[00:02:15] OK I just unbroke test [00:03:13] why does test not seem to have our features? [00:03:31] config time? [00:03:46] No, JS erro [00:03:48] r [00:03:50] hmm [00:03:57] even the prefs look wrong thoug [00:04:01] Code missing in wikibits.js , I need to sync [00:04:07] Yeah that's missing messages, will handle that too [00:04:21] but yeah, the ls errror... [00:04:22] hmm [00:04:26] mw is not defined [00:05:32] Yeah I need to sync wikibits.js [00:06:32] OK hard refresh now [00:06:51] Bleh I need to sync more files [00:28:47] OK so now I need to run scap, and all the ops folks are away or offline so I can't ask anyone if it's OK [00:29:33] TrevorParscal: Are Ariel and Fred at their desks? [00:39:58] no [00:40:00] neither [00:41:20] http://bits.wikimedia.org/skins-1.5/common/http://bits.wikimedia.org/w/extensions/UsabilityInitiative_alpha/WikiEditor/WikiEditor.combined.min.js?urid=257z3 [00:41:23] WTF [00:45:27] what the heck? [00:45:43] Fixing [00:45:49] OutputPage::addScript() doesn't like URLs [00:51:11] *werdna looks for RoanKattouw [00:51:13] oh there you are [00:54:50] I've got a request for a change to a bunch of WM skins, including vector [00:55:06] What's that? [00:55:08] and I figure if I can get you all to do it to fix vector, it will be faster [00:55:16] the Contact us link at the bottom of articles [00:55:29] it is offset a line because of the way that it is hacked in [00:55:43] hacked in? [00:55:50] basically [00:55:58] oooh [00:56:02] a Mediawiki message ends a list that it isn't supposed to [00:56:04] yeah, that's not how it's supposed to be [00:56:05] then starts another one [00:56:27] it should have been added to the UL that's classed footer-places [00:56:27] that works in monobook, but the vector (and everything else) structure is different [00:56:39] the problem is there is no way to add text to the footer-places area [00:56:41] in any skin [00:56:58] on the server or the client? [00:57:08] Contact us is added via mediawiki message, unlike all the other links [00:57:33] what's the message? [00:58:01] Wikimedia-copyright [00:58:58] and you added a "

my stuff here" or something? [00:59:02] that the hacking you mean? [00:59:10] it has been there a long time [00:59:22] its just that no one noticed it because everyone uses monobook [00:59:33]

Contact us [00:59:55] that works in the monobook HTML, but because of the seperate div in vector, it doesn't there [01:00:05] http://en.wikipedia.org/wiki/MediaWiki:Wikimedia-copyright [01:00:19] yeah, so that's really not ideal how that's being done to be honest [01:00:55] so, what are you trying to get done? [01:01:03] add the contact us to the places links, right? [01:01:17] right, so what I'd propose to do to fix it is to add a new mediawiki message, along the line of "$1 $2 $3" for the footer [01:01:24] so we can edit it [01:01:42] (where $1, $2, $3 call the three existing links) [01:02:01] I don't think forcing a contact us link in for all mediawiki sites would be a good idea [01:02:28] no, but it could be optionally shown [01:03:45] true but I imagine a lot of mediawiki sites would like the ability to edit the footer [01:03:52] any solution getting contact us on that list works for me though [01:05:55] basically, there are allot of links that could be in the footer [01:06:00] only those that are being used are shown [01:06:22] adding a contact link to the footer can be done in the skin, and it would only show up if it's being used [01:06:41] honestly I think the footer is a mess, so whatever works for you [01:07:48] I think the entire skin system is horrible [01:08:03] but I have not yet been given the task to fix it [01:08:15] and I assure you when I do get that task, it will be great [01:09:56] in the mean time we could add a contact item to the skins, and then you could just use a mediawiki message and it would be properly integrated [01:10:07] software change needed, yes, but it will be done right [01:10:29] much the way that the pricacy link is there because there's a http://en.wikipedia.org/wiki/MediaWiki:Privacy