in the comment? [18:58:06] yes [18:58:10] dang it [18:58:38] fixed it [18:58:54] so tim ripped all michael's work out, again [18:59:13] or what? [18:59:24] Not really [18:59:30] He restructured the JS2 emulation stuff [18:59:42] You know like the wrappers in js2stopgap [18:59:49] He renamed most of those functions [18:59:55] Moved the files, etc. [18:59:57] well, UsabilityInitiative.hooks.php no longer contains class names for the modules [19:00:02] so, what's with that? [19:00:23] Oh that [19:00:31] Yeah so I guess Tim did take out Michael's changes [19:00:45] He didn't want near-deployment trunk code to be contaminated with experimental branch stuff [19:00:50] (my words, not his) [19:00:52] http://svn.wikimedia.org/viewvc/mediawiki/trunk/extensions/UsabilityInitiative/WikiEditor/WikiEditor.hooks.php?&pathrev=61566&r1=61565&r2=61566 [19:01:26] I'm not really sure what's going on there, I will need to investigate and get control of this whole transitional situation [19:01:48] there's too many people (both michael and tim) who are making major changes to our code with little or no communication to us [19:02:29] Yeah so basically Tim undid most of Michael's changes [19:02:54] I'll write a wikitech-l post later telling people to please not pull this kind of crap any mroe [19:03:03] Not without consultation anyway [19:03:26] no - I will take care of this [19:03:31] Sure :) [19:04:01] I looked at Tim's changes and approved of them; feel free to look through and judge them yourself [19:04:20] sounds like it is time for a developer log alla the server admin log :D [19:05:04] We have that, it's called SVN [19:05:11] And wikitech-l :) [19:11:08] RoanKattouw: highlight is throwing an error for me in firefox: Index or size is negative or greater than the allowed amount" code: "1 http://localhost/~adam/wiki/prototype/extensions/UsabilityInitiative/js/plugins/jquery.wikiEditor.highlight.js?264 Line 199 [19:12:23] adam_miller: Try changing line 192 to if ( e.offset < e.length - 1 && e.offset < e.node.nodeValue.length - 1 ) { [19:12:35] That should shut up the error [19:12:40] Hopefully it'll also result in sane behavior :P [19:13:33] RoanKattouw: same deal [19:13:48] WTF [19:14:23] can we get rid of these warnings too? The 'charCode' property of a keyup event should not be used. The value is meaningless. [19:14:24] adam_miller: Inside that if, can you do something like if ( e.offset < 1 ) console.log(e.offset); [19:14:34] adam_miller: That's jQuery's fault [19:14:38] RoanKattouw: are you not seeing the same thing? [19:14:55] I'm busy doing other things right now [19:15:05] But I haven't seen that error [19:15:11] Do you have templateEditor enabled by any chance? [19:15:19] Without templateEditor that code path should never be taken [19:15:40] consistently -52 [19:15:55] i do not have templateEditor enabled [19:16:25] -52 :O [19:16:31] These offsets should never be negative [19:17:10] Can you pastebin the contents of the article you're witnessing this on? [19:18:44] OK guys I'm gonna hijack the deployment prototypes for about 15 minutes to test my hackery [19:18:58] So they might be buggy or outright broken for a bit [19:20:00] RoanKattouw: http://pastebin.com/d7558c8f2 -> this one is giving me -15 now [19:20:08] not sure which one i was looking at that was -52 before [19:21:56] OK that's good enough for me, thanks [19:22:06] I'll paste it into my wiki and see if I get the same error [19:22:24] Are you doing anything in particular to provoke this error? [19:23:11] clicking into the iframe, moving the cursor around [19:23:19] basically anything i do to it [19:24:14] Ah got it [19:25:23] Oh wait I'm getting it on prototype on [[San Francisco]] now, interesting [19:26:00] adam_miller: Does this error go away if you do $wgUsabilityInitiativeResourceMode='raw'; in LocalSettings.php ? [19:26:16] ok [19:26:19] where are we at here? [19:26:26] 22261? 