[02:14:39] hi mdale [17:58:52] nkomura: I assume we're gonna wait for Trevor a few minutes? [18:00:27] Either way, please poke me when you're starting the meeting. I'm trying to charge my phone as much as possible so it'll stay alive for an hour [18:00:29] yeah [18:02:49] what is the meeting and should i lurk? [18:04:49] It's our weekly usability team meeting [18:05:18] ah [18:05:43] roan? the user accounts dont work on prototype, correct? [18:05:52] so I do have to create an new account? [18:06:41] Yes, user accounts don't carry over from Wikipedia [18:06:58] cause all my user accounts are already in use [18:07:03] but my pw dont work [18:07:05] strange [18:07:23] adam_miller_away: Out of curiosity, how do the dialogs (especially the table dialog cause it's nonresizable) hold up in environments with long messages (e.g. &uselang=de )? [18:07:33] hannes-_-: Would you like me to purge some of those accounts? [18:08:26] RoanKattouw, i believe you have been reviewing a bunch of the Usability code, you think it is ok for you to mark that reviewed in CodeReview? [18:08:31] nevermind - i just created a new account [18:08:45] pdhanda: Well if Tim says so... [18:09:29] (For clarity: yes, I do mark revisions I've reviewed as OK in CR because Tim pretty much said I could) [18:09:48] yeah i chatted with him the other day and he said he was happy with you reviewing, if you have something specific you want him to look at we can make that happen tonight [18:09:49] More specifically, usability code is supposed to be cross-reviewed by ourselves and Tim would only do whole-file reviews [18:10:03] yeah makes sense [18:10:11] Tonight as in what time? [18:10:33] not sure, when he starts working [18:10:40] If you guys could do like a snapshot review of the UsabilityInitiative dir (minus a few files that I'll specify later) that'd be great [18:10:52] I have a few (three) trunk changes that need review too [18:10:56] works for me [18:11:23] With that snapshot review done, only revs after tonight would have to be reviewed individually [18:11:55] Hm it's 5:11 in Sydney now, so presumably Tim starts work in about 4 to 5 hours [18:12:20] That's 2-3 pm PT [18:13:00] yeah he usually comes online around 3pm Pacific [18:13:48] Cool [18:14:01] That's midnight my time but I was probably gonna work into the night anyway [18:14:18] Roan have you been marking things that you reviewed as ok? [18:14:18] do u ever sleep? [18:14:21] hehe [18:14:52] hey pdhanda: You are welcome to join our usability team meeting [18:15:04] Yes, I do sleep, as guillom remarked last night :) [18:15:05] we are going to start 10:30am instead of 10 today [18:15:11] h??h?? [18:15:24] But whenever I don't have to get up early in the morning, I take advantage of that and work Pacific office hours [18:15:49] if you are going to talk about the upcoming UsabilityInitiative push then i would love to [18:15:53] 5 pm PT is 2 am CET so that's not too bad [18:15:54] nkomura [18:16:03] *RoanKattouw forwards the agenda to pdhanda [18:16:09] thanks :) [18:16:40] i'm going to set-up the projector and etc in the mtng room [18:16:56] talk to you guys in 15 minutes [18:38:52] http://prototype.wikimedia.org/deployment-en [18:43:24] *Roan-meeting makes a mental note to tell hannes-_- about the uselang= trick [19:16:55] OK hannes-_- so do you know about the uselang trick? You can use it to test languages we don't have prototypes for [19:34:45] RoanKattouw: no - how does that work? [19:35:17] hannes-_-: Basically you append &uselang=es to the URL and you have a Spanish page [19:35:25] Or ?uselang=es if the URL doesn't contain a ? [19:35:52] ok [19:35:55] hmm I get a fatalerror on nl-prototype [19:36:06] when I click on random article [19:36:30] Text of fatal error? [19:37:34] hmm I get a fatalerror on nl-prototype [19:37:40] ups [19:37:51] Fatal error: Call to undefined method OutputPage::includeJQuery() in /srv/org/wikimedia/prototype/extensions/UsabilityInitiative/UsabilityInitiative.hooks.php on line 218 [19:37:58] Gah [19:38:06] Oh I know why that is [19:38:08] only on the