer. [17:45:05] jorm: Therefore recommend that the parsemag change be reverted; presumably FormatJson::encode() is the active part in fixing ? [17:45:09] omg socks [17:45:19] parsemag parses things like {{SITENAME}} only, parse does ''italics'' too [17:45:24] RoanKattouw: no [17:45:26] no no no [17:45:44] the parseinline thing was the culprit, I just did the formatjson thing so my eyes did not bleed [17:45:51] OK [17:45:53] <^demon|away> RoanKattouw: parsemag does plural too. [17:46:09] ^demon|away: Statically, not dynamically with client-side values [17:46:19] <^demon|away> well of course. [17:46:37] TrevorParscal: But, like I said, parsemag breaks ''italics'' in the "article has been ''revised'' since ... " mesasge [17:47:44] Also, whoever wrote the summary for that commit, it's misleading. Don't call it a performance fix when it's really a bugfix [17:48:51] that was me; the way the json change was explained made it appear like it was going to be faster. [17:49:32] <^demon|away> If you didn't profile it, it's not true ;-) [17:51:07] *TrevorParscal thinks you guys should go a little easier on jorm, he's new, and he's a graphic artist that's acting as a developer right now - forgiveness and understanding are in order [17:52:23] I would have said he's a developer acting as a graphic artist :P [17:52:38] Exactly :) [17:53:02] *RoanKattouw has been reading jorm's blog where he claims he's a good dev or something along those lines [17:53:02] yeah, i'm none of those things. [17:53:06] he was hired as a user interface designer... [17:53:15] Design hobo. [17:53:21] i know fuck all about php is all. [17:53:42] and we love that he can develop, but go easy, he wasn't hired as an MW dev like most of you were [17:53:44] Either way [17:54:07] I think reverting the Brandon/Trevor rev and using .html() to populate those fields instead of .text() should fix it [17:54:26] RoanKattouw: sounds fine to me [17:55:00] whatever works. [17:55:13] OK I'll revert, Trevor could you poke at the JS? [17:55:21] (Don't forget to bump style versions and run make) [17:56:27] curious: did i do something to piss you off recently, roan, or am i being overly sensitive today? [17:57:03] The latter, probably [17:57:15] that's entirely possible. [17:57:24] If I bitch about something you did wrong in SVN, don't take it personally ;) [17:57:40] RoanKattouw: I'm not working on this project right now, please don't suck me in... please! I beg you... [17:57:55] TrevorParscal: OK, I'll do it then [17:58:10] jorm: I'm probably also undersensitive [17:58:14] *RoanKattouw hides behind Asperger's [17:58:20] RoanKattouw's on adrenaline :) [17:59:53] jorm: Also, thank you for asking instead of just silently assuming malice and hating me forever ;) [18:01:23] uh, you're welcome, i guess? heh. i like to know if i've pissed people off so i can either fix it or exacerbate it, depending. [18:04:21] what... you mean y'all *aren't* just jerks? :D [18:08:22] oh, yes we are! [18:08:29] brion: howdy! [18:08:37] yo [18:11:24] how's status.net land? [18:12:14] OK the nbsp problem is solved [18:12:20] Y'all ready to deploy for real now? [18:12:30] pretty good dude [18:12:37] how's the mediawiki land? [18:12:51] sure. i was willing to deploy without the nbsp fix, but so it goes. [18:13:05] i'm gonna trust your judgement on this entirely, roan. [18:13:37] i should be on the en dark bit as User:Jorm (WMF) [18:14:24] I mean killing the dark bit and making it available to the world [18:14:34] sure, why not. [18:14:38] crank it up. [18:14:56] *guillom prepares to push the blog post [18:15:12] Anyone else I should inform? [18:15:24] nah, alolita is sitting with me. [18:15:31] The Word Is Given. [18:15:50] Alright then [18:16:05] i CANNOT WAIT to start telling people how much their articles suck. [18:17:37] let's get this over with so I can get out of this article feedback black hole and go do Real Work™ :) [18:18:40] Amen [18:18:43] And we are live [18:20:02] RoanKattouw++ [18:20:31] blog post published, announcement sent