22281? [19:27:06] RoanKattouw: no, i'm already in that [19:28:41] Interesting [19:28:47] I'm getting this bug on prototype but not on localhost [19:28:55] Although admittedly Firebug is kind of exploding on me now [19:29:01] Refusing to show the Script tab and all [19:29:05] I'll restart Firefox later [19:29:35] TrevorParscal: 22261 seems fine to me [19:30:11] 22281 is still problematic though [19:32:03] adam_miller: I will try and recreate 22261 and then resolve it [19:32:11] ? [19:32:55] pretty sure you wont be able to [19:33:05] i converted it from a table to divs yesterday [19:33:30] the input fields are being rendered on their own line [19:34:59] maybe prototype wasn't updated my change yesterday? or calcey was viewing a cached version of the page [19:35:37] That's what I suggested [19:36:54] so, 22281? [19:37:07] is this really a High importance Major bug? [19:37:24] I want to fix it, but it's not that severe in my opinion [19:37:45] yeah it works, but it's a little unpredictable [19:37:49] Which browsers is it even still broken in? [19:38:19] i'm seeing inconsistent behavior in all of them, but maybe I just don't understand how it's intended to work [19:38:41] OK so what you do is you clear the edit box (Ctrl+A Backspace), then click the list button repetitively [19:38:49] That should insert "*Foo \n* Foo\n* Foo" etc. [19:38:55] well, in chrome it adds to much extra space after a tool with ownLine = true [19:38:58] As opposed to not inserting the newlines or inserting them twice [19:39:14] Yeah ownline seems to be broken in Chrome [19:39:20] I fixed it for Firefox earlier [19:39:31] or rather, the cursor ends up on the next line rather than the end of the newly inserted line [19:39:38] which causes this effect [19:39:58] firefox is the oposite [19:40:05] It seems to be inserting "\n* Foo\n" [19:40:05] ownline stuff seems to not work [19:40:10] Wait what [19:40:17] It did work for me after I fixed it ^^ [19:40:29] well [19:40:40] it works if you select a portion of text in an existing paragraph [19:40:57] but if you repeatedly click the bullet button, it doesn't make a bullet on each new line [19:41:01] It broke when the cursor was at the very beginning IRIRC [19:41:04] it makes a series of them on one line [19:41:05] What does it do in Firefox? [19:41:07] Ah yes [19:41:11] That's what I *used* to get [19:41:14] But then I fixed it (I think) [19:41:15] in safari, I get "*Foo\n*Foo*Foo*Foo*Foo" [19:41:17] Lemme verify [19:41:20] k [19:41:48] OK WTF [19:41:54] It works at the beginning and end but *not* in the middle [19:41:56] *RoanKattouw debugs [19:42:08] RoanKattouw: I see show/hide left navigation bar on prototypes. please disable it at your next update [19:42:51] Deployment or non-deployment prototypes? [19:42:58] non-deployment [19:43:33] "my team and I" --> go Trevor :) [19:43:55] our toolbar is gone too [19:44:05] RoanKattouw: you like that? [19:44:09] Oh yeah [19:44:11] nkomura: Poking [19:44:26] can I get some lunch? [19:44:30] I will be right back [19:44:33] like 5 mins [19:47:03] nkomura: CollapsibleNav was on because of a typo in the config file, fixed now. I do see the toolbar on en prototype [19:47:55] Interesting [19:48:10] do you see the enhanced toolbar? [19:48:16] In Firefox, when I click the list button repetitvely while the cursor is in the iframe, it inserts them on the same line [19:48:17] i see the old toolbar [19:48:36] But when I click outside the iframe then on the buttons, it inserts them with double newlines [19:48:53] nkomura: I see the new one. Which URL are you looking at? Did you purge cache? [19:49:08] http://prototype.wikimedia.org/en-wp/index.php?title=Main_Page&action=edit [19:49:31] i'm refreshing again [19:49:34] still loading [19:49:57] OK I gotta go for a few seconds to restart Firefox, Firebug is all messed up [19:53:05] hi everyone [19:53:31] u see a toolbar here: http://prototype.wikimedia.org/en-wp/index.php?title=Main_Page&action=edit ?? I don t. [19:54:41] hannes-_-: I don't see it either. No NTOC either [19:54:49] de.prototype is working though [19:54:51] I see it, lemme purge cache [19:57:10] ok, got the new toolbar this time [19:57:15] no NTOC still tho [19:57:47] I get it just fine ^^ [19:58:23] hm, my pref is turned on too [19:58:30] how would that be possible [19:58:41] are you FF3.5? [19:58:48] I'm still FF3 [19:58:59] Yes [19:59:03] I guess that's possible [19:59:14] Any dev amongst us with FF 3.0 ? [19:59:51] Otherwise, maybe a dev should walk over to Naoko's desk, install Firebug for her and use it to figure out the error message [20:00:07] i'm not in the office [20:00:14] i can start FF3.5 [20:00:38] i have ff3.5 [20:00:58] hannes-_-: will you check with FF3.5? [20:01:11] ohhh [20:01:23] what happend to the font in the