nl prototype [19:38:17] Ryan_Lane added new prototypes but didn't update the scap script [19:39:19] Oh wait he did update it [19:39:59] Lemme run scap.sh and see what's going wrong for nlwiki, I vaguely remember it throwing an error last week [19:41:42] RoanKattouw: do u know which letters stand for bold and italique in polish? [19:41:53] No idea [19:42:00] I'll ask Tomasz [19:43:40] Aah, it was me that forgot to /remove/ pt from the list, that caused the breakage [19:54:57] RoanKattouw: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/61507 [19:55:23] Yeah I saw that [19:55:29] are you planning on investigating, or should we just accept that IE is unstable [19:55:44] So you actually encountered cases where ca1.parentNode was an object that had .firstChild undefined or null? [19:55:51] yes [19:55:54] Do you realize just how absurd that is? [19:55:57] yes [19:56:03] it took me a while to accept it myself [19:56:18] and it was only on IE8 [19:56:20] i think... [19:56:26] wow. Had the original firstChild been deleted or removed? [19:56:28] or moved? [19:56:35] Nope, nothign [19:56:38] or is it just retardation [19:56:47] the latter afaik [19:56:50] All that happens is it does a DOM traversal with repeated .parentNode accesses [19:57:01] Then immediately after it calls .firstChild on the result [19:57:14] what kind of node? [19:57:26] All kinds, usually textnodes though [19:57:29] ah [19:58:05] Oh wait no, parents of textnodes [19:58:19] Textnodes never have children so they can't be reached using an upward traversal like that [19:58:27] right [19:58:58] on the other hand, textnodes are often merged, and the number of textnodes inside an element is sort of arbitrary from parse to parse [19:59:17] that's still some impressive brain damage though. [19:59:29] There's usually one per element or between elements in Firefox [19:59:33] right [19:59:42] IE usually does the same except it puts leading whitespace in its own textnode [19:59:46] exactly [19:59:48] But otherwise the number of textnodes is minimized [19:59:51] hey hannes-_- check out what i just stumbled upon: http://bs.wikipedia.org/wiki/Po%C4%8Detna_strana [20:00:02] Until of course we start splitting them up ourselves so we can wrap parts of them [20:00:16] hi parutron [20:00:22] bs.wikipedia.org ? :P I wonder how many people take that seriously [20:00:26] rounded corner style? [20:00:40] just musing about why it could be, lack of .firstChild is still a bug [20:00:43] rounded corner, different font, spacing, etc [20:00:48] i clicked on it by accident [20:00:55] it's like a monobook bandaid [20:01:04] beware, CSS3 rounded corners do not work on most browsers [20:01:25] It's mostly the main page design though, and that's skin-agnostic for the most part [20:01:39] oh tee hee [20:01:43] i didn't click past the mainpage [20:03:15] Most Credible Wikipedia Subdomain Ever [20:03:24] hmm I am checking the language specific icons and it looks like some are wrong. who can change that? roan you? [20:03:51] afaik, trevor, roan, or adam [20:03:57] nimish is too cool for icons. [20:04:26] Yes they're my baby [20:04:30] Which are wrong and why? [20:04:46] also, what's up with the dropdown being on the left tabs here: http://prototype.wikimedia.org/zh-wp/index.php?title=%E9%A6%96%E9%A1%B5&variant=zh [20:05:05] (i.e. with the page types "Article" and "Disucussion") [20:05:13] That's the variant dropdown [20:05:29] what's that, i've never heard of it [20:05:53] RoanKattouw: I ll send u a mail about that [20:06:04] We do auto-transliteration between certain variants (alphabets) of Chinese and some other languages like Serbian (Latin and Cyrillic variant) [20:06:17] Try srwiki to get a better idea :) [20:06:30] will do [20:07:24] Meh sr prototype is broken, fixing [20:13:20] i just checked out the font size for the character-language dialogues (the ones that came up from our beta survey feedback) [20:13:45] looks ok to me, but i noticed, curiously the characters in the editor are quite smaller than any of our skin/layout elements [20:13:51] i wonder how