| 00:01 | <Hixie> | the tools will save us! |
| 00:01 | <Hixie> | except apparently in this case: http://www.shelter-systems.com/grip-clips.html |
| 00:01 | <Hixie> | where the user got confused between "list bullet" and "radio button" |
| 00:03 | <gavin_> | heh |
| 00:04 | <kingryan> | yeah, Hixie, I can't wait for the day, far in the future, when the tools will save us |
| 00:04 | <kingryan> | of course, then they'll develop AI, violate the 3 laws and destroy human civilization |
| 00:05 | <marcosc> | but at least we will be saved! |
| 00:05 | <kingryan> | our robot masters will save us from ourselves |
| 00:09 | <Dashiva> | They won't save us, they'll just save HTML |
| 00:12 | <Philip`> | jacobolus: I was using pstoedit (via Inkscape) to convert PS->SVG, and that appears to only work correctly if you make it convert all text into curves |
| 00:12 | <Lachy> | looks like they wanted to create an inline list, rather than a vertical bulleted list, and wasn't aware of either stylesheets or characters like U+2022 Bullet • |
| 02:01 | <Lachy> | I wonder why Křištof is arguing about the meaning of "checked"? http://lists.whatwg.org/pipermail/whatwg-whatwg.org/2007-September/012520.html |
| 02:22 | <Hixie> | i dunno, but i hope nobody tries to correct him |
| 03:07 | <Webtester01> | Hello. |
| 03:08 | <Webtester01> | I have an idea that would be great for web applications. |
| 03:11 | <Webtester01> | Related to javascript and html5 |
| 03:11 | <Webtester01> | . |
| 03:20 | <Webtester01> | Hello. |
| 03:23 | <Webtester01> | I propose two attributes to the the script tag: private and allowid. The private attribute would isolate a script's variables from other scripts on the page, and the allowid attribute would contain ids of scripts or objects allowed to access the script marked private. |
| 03:25 | <deltab> | you can do the former with (function () { ... })() |
| 03:26 | <Webtester01> | Scripts marked as private would be allowed to make persistant socket connections, ie: var sock = XMLSocket(host,port), or some other syntax. |
| 03:26 | <Lachy> | where is XMLSocket defined? |
| 03:26 | <Webtester01> | This is just a hypothetical idea. |
| 03:26 | <Lachy> | maybe you want the TCPConnection interface |
| 03:27 | <Lachy> | that's already in HTML5, or maybe you actually want <event-source> |
| 03:27 | <Webtester01> | Ok. |
| 03:27 | <Webtester01> | So I can connect to a host other than the domain, on which the html 5 page resides? |
| 03:28 | <Webtester01> | For example, a web-based IRC client, that uses html and styling to produce a gui. |
| 03:29 | <Webtester01> | Or a terminal? |
| 03:29 | <Lachy> | you would have to proxy messages through a custom server to do that |
| 03:30 | <Lachy> | cross-domain security restrictions apply |
| 03:30 | <Webtester01> | Yes. |
| 03:30 | <Webtester01> | That's why I had some security ideas in place, regarding the idea I mentioned earlier. |
| 03:31 | <Lachy> | the way you described the private attribute wouldn't actually solve those problems |
| 03:32 | <Lachy> | Webtester01, http://www.whatwg.org/specs/web-apps/current-work/multipage/section-network.html#network |
| 03:32 | <Webtester01> | Well the private script would not be able to access the pool from the other scripts either. |
| 03:33 | <Lachy> | so? |
| 03:33 | <Lachy> | it would still be able to modify the DOM and insert information into that. Other scripts could then read from the DOM |
| 03:35 | <Lachy> | what do you want cross-domain TCP connections for? Do you really want to implement an IRC client in javascript? |
| 03:37 | <Webtester01> | Well I wrote one in java for the socket connection, then javascript creates the gui. |
| 03:38 | <Lachy> | can create connections to any arbitrary host? |
| 03:38 | <Lachy> | can *java create ...? |
| 03:39 | <Webtester01> | Not cross domain. |
| 03:39 | <Lachy> | ah, so similar restrictions apply |
| 03:40 | Webtester01 | will be back soon. Dinner. |
| 03:44 | <Lachy> | Hixie, in TCP connections, there's an issue that states "We currently don't allow connections to be set up back to an originating IP address, but we could, if the subdomain is the empty string." |