editor? the code looks different [20:01:54] refering to this prototype: http://prototype.wikimedia.org/d-en [20:01:56] correct? [20:02:44] i was referring to prototype, not deployment [20:03:15] hannes-_-: I saw that, on it [20:03:26] Like I said, I'm messing around with deployment a bit [20:03:51] RoanKattouw: you mean the font? [20:04:37] Yes [20:05:08] puhh so slow..... [20:06:19] Aye [20:06:42] RoanKattouw: the code is in timesroman or something similar, right? so can u change the font easily? [20:06:52] I'm working on it [20:07:03] Ah there we go [20:07:11] Updating some style versions should do [20:07:57] it feels stange to see the code beeing a serif font [20:08:01] OK try now [20:08:07] very interessting [20:08:10] I assure you that wasn't intentional :) [20:08:14] yeah its courier [20:08:23] The file telling it which font to use was being pulled from the wrong place [20:08:39] So the browser was never told to use Courier and just decided on Times itself [20:08:42] no worries - i actually enjoyed it [20:08:43] ;) [20:09:04] but courier works better. lets keep it [20:09:39] by the way: nkomura I see the old toolbar as well, in FF3.5 [20:10:12] hannes-_-:, how about NTOC? do you see it? [20:10:16] RoanKattouw: now the bold and italic icons don t show [20:11:40] I know [20:11:43] Should be fixed now [20:12:34] are you guys talking about prototype or deployment? [20:13:04] deployment [20:13:13] but I am talking about the protoype with u [20:14:17] hannes-_-: RoanKattouw is testing his deployment script on deployment environment, so there is no point in providing feedback right now [20:14:54] Yeah please stop telling me what's broken until further notice :) [20:15:05] nkomura: so now everythin seems to be fine: I see the new tollbar * ntoc [20:15:48] good for you, i still don't have love with NTOC on FF3.5 [20:15:58] doesn't the extended toolbar have a nowiki button? [20:16:03] how weired is that [20:16:29] Platonides: Nope [20:16:37] Platonides: it will incorporate nowiki in the next release later in feb [20:17:00] TrevorParscal: I think your

fix may be causing dirty diffs on Firefox now [20:17:10] It's throwing off EditWarning at least [20:17:21] for a button I actually use... :P [20:17:32] Platonides: i hear you [20:17:49] we got similar feedback from editors [20:18:21] RoanKattouw: grrr [20:18:37] we could make it conditional, but that's kinda scary [20:18:37] nkomura, I think that's because it's long to type [20:18:45] as opposed to pressing a button [20:18:54] while adding '' is faster with the keyboard [20:18:55] Platonides: if u have a good innovative idea on how that icon could look like: tell me! I still can't think of something better than a crossed W [20:20:20] TrevorParscal: Wait nm that's not it [20:20:26] It's entities grr [20:20:31] ah [20:23:49] OK guys prototypes should be mostly stable now, feel free to hit it again and bug me when stuff is wrong [20:24:08] i have no love seeing NTOC on en.wp prototype with FF3.5 or FF3.0 [20:24:28] Weird [20:24:31] It works fine for me [20:24:35] i shift+refresh many times so i'm going to give up now until the next rev [20:24:44] i know, it works for you and hannes-_- [20:25:09] if it is still persistent with next rev, i will file a bug [20:25:36] TrevorParscal: kind reminder to disable syntax highlighting from the preference :-) [20:25:47] Ah yes [20:25:53] Needs to be enabled but invisible [20:26:04] yes, global = true, user = false [20:26:13] Oh of course [20:26:18] Well, no [20:26:26] That'll enable it for literally everyone [20:26:28] We don't want that [20:26:42] Ideally we'd enable it only if the NTOC is enabled [20:26:48] if syntax highlighting is enabled, but the toolbar is not... [20:26:52] hmm [20:26:58] is that really an issue? [20:27:10] it would literally be highlihgting nothing [20:27:25] I'm talking about non-beta users [20:27:28] it would be like, let's mark some stuff, hey there's nothing to mark, nevermind [20:27:37] It'd be highlighting nothing at the expense of loading a lot of JS [20:27:53] the js is combined [20:27:55] they get it anyways [20:29:13] No they don't [20:29:16] Not if they're not opted in [20:29:26] The point is that people that aren't opted in shouldn't get our JS [20:29:26] hmm [20:29:36] But with highlight globally enabled they will [20:29:41] hmm [20:29:45] So ideally