to wikimediaaannounce [18:20:41] going to forward to foundation-l and wikien-l shortly [18:21:03] roan: good stuff thanks :-) [18:21:28] RoanKattouw+++ [18:33:20] wait [18:33:33] guillom: what happened? [18:33:54] someone tells me you have to be logged in to see the article feedback tool [18:34:00] *guillom checks [18:34:15] no [18:34:21] *guillom is relieved [18:35:03] Squid caching, my friend [18:35:13] Anons won't see the tool immediately [18:35:24] guillom: good testing [18:35:34] ok RoanKattouw :) [18:47:07] does anyone know why wikien-l is the only list to have a pretty info page? https://lists.wikimedia.org/mailman/listinfo/wikien-l [18:47:43] (also, I'm all lost with their unconventional design) [18:47:58] wikinews-l used to have a pretty info page (but doesn't anymore) [18:48:44] yeah the pages can be customized by list admins; some have been some aren't [18:48:51] and there's no consistency of design for those that are customized :) [18:49:27] i still kinda want to write a non-shitty mailing list manager, though nobody seems to get too excited about mailing lists anymore [18:52:57] Haha, the new XKCD is hilarious [18:53:09] Having been to California I totally get it [18:53:35] <^demon> Haha [18:55:54] :) [18:56:37] we're not ALL douchebags. [18:58:57] jorm, http://lists.wikimedia.org/pipermail/foundation-l/2010-September/061058.html [19:23:35] jorm, https://twitter.com/jeblad/statuses/25235177538 :D [19:47:06] he spelled "usability" wrong. [19:47:30] Perhaps it was intentional. [19:48:44] I've asked Howie to follow-up on foundation-l, answering those questions is his job, after all. [19:50:13] What does Howie do? [20:07:30] haters gonna hate. [20:53:52] Hmm, right. [20:54:34] Ashlee, « Senior Product Manager, Strategy » afaik [20:54:45] Right. What does he do? [20:56:34] http://lists.wikimedia.org/pipermail/wikimediaannounce-l/2010-May/000007.html [20:56:50] I don't know much more than that myself. [20:57:22] other than he's going to work with Erik in the Strategic Product Department thing [20:57:44] It's like these people have swallowed a dictionary of buzzwords and just randomly throw up. [20:57:56] :D [20:57:59] > In the longer [20:57:59] term, Howie will help us to build Wikimedia's product development [20:57:59] roadmap by commissioning and assessing research and analytics, and by [20:57:59] engaging in broad consultative processes. [20:58:12] So much noise, so little signal. [20:59:05] Well do you remember how communication around the phase V (final phase, remaining wikis) of Vector was messy? [20:59:20] Not really. [20:59:20] (Banners going up late, etc.) [20:59:31] I don't most of that was done publicly. [20:59:32] I guess the outside world didn't really notice, good [20:59:39] That's not really good. [20:59:42] On either side. [20:59:47] Either way, that was when Howie was out of the office for 2 weeks [20:59:51] You had little transparency and a lot of fuck-ups... [20:59:57] Okay. [21:00:06] He usually coordinated those things with communication and CentralNotice, TWN, God knows what else [21:00:07] So he's a RobLa type? [21:00:13] kind of. [21:00:14] And that kinda all didn't happen [21:00:17] Sort of but not really [21:00:36] Ashlee: By 'good' I meant that we didn't fuck up that badly then [21:01:11] RoanKattouw: A tree falling in the forrest still makes a sound. [21:01:19] I know [21:01:31] But it doesn't kill any people standing below it [21:01:43] Sure, it just rots unattended. [21:02:39] Other than that, Howie's been involved with some strategy decisions and statistics analysis back in the Stanton usability days [21:02:56] Couldn't exactly tell you what his entire job description is, I suggest you ask him :) [21:03:37] All right. [21:55:01] *robla reads the backlog to figure out what a "RobLa type" is supposed to be [22:06:18] robla: I think it was meant to mean "someone who involves the community in their decision making process" [22:13:31] Hah. [22:13:43] I was making a distinction between developers/coders and managers.