| 03:44 | <Lachy> | then that's followed by "Then, if the subdomain argument is null or the empty string, the target host is the domain part of the script's origin." |
| 03:45 | <Lachy> | so it seems that that issue has been resolved already |
| 03:45 | <Hixie> | really? |
| 03:45 | <Lachy> | yeah, unless I'm misunderstanding something |
| 03:49 | <Hixie> | oh well |
| 03:49 | <Hixie> | i'm sure it'll come out in the wash when that section gets its next hose down |
| 03:51 | <Lachy> | it's also not clear how often the read event is fired for the Connection interface |
| 03:52 | <Lachy> | I assume it's probably once for every TCP packet (for the TCPConnection) |
| 03:52 | <Hixie> | yeah i wouldn't look too closely at that section |
| 03:52 | <Hixie> | it needs much work |
| 03:52 | <Lachy> | I realise that (given the number of red boxes in it) |
| 03:53 | <Hixie> | :-) |
| 04:12 | <Lachy> | Hixie, when are planning to edit the spec some more? it's been almost a month since the last commit |
| 04:13 | <Hixie> | yeah, been at burning man and on a boat trip |
| 04:13 | <Hixie> | this part of the year is my vacations |
| 04:13 | <Hixie> | :-) |
| 04:13 | <Hixie> | that's over though |
| 04:13 | <Hixie> | i've been editing today in fact |
| 04:13 | <Hixie> | writing up the datatemplate algorithms |
| 04:14 | <Hixie> | spec on the site is regenned |
| 04:14 | <Hixie> | still incomplete though |
| 04:17 | <Lachy> | oh cool, I was wating to find out more about datatemplate |
| 04:19 | <Lachy> | I'll have to read it later though, right now I'm trying to install MySQL |
| 04:20 | <Lachy> | hi othermaciej |
| 04:21 | <othermaciej> | hey Lachy |
| 04:21 | <othermaciej> | colloquy keeps messing up my nick |
| 04:21 | othermaciej | swears |
| 08:46 | <Dashiva> | http://www.joelonsoftware.com/items/2007/09/18.html?provocative_title |
| 09:00 | <Hixie> | Dashiva: he's probably right |
| 09:00 | <Hixie> | though google already has something like what he describes (GWT) |
| 09:00 | <Hixie> | compiles Java straight to JS+DOM, cross-browser |
| 09:02 | <othermaciej> | Google also compiles JS down to JS |
| 09:02 | othermaciej | is on the receiving end of that a fair bit |
| 09:03 | <kfish> | http://haskell.org/haskellwiki/Yhc/Javascript |
| 09:04 | <kfish> | ((((a real language, but no need for endless trailing parentheses)))) |
| 09:04 | <Hixie> | othermaciej: true |
| 09:11 | <KevinMarks> | you can usually append something to get it uncompiled |
| 09:25 | <KevinMarks> | isn't HTML5 at least part of Joel's answer? |
| 09:25 | <Hixie> | HTML5 is a level below what joel's talking about |
| 09:25 | <Hixie> | it's like a new libc |
| 09:26 | <KevinMarks> | I was thinking this bit: "Right now the big hole in the portability story is — tada! — client-side JavaScript, and especially the DOM in web browsers. Writing applications that work in all different browsers is a friggin’ nightmare. There is simply no alternative but to test exhaustively on Firefox, IE6, IE7, Safari, and Opera, and guess what? I don’t have time to test on Opera. Sucks to be Opera. Startup web browsers |
| 09:26 | <KevinMarks> | don’t stand a chance." |
| 09:27 | <Hixie> | well, only insofar that html5 does a small amount to help interop |
| 09:27 | <Hixie> | but really interop problems are largely up to the browser vendors |
| 09:28 | <Hixie> | just having a detailed spec doesn't solve the problem, the browsers still have to be fixed to match the spec |
| 09:29 | <Hixie> | http://www.whatwg.org/specs/web-apps/current-work/#datatemplate |
| 09:29 | <Hixie> | behold |
| 09:30 | <hsivonen> | zcorpan: thanks. bug documented and queued for the great XHTML 1 refactoring |
| 09:31 | <othermaciej> | HTML5 can foster interop directly (by providing useful spec language and backing it by exhaustive tests |
| 09:31 | <othermaciej> | but also indirectly by creating a forum where the players talk about interop |
| 09:32 | <othermaciej> | and by fostering a culture where interop is aprimary value |
| 09:32 | <KevinMarks> | the comment about cut and paste is interesting |
| 09:32 | <othermaciej> | I'm not sure what question Joe is trying to answer |
| 09:32 | <KevinMarks> | when I cut/paste within firefox, it seems to just work |
| 09:32 | <othermaciej> | *Joel |
| 09:33 | <Hixie> | ok i just checked in several hundred lines |