we hack something up where highlight is linked to toc [20:29:50] interesting [20:29:53] ok - we can do that [20:29:59] For all I care we hack the highlight module to look at the TOC pref [20:30:04] That'd be easiest [20:30:16] hmm [20:30:18] that's a great [20:30:20] idea [20:30:26] change one pref key [20:30:29] should be easy [20:32:44] Aye [20:32:57] Then we can just do global=>false, user=>false for highlight [20:33:06] right on [20:33:08] Hack this in branches/wmf-deployment/extensions/UsabilityInitiative_alpha please [20:33:31] did you branch that already? [20:36:30] Yes [20:36:35] That branch should be runnable [20:36:42] In fact, it's running the deployment wikis already [20:36:55] And extensions/UsabilityInitiative on deployment is already back to the old version [20:37:02] So my hackery worked and we're now ready for per-wiki rollouts [20:37:21] Except of course for this entities bug that I hope to have a fix for [20:37:43] Works in Firefox, firing up my VM [20:37:46] RoanKattouw: ja prototype does not display new toolbar nor NTOC [20:37:56] Will poke [20:38:12] nkomura: deployment-ja or ja-wp ? [20:38:20] ja-wp [20:41:24] I see it [20:41:34] Shift+Refreshing now [20:41:57] i restarted the browser after you said the prototype was ready... [20:42:27] prototypes are unbearably slow [20:43:33] I know [20:45:23] okay, the page loaded the new toolbar [20:45:29] but NTOC is empty [20:45:42] Weird [20:45:47] It must be some JS error I guess [20:46:02] NTOC box is there, but headers are not loaded [20:48:39] did you try turning it on and off again? [20:48:44] Heh [20:49:12] TrevorParscal: Could you guide Naoko in opening the JS console and looking for JS errors? [20:50:44] TrevorParscal: i haven't shut down my computer yet, but i closed the browser and shift+refresh many times ;) [20:51:27] perhaps you can replicate the problem just by following this link [20:51:42] the article is 219 kb long [20:52:00] http://prototype.wikimedia.org/ja-wp/index.php?title=%E7%AC%AC%E4%BA%8C%E6%AC%A1%E4%B8%96%E7%95%8C%E5%A4%A7%E6%88%A6&action=edit [20:52:12] not a typical length [20:52:19] i'm checking on deployment right now [20:52:40] it works on d-ja [20:52:50] http://prototype.wikimedia.org/d-ja/index.php?title=%E7%AC%AC%E4%BA%8C%E6%AC%A1%E4%B8%96%E7%95%8C%E5%A4%A7%E6%88%A6&action=edit [20:54:25] nkomura: d-ja is running the exact same code I intend on deploying [20:54:35] k [20:54:49] Updated that code just now, Trevor's most recent and my most recent commit are now included [20:54:59] ja-wp article just loaded headers in NTOC after another refresh [20:55:04] Hm [20:55:06] Very weird [20:55:22] Seems intermittent and I haven't been able to reproduce it once [20:55:40] Could also be a connection issue with prototype, I'm thinking maybe their bandwidth isn't great [20:55:44] Either way [20:55:50] it may be due to the slowness [20:55:58] Exactly, stuff could get cut off [20:55:59] yeah [20:56:13] can't wait to get out of linode [20:56:25] I'm all set to deploy the code now running d-en (and d-ja etc.) to any list of Wikimedia wikis [20:58:05] Give me the green light and I'll have it rolled out within 15 mins [20:58:39] our first step is to go to test.wikipedia.org and monitor for a while [20:59:14] OK [20:59:16] Wanna do that now? [20:59:28] I do not know why at first glance prototype would be slow [20:59:33] the system has zero load. [20:59:40] sorry, up to 0.4 now ;P [20:59:41] I was thinking maybe bandwidth? [20:59:54] Or generally the fact that it's virtualized? [20:59:59] they could be throttling our bandwidth, streaming backups put us about 300% over allocation. [21:00:03] but i dont think so [21:00:43] hm, so it is our software that is slow? [21:00:46] I will email them and ask if there are throttling measures in place or if we can just pay for excess. [21:00:48] I have no idea. [21:00:52] let me do some speed tests. [21:00:57] Sure [21:00:59] thanks for checking [21:01:18] nkomura: No it's definitely not our software that's slow, unless by our software you mean MediaWiki core :) [21:01:33] ;) [21:01:40] *TrevorParscal takes no responsibility for core performance [21:01:58] TrevorParscal: can you assess the readiness for deploying the code to test.wikipedia.org? [21:02:16] RoanKattouw: are prototypes and deployment up-to-date? [21:02:22] Deployment is [21:02:37] Not