message [01:10:37] you could also do so with a contact message [01:10:52] it just needs a few lines to be added to some skins [01:11:09] line 424 of Vector.php [01:11:12] for instance [01:13:07] RoanKattouw: did you enable the syntax highlighting thing the way we were saying [01:13:14] allong with the TOC [01:13:41] TrevorParscal: monobook is the most important one [01:14:03] Prodego: well, there's allot of people on Vector too [01:14:03] TrevorParscal: Still poking, almost done [01:14:35] TrevorParscal: I've been trying to decide if its worth getting someone to redo the massive js overhead I have to customize vector [01:14:40] instead of using monobook [01:14:42] over 182,000 [01:14:50] context.data.outline is undefined [01:14:52] http://bits.wikimedia.org/w/extensions/UsabilityInitiative_alpha/js/plugins.combined.min.js?urid=257uxz3 [01:14:53] Line 620 [01:15:01] *RoanKattouw wonders why [01:15:07] Like I don't need a 'read' tab [01:15:32] Prodego: Can't you hide it with CSS? [01:15:35] Prodego: you can stick to monobook, or use a user style to hide that tab [01:15:46] Our user research shows that that tab helps people [01:15:58] *RoanKattouw points Prodego to the topic [01:16:00] RoanKattouw: well /I/ can't, but once I find someone who can do it for me :) [01:16:19] Oh I'm sure its helpful if people don't know what they are doing yet [01:16:21] And the people, like you, who might wish it were not there tend to be the same people who can hide it themselves using a user CSS message [01:16:33] Oooh wait a second [01:17:11] RoanKattouw: I have a rather large amount of cusomization so it'll be a serious project at some point [01:17:23] but I do like how vector looks, so at some point I'll probably do it [01:17:44] so please, http://en.wikipedia.org/wiki/User:Prodego/Vector.css [01:18:10] #ca-view { display: none } [01:18:13] and you're all set [01:18:31] well no, because there are a million other things I'll have to change :) but that's a start, thanks [01:18:54] Prodego: s/have/want/ [01:18:57] :) [01:19:13] that's the joy of user styles and scripts! [01:19:19] TrevorParscal: In full glory: http://test.wikipedia.org/w/index.php?title=Main_Page&action=edit [01:19:20] and soon, you will have jquery on every page [01:19:27] so, it will be SUPER easy to do magic [01:19:32] And I was planning to go to sleep around midnight, now it's 2:19 am :( [01:19:37] did you do research on the star? [01:19:46] its rather unintuitive [01:20:07] The star icon is very common for favorites systems [01:20:22] the concept of a watchlist is very foreign to new users [01:20:29] but they understand favorites systems [01:20:34] cf. the star in the address bar in Firefox [01:20:35] which the watchlist is very similar to [01:20:47] *RoanKattouw moves that i'm sure a lot of folks will hate it, but it will certainly make it easier for my mother. be the official motto of our team [01:20:56] so, it helps people understand the basic functionality of that button at a glance [01:21:10] "Add to watchlist" doesn't seem to [01:21:14] perhaps, we'll see about that one [01:21:28] nkomura_mtng: After a painstaking process of nearly 3 hours, we have successfully deployed to test :) [01:21:39] Or 'watch' even [01:21:41] That's what is awesome about user studies, we film people using the stuff, and the post it online [01:21:46] so, yes, we will see indeed! [01:21:50] But the in-person tests we've done are much more basic than that [01:22:03] Like getting people to even see the edit button that the watch button is close to [01:22:15] Vector's design seems to help there [01:22:21] Lets see what else will I need to do, I'll need to move that watch button to a menu, replace the down arrow menu with words, rename all the 'My talk' links, etc to 'talk' 'prefs'... might take a while :) [01:22:37] RoanKattouw: TOC? [01:22:41] vector makes basic functionality easier, but more advanced functions harder [01:22:41] I get a blank area [01:22:52] Prodego: intentionally [01:22:55] TrevorParscal: Hard refresh [01:22:58] and it wasn't designed for