users feel about that. [20:16:20] OK sr prototype is fixed now [20:17:52] parutron: Go to http://prototype.wikimedia.org/sr-wp/index.php?title=%D0%93%D0%BB%D0%B0%D0%B2%D0%BD%D0%B0_%D1%81%D1%82%D1%80%D0%B0%D0%BD%D0%B0&variant=sr and that dropdown you mentioned will show "latinica" as a second alternative, pointing to the Latin transliteration of Serbian [20:20:12] hmmm, am i missing something? [20:20:26] what| [20:20:28] ? [20:20:31] it's a language variant [20:20:41] what is the variant accomplishing? [20:20:59] It turns all the Cyrillic gibberish into Latin equivalents using (mostly) our alphabet [20:21:04] OHHHHHHH [20:21:24] Both variants (Latin and Cyrillic) are common in Serbian, and Chinese has like half a dozen of them [20:22:06] The important thing to see here is that the article content changes too; this is because there's easy recipes for transliterating back and forth so no manual translation is needed [20:22:43] yeah. [20:22:47] how was this accomplished before? [20:22:54] In Monobook you mean? [20:22:56] yeah [20:23:24] http://sr.wikipedia.org/wiki/%D0%93%D0%BB%D0%B0%D0%B2%D0%BD%D0%B0_%D1%81%D1%82%D1%80%D0%B0%D0%BD%D0%B0 --> rightmost pair of tabs, one for Cyrillic one for Latin [20:23:30] i see [20:23:45] tee hee, we were looking up the same site simultaneously [20:34:33] :/ [20:35:17] for once i think about participating in IRC office hours, and it's planned in the one week a year i definetly do NOT have internet :D [20:36:11] parutron: are u there? [20:37:03] i am now [20:37:07] what's up? [20:37:10] Heh [20:38:08] nkomura: good blogpost. thank you ! [20:38:24] parutron: aim ? [20:38:45] sure thing [20:38:47] one sec [20:42:20] nkomura: Are these multimedia office hours the same thing as Cary's staff office hours thing? [20:42:32] Or are they recurring events like the strategy office hours? [20:42:33] yup [20:42:41] *RoanKattouw complained about the term confusion in the other channel [20:42:58] it's WMF office hours [20:43:19] i believe the strategy team has their own recurring schedule [20:43:39] hannes-_-: Instead of switching pl to B/I I'm gonna leave it alone for now so it follows the default. The default is currently A/A but we may wanna change that to B/I [20:43:41] Yeah they do [20:44:04] I was confused earlier today, thought the strategy office hours being announced were the WMF staff thing [20:45:11] RoanKattouw and hannes-_-, parutron and i are talking about switching the default to b/i instead of A/A [20:45:21] Rigth [20:45:27] we should make a decision soon [20:45:31] siebrand never really weighed in [20:45:50] i think it's better at this point to make a choice as a team. [20:45:53] Hannes asked some Polish people and they said they all use B/I , so I first said I'd change Polish to B/I, then realized this is true for a lot of other languages (incl. Dutch) as well [20:46:00] Yeah [20:46:04] i've given tons of thought to this, and i think they are both good solutions. [20:46:12] if i'm leaning anyway, it's to B + I [20:46:16] I'm for it per my previous comment [20:46:21] (for B/I that is) [20:46:34] hannes-_-? do you concur? [20:46:47] A/A is nice and neutral but the fact is that B/I is more widespread and probably clearer [20:47:24] yeah, the rich B/I will get richer, but i'm cool with that. [20:47:50] open source apps do tend to use A/A but they cannot afford the language variations we are providing.... [20:48:26] B/I: 2 A/A: 0 [20:48:36] hannes-_-, parutron: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/61536 [20:48:43] Pushing to prototype in a sec [20:49:12] nkomura: is any reasoning pushing you towards one over the other? [20:49:26] polish ms word and openoffice seems to use B and I [20:50:11] (relooking up "bold" and "italic" in spanish and italian) [20:50:13] tee hee [20:50:17] Same for Dutch Word [20:50:23] parutron: i did [20:50:37] and checked with people in es-IRC [20:50:44] oh great. [20:50:56] i put my big list somewhere, i suppose i'll put it on our language how-to somewhere. [20:51:09] Put what on your language how-to? [20:51:21] hannes-_-: you and i should look at that since we haven't looked at it in a while. [20:51:27] parutron: japan, china, and korea are MS dominated country and people are used to see b for bold i for italics [20:51:35] but i can't speak for other languages [20:52:05] seems we have past use/experience in the east and west. [20:52:16] B and I it is. [20:52:51] parutron: will you can send your recommendation to the team alias? [20:53:06] i'll write to siebrand again and see if he has any major concerns we haven't thought of ourselves, but i feel good in our decision. and the A and A are available if a language wiki does not feel so english-centric. [20:53:16] will do. [20:53:49] k hannes-_- my prototype has finally loaded. will meet you back on gtalk [20:55:24] parutron: ok [20:57:34] TrevorParscal: Did you read my e-mail with the subject "Skype call" (a few days old)? Could you reply (or react here)? [21:07:32] parutron: i don't find Korean fonts in editor that small though [21:07:39] on ko.prototype [21:08:30] Also guys, I've noticed that certain languages don't have deployment prototypes and I suspect ko may be one of them. I'll get that set up [21:09:44] ko is already available [21:10:01] ah you meant deployment [21:10:26] Yes [21:10:59] *RoanKattouw shakes fist at prototype for dropping his SSH connections after relatively short idle times [21:11:54] easy there [21:12:45] *nkomura has to reboot her computer brb [21:23:15] RoanKattouw: what is the current difference between prototypes and deployments code wise? [21:23:56] UsabilityInitiative extension is the same, MW core is different (latest trunk vs. wmf-deployment branch) [21:24:08] wmf-deployment is the code we run on production [21:25:15] got it [21:32:15] has the font size inside the editor been changed? [21:34:01] nkomura: the vmserver is in [21:34:03] I am going to rack it, but it may not get its IP and port assignments until tomorrow [21:34:23] RobH: great news!!! [21:34:36] Ryan_Lane: are you there? [21:34:51] the QA automation/virtual server has arrived [21:35:11] i just sent you all an email [21:35:15] i need to know what you want the hostname to be [21:35:19] otherwise it should be online for you tomorrow [21:36:16] i'll put my 100% brain power to come up with a good name ;) [21:36:44] do you have suggestions though? [21:36:56] nice [21:38:04] *nkomura is accepting a host name recommendations for QA/virtual server [21:38:25] RobH: is globe taken? [21:38:39] globe? [21:38:46] is that a person? its not used. [21:38:51] as a host name [21:38:54] it is not a person's name [21:39:01] Hosts are generally named after people AFAIK [21:39:03] does it has be a person's name? [21:39:06] k [21:39:13] At least the more modern ones are [21:39:16] RoanKattouw: no love with NTOC in en-deployment [21:39:30] en.prototype either... [21:39:34] Looking [21:39:45] We now have deployment prototypes for es, ko, nl and zh BTW [21:40:30] great [21:41:18] RoanKattouw: my bad on en-deployment [21:41:26] i didn't check my preferences [21:41:30] i need to turn it on [21:41:38] Right [21:41:42] but NTOC for en.prototype returns blank [21:41:48] *RoanKattouw notices that parsing is slow for long articles [21:42:07] WFM on deployment-en [[San Francisco]] [21:42:12] nkomura: I explained host naming in the email I sent you =] [21:42:22] anyhow, it wont be online until tomorrow [21:42:24] thanks [21:42:33] while I am racking it now, I have been in here awhile and I am tired of the DC [21:42:34] heh [21:43:16] it must be tiring there [21:43:26] when do you need the host name by? [21:43:39] tomorrow =] [21:43:43] it cannot be put online without it [21:43:48] we can do that ;) [21:43:54] (I can make one up of course, but I figured i would let you all do it) [21:44:08] mostly though I am tired of looking up usability figures and determining if they are still alive [21:44:13] its a bit morbid ;] [21:44:22] nkomura can you forward me the host naming email please [21:44:31] no worries, we can come up with a name today [21:44:51] pdhanda: tomorrow I plan on reinstalling srv89 as a new