sure about prototypes, but at this point I don't care too much about them TBH [21:03:37] it'd be good to be synched as we still use it as test beds [21:05:00] i'm just reviewing this patch [21:05:15] its throttled. [21:05:22] i am pushing files to it, and to another linode [21:05:29] and prototype goes at a third of the speed. [21:05:36] I am going to email them right now to open a ticket. [21:05:57] great find, RobH! [21:06:24] Thanks dude [21:07:10] atleast I assume its throttled, it has free memory and doesnt look like swap death [21:07:19] disk IO is too low. [21:07:28] plus swap death usually means the box is busy =] [21:07:40] Yeah it'll just not be reachable [21:07:53] nkomura: Update is running on the normal prototypes now, takes a while [21:08:03] thanks RoanKattouw [21:15:46] OK guys so do we want to deploy to testwiki now? [21:16:09] If I have to wait longer that's cool but I'll be poking at performance in the meantime then [21:16:16] i don't think TrevorParscal is ready [21:16:50] i'm going through some SVN nightmares getting a deployment candidate locally running so I can verify some of this stuff [21:16:55] almost done I think [21:16:56] sorry [21:17:15] Just checkout branches/wmf-deployment and run extensions/UsabilityInitiative_alpha [21:17:16] no worries, it is better to be thorough than sorry [21:17:27] The deployment prototypes also run that exact code (although they are slow) [21:17:39] TrevorParscal: Also after all this is over we should have our Skype call :) [21:18:24] yes [21:18:41] i'm going to grab some food [21:18:45] k [21:18:49] I will meet with Roan [21:18:58] and when you are back from eating we will push to test [21:19:03] OK that's cool [21:19:19] i can wait if you guys are ready [21:20:06] i'll be back in 30-40 minutes [21:24:19] TrevorParscal: I'm all set for Skyping [21:24:23] k [21:24:30] let me grab a conference room [21:24:40] one sec - letting a checkout finish before unplugging [21:28:38] can you give me the settings we're using for deployment? [21:28:52] like, the lines in LocalSettings.php that are particular to our stuff [21:29:15] Sure just a sec [21:29:20] pastebin it [21:30:13] Well first off, include everything from UsabilityInitiative_alpha [21:30:20] yes [21:30:25] I've got a nice trick for that: $UsabilityInitiative = defined( 'DEPLOYMENT' ) ? 'UsabilityInitiative_alpha' : 'UsabilityInitiative'; [21:30:43] Then :%s/\/UsabilityInitiative/\/\$UsabilityInitiative/g [21:31:03] maybe I will do this over a skype call [21:31:06] one sec [21:31:08] finiding a room [21:32:12] http://pastebin.com/m24d97bcf is what we're using, not sure we need all the DefaultUserOptions stuff [21:32:17] Actually we don't need that on Wikipedia [22:53:53] TrevorParscal and RoanKattouw, are you ready for test deployment yet? :-) [22:54:37] We're still on Skype [22:55:18] OK [23:07:23] nkomura: where are you at? [23:08:55] waiting for your patch review to be done [23:09:15] and if it is, we are good to go for test.wikipedia.org [23:10:38] TrevorParscal: I tracked down that JS error [23:11:22] do we need to do another update to prototypes and deployment? [23:11:32] We will, I'm still poking at something [23:12:12] i also filed this case about link status [23:12:24] https://bugzilla.wikimedia.org/show_bug.cgi?id=22291 [23:13:33] Hm [23:13:41] *werdna waves [23:13:43] it is not a show stopper, but we need some measure for informing a user when a wiki link is switched to external link [23:13:49] hi werdna [23:13:49] ooh, deployments [23:13:51] how exciting [23:13:53] I think Trevor asked Adam to make it so that it always shows the external link icon when the extlink radio button is selected [23:13:55] Morning werdna [23:14:27] i think RoanKattouw and adam_miller were still discussing AFAIK [23:14:42] it is not a stopper for today's test deployment [23:15:06] I'm working on that guy AS WE SPEAK [23:20:07] I just optimized the TOC allot [23:20:41] Yeah [23:20:48] I'll svn up and profile [23:20:54] Roan and I figured if we build hashes of the outline really early on and did basic string comparison to only update the TOC further if something actually changed, it might speed things up a bit [23:20:58] Don't you love it when performance optimization is that easy? :) [23:21:09] so far, it's noticable more responsive on large articles [23:21:23]