admins [01:23:11] it was designed for readers and new contributors [01:23:13] TrevorParscal: so when will you guys be designing that skin? :) [01:23:19] to - as the grant says "lower the barriers to contribution" [01:23:20] Prodego: " i'm sure a lot of folks will hate it, but it will certainly make it easier for my mother." [01:23:32] Prodego: hopefully soon [01:23:35] RoanKattouw: I know, its a good idea, its well done [01:24:09] its hard to put myself in the place of someone who knows nothing to ask what's improved [01:24:55] Prodego: that's exactly why we have to do allot of research, because it's a hard thing to see on your own [01:25:08] TrevorParscal and RoanKattouw, congrats for deploying to test! [01:25:23] Yay [01:25:23] so we put other people behind the software and watch them fail, and then fix stuff, and watch them succeed - or fail and then fix more stuff.. etc. [01:25:27] in basic terms [01:25:36] It took me some time to sort out all kinds of weird issues [01:25:42] RoanKattouw: still no TOC [01:25:51] so the config issues have been resolved? [01:25:55] TrevorParscal: WFM after hard refresh [01:26:05] WFM? [01:26:18] nkomura: Yeah there was the config and there was the hackery I did to make the old and new code coexist [01:26:30] I obviously didn't get that right in one go :) [01:26:34] TrevorParscal: Works for me [01:26:44] hmm [01:27:51] I got that empty TOC before, along with a context.data.outline is undefined error [01:28:02] So I fixed the bug on the server and did a hard refresh [01:28:26] shift+refreshed many times [01:28:27] http://img407.imageshack.us/img407/9011/picture2is.png [01:28:36] JS errors? [01:28:50] wgNavigableTOCCollapseEnable=null, [01:28:50] 75wgNavigableTOCResizable=null, [01:28:52] um [01:28:55] oh vector has HTML errors in it [01:28:58] you forgot the globals [01:29:29] http://validator.w3.org/check?uri=http://en.wikipedia.org/wiki/Special:BlankPage%3Fuseskin%3Dvector [01:29:32] they are supposed to be $wgNavigableTOCCollapseEnable = true; $wgNavigableTOCResizable = true; [01:29:33] I forgot to mention this [01:29:35] in the PHP [01:30:04] Prodego: hmm - yeah we should check that out [01:30:08] Aah [01:30:09] it validated at one time [01:30:22] I wonder if this is core skin stuff or localized hackery [01:31:32] ok [01:31:55] RoanKattouw: strange that it worked for you [01:32:09] i don't like those globals [01:32:14] they need to die soon [01:32:31] Yeah although mine probably wasn't collapsible [01:32:38] And yes those globals must be killed with fire [01:33:04] death fire [01:33:20] I need to get going [01:33:41] Me too :) [01:33:52] my daughter is having a bad day, so my wife is pretty much also having a bad day... (cadence this is) so today is a bad day to stay late [01:34:03] and yeah, it's like 3am over there or something [01:34:11] 2:30, yes [01:34:16] change those globals and call it a night I say [01:34:26] I intended to go to bed around midnight [01:34:31] ha ha [01:34:33] Yeah already did that, works for me now incl. collapsible [01:35:49] k [01:35:51] i'm out [01:35:52] cyall [17:08:08] good morning [17:13:58] Morning [17:14:24] how are you RoanKattouw? [17:14:30] I'm good [17:14:41] did you manage to get to University today? [17:14:49] I just spent an hour and a half on interesting Math Olympiad problems my brother brought home :) [17:14:50] you were up so late last night [17:14:52] Yeah [17:15:06] I couldn't sleep till like 5:15 [17:15:24] =-O [17:15:56] i've been poking at test.wp since yesterday [17:16:08] but NTOC does not load [17:16:19] Weird [17:16:35] It works for me after Shift+Refresh [17:16:54] i've done that many many times [17:18:53] Interesting, I'll investigate that [17:18:59] What's your Firefox version? 