| 09:33 | <Hixie> | hopefully that'll make up for the few weeks of nothing |
| 09:33 | <KevinMarks> | he's not traying to answer one, he's just being a old fart and saying he's seen ti all before |
| 09:33 | <Hixie> | KevinMarks: you can't cut and paste a contact from gmail and paste it into yahoo mail |
| 09:33 | <Hixie> | his concern is valid, imho |
| 09:33 | <Hixie> | othermaciej: did you decide you wanted to visit us for the offline stuff? |
| 09:33 | <KevinMarks> | he said an image |
| 09:34 | <KevinMarks> | but yes, a contact woudl be good |
| 09:34 | <KevinMarks> | *cough* hCard *cough* |
| 09:34 | <Hixie> | you wouldn't want to use hCard |
| 09:34 | <Hixie> | vCard would make more sense imho |
| 09:34 | <KevinMarks> | I would |
| 09:35 | <KevinMarks> | firefox does HTML copy paste nicely |
| 09:35 | <KevinMarks> | hCard is HTML |
| 09:35 | <Hixie> | html5 actually does introduce proper APIs for copy/paste though |
| 09:35 | <KevinMarks> | the other browsers need to catch up there |
| 09:35 | <Hixie> | so i guess in a way that's part of his SDK |
| 09:35 | <hsivonen> | KevinMarks: does HTML copy paste nicely within two text/html views? or otherwise? |
| 09:36 | <KevinMarks> | hm |
| 09:36 | <Hixie> | pasting html into a web page wouldn't really help |
| 09:36 | <Hixie> | you want to paste into the app, not into the page |
| 09:36 | <Hixie> | anyway |
| 09:36 | <Hixie> | i should sleep |
| 09:36 | <Hixie> | nn |
| 09:36 | <kfish> | javascript to watch selections and override the selection buffer (with an appropriate arbitrary xml representation of the selected data, contact etc.)? |
| 09:36 | <KevinMarks> | well, paste works in gmail and writely |
| 09:36 | <hsivonen> | ok |
| 09:37 | <KevinMarks> | with copy from $arbitrary_webpage |
| 09:37 | <hsivonen> | KevinMarks: have you tried copying from a page served as application/xhtml+xml? |
| 09:38 | <KevinMarks> | nah, they always show up as invalid anyway ;) |
| 09:38 | <hsivonen> | I had a good look at the Gecko clipboard code a bit over a year ago. I can't remember what happens with XML, but I'd expect trouble |
| 09:38 | <KevinMarks> | per Hixie, finding one is hard.... |
| 09:38 | <KevinMarks> | .02% wasn't it? |
| 09:38 | <hsivonen> | Planet Intertwingly |
| 09:39 | <KevinMarks> | or shelley I suppose |
| 09:39 | <KevinMarks> | come to think of it, I haven't seen anything by her in a while |
| 09:40 | <zcorpan> | KevinMarks: 0.014% had some xml mime type iirc (might not be xhtml, more likely rss) |
| 09:41 | <othermaciej> | KevinMarks: how come Pages and Pesentations have an HTML editing engine that utterly fails in Safari, but GMail and Writely work fine |
| 09:42 | <KevinMarks> | google is large, it contains multitudes |
| 09:42 | <hsivonen> | the only application/xhtml+xml blog that I've read repeatedly and on which I haven't seen the yellow screen of death is Jacques Distler's Musings |
| 09:42 | <othermaciej> | it contains a knowledge exchange problem |
| 09:45 | jgraham | thinks he remembers seeing the YSOD on Musings (at least, I remember Jacques said there had been a problem caused by a trackback with an unexpected character encoding) |
| 09:47 | <hsivonen> | well, I've had a vicarious YSoD on Musings. I instructed another person read something over there and he replied that he can't because of YSoD. (He had a Gecko build without the MathML pseudo-DTD in the catalog--Netscape 6.something.) |
| 09:51 | <KevinMarks> | did you see the footnote here, maciej? http://ejohn.org/blog/freeing-the-grid/ |
| 09:55 | <hsivonen> | interestingly, when Joel mentions compilation to cross-browser JS, he doesn't mention Google Web Toolkit |
| 09:55 | <othermaciej> | KevinMarks: the whiny part or the cnfusing part? |
| 09:57 | <othermaciej> | UI toolkits do soeem to have very aggressive expectations of the pace of new browser version deployment |
| 09:59 | <hsivonen> | btw, Joel is right about download times. Flickr sucked on GPRS (not because of photos but because of JS). Moving to HSDPA fixed it better than code optimization. |
| 10:00 | <othermaciej> | a fair chunk of Safari users are still on Safari < 2 |
| 10:01 | <othermaciej> | .63% of total market |
| 10:01 | <othermaciej> | close to Opera's total share |