host for you to use for project manangement software testing [21:45:01] I meant to do it today, but did not get to it =P [21:45:36] np, I'm busy with stuff today anyway [21:50:11] awesome [21:52:10] Holy crap [21:52:21] pdhanda: I'm profiling stuff now and it does seem to be slow on large pages [21:52:35] Parsing [[San Francisco]] (160 KB or so) for headers takes like 4 s [21:58:49] I'm not seeing the precision of NTOC I was seeing last week [22:01:19] What do you mean? [22:02:21] when i edit short article like this [22:02:46] NTOC does not bring up the header to the top of the editor [22:03:06] but when i edit a longer article [22:03:19] the jumping works more precisely [22:03:44] Link to a short article? [22:04:02] http://prototype.wikimedia.org/en.wikipedia.org/2009_Yambol_bus_crash [22:04:15] haven't tested on deployment yet [22:05:47] it behaves the same for deployment [22:07:32] *TrevorParscal waits a few minutes for the page to load [22:07:35] longer article likes the following one works good [22:07:36] http://prototype.wikimedia.org/d-en/index.php?title=Manchester_Small-Scale_Experimental_Machine&action=edit [22:07:55] *TrevorParscal starts waiting all over again [22:08:21] neither of those articles are working for me in safari [22:08:28] really? [22:08:32] something broke [22:08:39] because it's not workign locally anymore either [22:09:08] every click in safari sends it to the top of the article, if it's not already there [22:09:28] hmmm [22:09:34] what's been changing here? [22:10:12] something's gone crazy [22:10:17] I've been profiling stuff, turns out that to reparse [[San Francisco]] we call $() like 6000 times [22:10:19] the TOC is gone in IE [22:10:31] oooh [22:10:42] it's not too bad with FF [22:10:53] just the precision for short articles degraded [22:10:56] What you mean the slowness? It is bad if the article is large enough [22:11:20] i am only talking about the precision here [22:11:33] nkomura: Works perfectly for me. The last two/three sections can't be scrolled to the top because there's less than a screenful of text below them [22:11:36] precision on short articles? [22:11:45] yes, what he said [22:11:46] The cursor is still moved to the right place though [22:11:59] definitely [22:12:09] I'm getting allot of invalid argument in IE, especially on resizing [22:12:34] did our 'NaNpx' trick get lost? [22:13:00] Did you hack jQuery to get that done? Then yes, maybe [22:13:14] I'll look through the log for js2stopgap.js [22:13:17] epse is going crazy! [22:13:21] *eclipse [22:13:23] TrevorParscal: i noticed if there are multiple headers are visible in the editor, jumping from NTOC does not bring the selected header to the top [22:13:24] RoanKattouw: yes [22:13:33] but it sounds like you guys are dealing with a bigger issue [22:13:46] so we can revisit it later [22:13:49] no, it's a simple one [22:13:50] hld [22:13:57] TrevorParscal: Nope, it's still in there [22:14:34] nkomura: That's not fixable because in order to scroll something to the top, there needs to be stuff filling the rest of the box. So headers that are near the bottom can't scroll to the top because that would mean scrolling past the bottom [22:17:35] interesting [22:17:39] RoanKattouw: TrevorParscal came over and explained it to me [22:17:43] so i got it, thanks [22:18:01] it's giving me the invalid argment on the line inside the statement we added the NaNpx test in [22:18:10] RoanKattouw: must be something else [22:18:18] Hm [22:18:19] can you reproduce on your IE8? [22:18:39] line 125 char 1 of js2stopgap.min.js [22:19:10] we are passing bad arguments into a CSS selector, which is probably a bad thing for us to do [22:19:19] but I'm not sure where it's coming from yet [22:21:13] Steps to reproduce? Also, line numbers in a non-minified file are more useful [22:23:31] Whoo [22:23:40] *RoanKattouw is a ninja [22:23:43] ? [22:23:47] what did you do now? [22:23:49] I just cut like 40% off parsing time with one single change [22:24:07] Apparently doing $( domElement ).hasClass( 'foo' ) in a tight loop that's