3.0, right? [17:19:01] *RoanKattouw --> dinner [17:19:47] right FF 3.0 [17:47:30] hi everyone [17:47:39] nimish_g: hi naoko [17:48:27] I am here - I cancled my meeting at 18:00 cause roads over here are insane at the moment. [17:48:43] hi hannes-_- thank you [17:49:41] howdy [17:56:48] heyyo [17:58:23] hi nimish [17:58:36] how will we meet? conference call? [17:59:15] RoanKattouw_away: adam_miller how's it going? we may want to implement an undo-redo system today, I've got some example code from CodePress of how they do it by saving states and intercepting keys [17:59:47] OK [17:59:51] sounds interesting [17:59:57] where can we look at them? [18:00:08] I'll be poking at the bugs Calcey's been filing last night [18:00:43] we do have a call right now, right? [18:00:55] Yes, that too [18:01:00] *RoanKattouw grabs phone [18:01:01] one strange thing that happens, in FF anyways, is that when you are at the begining of a paragraph tag, and press backspace, it deletes the whole paragraph or does other crazy things [18:01:41] http://codepress.sourceforge.net/index.php [18:01:44] Hm Naoko said it was 6pm-7pm CET but I'm sure she miscalculated, because 10am-11am PT is definitely 7pm-8pm CET [18:02:00] look at one of the engine/[browser].js files [18:02:01] Unless you guys are like in DST already :P [18:02:11] not yet [18:02:45] *TrevorParscal_ has switched to wireless [18:04:03] TrevorParscal_: I can't replicate the strange thing you mentioned a minute ago [18:29:15] http://img168.imageshack.us/img168/6914/picture2ph.png [18:29:24] yeah, no wonder it's not working [18:34:49] and this is on the edit page -> http://img96.imageshack.us/img96/1163/picture1sv.png [18:35:04] so, basically the settings are screwed up somehow [18:53:32] http://en.wikipedia.org/wiki/MediaWiki:Sidebar [18:56:48] http://groups.google.com/group/openweb-group/files [20:08:11] no tollbardiscussion now? [20:09:16] nkomura: are u there? [20:16:01] nimish_g, TrevorParscal : When will you guys be having lunch? I have another phone call to make so I could adapt to making that call during your lunch so we'll have more time to work together [20:16:27] ok, Trevor just left to get food and I'm about to leave in a min, so that would work [20:17:46] Cool [20:47:18] Anyone around who could at one point reproduce the empty TOC bug? I deployed a tweak, could you try if it worked or report any JS errors from the JS console or Firebug otherwise? [20:48:58] RoanKattouw: I have ff3.57 that doesn t help, does it? [20:50:03] nimish_g: RoanKattouw can I talk to one of you about some minor stuff with the link dialogs? [20:50:06] Well could you try? [20:50:19] Go to test.wikipedia.org and see if NTOC works for you [20:50:25] Is this the link image again? [20:50:47] nope - hold on roan I ll checkl the test.wiki first [20:51:23] looks good to me [20:51:57] OK [20:52:13] everything works perfect [20:52:31] ok so about the link dialog [20:52:54] are the translations on the language prototypes final? [20:53:07] Translations are never final [20:53:12] hehe [20:53:18] Even when we deploy things, translations are added every 24 hrs [20:53:25] This applies to all software running on Wikipedia [20:53:26] so maybe we wanna change the polish one [20:53:31] becuase that end with an "-" [20:53:49] *RoanKattouw suggests #mediawiki-i18n for that [20:54:06] ? [20:54:24] That's where you should poke people to change translations on TranslateWiki [20:54:50] ok [20:56:35] oh and than I noticed something about the left nav on test.wiki [20:56:44] can I tell u or shall I use bugzilla? [20:57:15] What's wrong with the left nav on test? [20:57:21] It's not collapsible, on purpose [20:57:52] when I first got there the links were displayed in two different blues [20:58:06] like if I had already visited some [20:58:41] That's not our software's doing [20:58:49] And one of them links to the blog, you may have visited that [20:59:37] strange [21:01:58] enough for today.... good night everyone [21:02:03] Either way, it's not our code and no one cares about test :) [21:02:05] Night [21:29:40] RoanKattouw: still awake? [21:29:46] Yeah [21:29:49] It's only 10:30 [21:31:19] nimish_g: So yeah I was gonna tell you something about the caching changes [21:31:28] yes [21:31:50] Before, we used to serve stuff from http://en.wikipedia.org/w/extensions/UsabilityInitiative/.. and from http://de.wikipedia.org/w/extensions/UsabilityInitiative/... etc. [21:31:58] So that's 812 different URLs for the exact same file [21:32:18] Now we serve everything from http://bits.wikimedia.org/w/extensions/UsabilityInitiative/... [21:32:21] On all wikis [21:32:41] for all extensions? [21:32:51] No just for us and for core skin stuff [21:32:57] ooh interesting [21:33:00] Other extensions can be converted as needed [21:33:12] Because it requires changing their code a bit to use $wgExtensionsAssetsPath [21:33:14] that still causes cache issues [21:33:30] Just not as bad as before [21:33:40] right. ok. got it [21:33:41] Before, enwiki and dewiki were completely independent in terms of JS/CSS cache [21:34:01] You're right we should ideally deploy this at low traffic time [21:34:19] Or turn it on that is [21:34:30] Because new clients will be starting to hit bits like crazy [21:34:49] But the other advantage is that bits runs on a dedicated bunch of Squid/Varnish servers (not sure which) [21:35:08] So at least the impact of CSS/JS avalanches is isolated [21:39:21] adam_miller: There's quite a few issues with the textbox description text in the link dialog being inside the textbox [21:39:47] link? [21:39:53] For instance, the dialog will happily insert [[Page title or URL]] [21:40:13] The descriptions of the two textboxes are the same, which is probably not intended [21:40:39] And the presence of "Page title or URL" in the upper textbox triggers the existence check, which reports "Page does not exist" [21:40:48] And the text in the upper textbox doesn't get greyed out [21:40:58] Tested on trunk, probably works identically on test.wikipedia.org [21:41:06] ok so these are just your observations, not logged in bug reports yet? [21:41:15] Some of them are, lemme dig [21:42:16] https://bugzilla.wikimedia.org/show_bug.cgi?id=20867#c6 is what led me to it [21:42:35] ok i'll start fixing those issues [21:42:54] That BTW is a separate issue altogether [21:42:55] Cool [21:58:19] ok [21:58:32] I'm finally back to coding [21:58:49] RoanKattouw: what's the status of the test.wikipedia.org deployment [21:59:04] there seemed to be some issues with the highlight plugin not being turned on [21:59:11] I hacked that [21:59:17] right on [21:59:21] Look at your WikiEditor.combined.min.js file [21:59:23] yes [21:59:24] I see it [21:59:37] The highlight code should check for wgWikiEditorEnabledModules.toc instead of .highlight [21:59:48] There was a previous version that did check for .highlight [22:00:08] So could you confirm that it really does check for the right one and that .toc is true? [22:00:31] I just realized the reason this works for me and Hannes and not for you and Naoko could be Squid cache: Amsterdam vs. Tampa [22:00:58] ah [22:01:00] interesting [22:01:30] seems to be working now [22:01:33] excellent [22:01:58] Naoko said she hard refreshed a million times but it didn't work [22:02:03] so we still have some outstanding bugs, plus I was hoping to refine the interaction with text a bit [22:02:23] If anyone still has this issue pop up the JS console and look for errors plus check the .min.js file [22:02:25] Yeah [22:02:31] I asked Adam to handle the dialogs issues [22:02:48] Specifically, the fact that the textbox tooltip is now inside the textbox causes a lot of issues [22:02:48] do some key interceptions for undo, redo and deleting to prevent paragraphs from disapearing unexpectedly (cursor at begining, then pressing backspace deletes entire paragraph) [22:02:58] Inserting the tooltip, running an existence check on it, etc. [22:03:07] Yeah you mentioned that [22:03:17] So why don't you work on the undo/redo/paragraph deletion thing [22:03:28] While I sift through Calcey bugs [22:03:29] pdhanda: you working at home? [22:03:45] *RoanKattouw hopes pdhanda will feel better this weekend [22:03:48] RoanKattouw: sounds great [22:04:04] adam_miller: you doing ok with the dialog stuff? [22:04:16] i just started on it [22:04:42] safari isn't throwing events when you delete characters from a text field apparently [22:05:33] what event are you listening for? [22:05:47] keypress paste cut [22:06:21] did you see this? http://www.quirksmode.org/js/keys.html [22:06:50] Most special keys (backspace, arrows) don't throw keypress in non-Firefox browsers [22:06:58] Only keydown [22:07:05] Even keyup is not reliably supported for them IIRC [22:07:25] what else could we listen for? change doesn't seem to do it either [22:07:31] this is why I setup that event filter system in wikiEditor core [22:07:49] so we could potentially listen for all of them, and then only pass on the ones we want to [22:08:04] so you could capture keyup and keypress [22:08:51] adding keyup did the trick [22:08:55] then add an event filter in wikiEditor core that says if you are a keyup and a delete go ahead, otherwise die, if you are a keypress and a delete die, otherwise go ahead [22:09:24] that way we don't end up with double events in a browser that's letting both through [22:09:29] are there examples of filters in wikiEditor right now? [22:09:57] yes, line 263 [22:09:58] nm, found them [22:10:00] context.evt [22:10:01] :) [22:10:08] It's really just return true/return false [22:10:10] if you return false, it kills the event [22:10:21] yes, but you can also add data to the event object which will continue on [22:10:24] TrevorParscal: I also fixed an interesting bug you introduced in our onbeforeunload hook [22:10:33] Apparently return null; in IE doesn't do what you want [22:10:34] RoanKattouw: oh yeah? [22:10:39] interesting [22:10:45] IE will coalesce *any* return value, including null and false, to a string [22:10:53] The only thing that works is not returning anything [22:11:08] Of course that subsequently broke the !== null check [22:11:17] Changing that to !== undefined now [22:13:03] would using preventDefault() be a better approach? [22:14:10] or i guess that's not sufficient to halt the chain of bound events [22:14:18] *event handlers [22:17:06] No I mean for event filtering in evt: [22:17:18] You're /supposed/ to use return true/false there, that's how Trevor designed it (right?) [22:17:32] *RoanKattouw hates IE [22:17:48] I previously wrote about how IE forgets the selection in the iframe on blur [22:18:07] This is avoidable for toolbar buttons, but obviously not for dialogs *hate* [22:18:31] So I'm gonna need selection saving voodoo [22:19:12] yes, return false [22:19:32] there's a test on line 308 [22:19:50] and if the filter returns false, it stops propagation [22:20:22] well i'm going to test just swapping keypress for keyup [22:20:30] works in safari and FF [22:20:45] just be on the lookout for side effects [22:21:06] I know that mouseup rather than click returns events for any mouse button rather than just the primary one [22:21:34] the keyup event will be triggered when you press the non character keys too [22:21:54] which is probably good for us, but, yeah, just be looking for that [22:23:27] yeah i don't think there's any harm in that [22:24:25] RoanKattouw thanks, yeah i just threw out my back so hopefully will be better by tomorrow if i stay home and ice it today [22:25:13] Yeah the last thing you wanna do is sit in a plane seat for 10 hours [22:26:30] depends how comfortable the plane seat is [22:27:08] Well she booked it through WMF so it's not business class :) [22:27:33] and probably not very comfortable [22:27:50] Well it is a long-haul flight [22:28:02] Not that crap they give you on domestic flights [22:41:26] hoprefully it will be ok by thursday, it feels a lot better than yesterday already [22:42:09] *RoanKattouw hates IE more [22:42:23] Not only does it blur the iframe when I don't want it to, it also doesn't seem to throw the blur event [22:55:37] *TrevorParscal decides to ditch the iframe and use HTML5 canvas [22:56:01] no but seriously, eventually, that's not a bad direction [22:56:23] Mozilla's Bespin could totally be supplimented to support wikitext [23:35:05] TrevorParscal: if you attempt to use any of the insert dialogs in safari without first clicking into the iframe, they wont work [23:35:16]