run 5000 times is not very fast [22:24:07] isn't that what we call chance? :P [22:24:22] So I did a manual check on domElement.className [22:24:35] awesome! [22:25:24] That eliminated 5172 calls to the jQuery constructor and made stuff 1521 ms faster [22:25:42] So now parsing [[San Francisco]] takes "only" 2707 ms. That's down from 4228 but still way too much [22:26:09] adam_miller: got a min? [22:26:18] sorta, whats up? [22:26:21] RoanKattouw: you are awesome [22:26:25] adam_miller: on the link dialog [22:26:42] I'll need to be a lot more awesome to get this down to a reasonable number [22:26:52] We may want to consider raising that 250 ms timeout for larger articles [22:26:59] can we make it so that anytime the external link radio button is selected, no matter if it is selected manually or automatically, the status icon shows external link? [22:27:03] Or just put our bets with jQuery 1.4 :) [22:27:22] RoanKattouw: timeout as function of article size on load [22:27:26] ? [22:27:37] it's an idea [22:27:51] Yeah [22:27:54] With an upper bound though [22:27:59] sure [22:28:07] I do that sort of thing with the animation of the toolbar [22:28:12] the speed is based on the distance [22:28:22] so there is a stable pixels per second animation speed [22:28:42] we could base delayedBind on the length of the article [22:29:05] short articles are near realtime, long ones only update after 1 sec of idle [22:29:17] 250ms - 1000ms is a goo range [22:29:20] *good [22:29:29] I will do that [22:30:29] TrevorParscal: I don't follow 100%. So say i type 'Boeing 787' into the Target page field, and then select the external page radio button. You want it to show the external link icon? [22:32:17] *nkomura is enjoying selecting text [22:32:18] yes [22:32:35] because the link will be made an external link or not based on the radio buttons [22:32:41] so, we need to refelct that [22:32:43] should be simple [22:33:44] OMG wait a second [22:33:54] highlight is still tokenizing {{ and }} [22:33:58] THAT's what's taking so long [22:34:23] I need to change it so it only picks up tokens from enabled modules [22:37:13] And that' [22:37:20] And that's another 58% off :) [22:39:33] YO i know you guys are into some other issues, but safari (and presumably other webkit browser) are consistently returning 0 for $element.offset().top in scrollToTop, which is breaking the ntoc [22:40:47] RoanKattouw, or TrevorParscal got any ideas why? [22:41:28] Don't look at me [22:41:35] I assume it's doing this, even when the offset().top isn't actually 0 [22:41:54] yeah it's doing it for every heading [22:42:05] RoanKattouw: will u have a look at what's going on with arabic prototype when you have a moment [22:42:14] What about it? [22:42:16] it is giving me jQuery fatal error [22:42:22] when i opt-in [22:42:31] it gives me an fatal error [22:42:41] Fatal error: Call to undefined method OutputPage::includeJQuery() in /srv/org/wikimedia/prototype/extensions/UsabilityInitiative/UsabilityInitiative.hooks.php on line 218 [22:43:36] Ah ar-wp locked itself up, so it didn't get updated [22:43:41] the core code is updated right? [22:43:41] On it [22:43:42] aah [22:44:36] TrevorParscal: were the marker div's always empty? or were we placing the heading text inside the marker divs previously? [22:45:24] they weren't before, but they are now [22:45:34] i'm pretty sure that's the problem [22:45:39] it helped us solve this issue of the cursor being moved outside of the heading upon parsing [22:45:58] it's very important that they aren't changed back to surrounding the heading [22:46:02] safari isn't positioning them [22:46:14] so, we need a work-around [22:46:28] can we put something inside them? [22:46:43] RoanKattouw: adam is thinking his issue has to do with the marker div's being empty [22:46:53] we don't want any visible characters.... [22:46:59] Interesting [22:47:00] you could put an empty span [22:47:02] ? [22:47:15] Well can we put random crap in them and apply display:none; ? [22:47:49] it'd still get included when they save tho, right? [22:48:02]