| 10:01 | <othermaciej> | otoh developing for Safari 1.3 (or god forbit 1.1) would be painful |
| 10:04 | <KevinMarks> | I don't like dojo's basic model of crufting up the html with their own extensions |
| 10:09 | <hsivonen> | what are considered the least biased sources of browser statistics these days? |
| 10:09 | <KevinMarks> | measuring your won site |
| 10:09 | <KevinMarks> | *own |
| 10:10 | <KevinMarks> | it was an education seeing Orkut's browser stats |
| 10:12 | <hsivonen> | http://msdn2.microsoft.com/en-us/library/ms534654.aspx looks self-contradictory to me |
| 10:12 | <hsivonen> | "An element can have focus if the tabIndex property is set to any valid negative or positive integer." |
| 10:12 | <hsivonen> | shouldn't that be zero or positive? |
| 10:14 | <hsivonen> | hmm. perhaps I misunderstood and tabbability and focusability are different in IE. hmm. |
| 10:15 | <hsivonen> | othermaciej: has Apple said anything about ARIA in public? |
| 10:16 | <othermaciej> | hsivonen: not that I know of |
| 10:16 | <othermaciej> | I look at http://marketshare.hitslink.com/ usually |
| 10:17 | <othermaciej> | (re market share) |
| 10:17 | <othermaciej> | I don't have any specific reason to think their methodology is way better than the competition |
| 10:17 | <othermaciej> | but their numbers seem so align with my expectations |
| 10:18 | <othermaciej> | and it's not a one-site survey of a specialist site |
| 10:18 | <hsivonen> | hmm. they don't count engines |
| 10:19 | <hsivonen> | instead they isolate some branded products: Netscape, PSP and Playstation |
| 10:19 | <othermaciej> | you have to know what product is what engine |
| 10:19 | <othermaciej> | you can also get a more detailed version breakdown |
| 10:20 | <hsivonen> | when I write my perfect log analysis tool, it'll count engines |
| 10:21 | <othermaciej> | this is pretty detailed: http://marketshare.hitslink.com/report.aspx?qprid=6 |
| 10:21 | <othermaciej> | I don't know what the PSP browser is |
| 10:22 | <othermaciej> | Blazer seems to be the top entrant to clearly not use one of the top 4 engines |
| 10:22 | <hsivonen> | othermaciej: I've been told that PSP in NetFront |
| 10:23 | <othermaciej> | series 60 shows up in os share but I don't see a browser that is obviouslly theirs |
| 10:26 | <hsivonen> | weird. Opera Mini reveals my phone's built-in UA to servers in a custom header |
| 10:27 | <hsivonen> | the headers Mini sends are more crufty than headers that cause size concern in desktop Gecko... |
| 12:24 | <Lachy> | the copy/paste issue Joel wrote about in that article wouldn't be too hard to solve |
| 12:24 | <Lachy> | it's not that hard to implement a clipboard: http://html5.lachy.id.au/clipboard |
| 12:25 | <Lachy> | just need to develop the API a bit more, make it support multiple users (probably backed by a DB instead of a simple text file) |
| 12:26 | <Lachy> | then just get different web apps to hook into the API, and then you've got an instant shared clipboard between apps that works anywhere |
| 12:27 | <Dashiva> | And a huge giant SPOF |
| 12:28 | <Lachy> | SPOF? |
| 12:29 | <Lachy> | ah, Single Point of Failure? |
| 12:29 | <Lachy> | then make it a distributed system. |
| 12:30 | <hsivonen> | the browser should probably perform token exchange choreography between the sites and then initiate a site-to-site transfer with the browser-negotiated token |
| 12:30 | <hsivonen> | not simple |
| 12:31 | <hsivonen> | (for stuff like enabling copy-paste of large media files without round-tripping originals through the browser) |
| 12:32 | <Lachy> | if the API is standardised, web apps could talk directly to each other without having to round trip everything through the browser |
| 13:25 | <Lachy> | othermaciej, have you made any progress on the design principles today? |
| 13:55 | <jeremyb> | does dave raggett come by here? |
| 13:56 | jeremyb | wonders about the topic |
| 13:56 | <jeremyb> | or anyone know much about slidy/want to discuss it? |
| 13:57 | <jeremyb> | ooooh, /me likes the ability to highlight nicks in the logs and doesn't remember that from last time he looked at them |
| 14:01 | <Lachy> | jeremyb, the highlight feature was added about a week ago |
| 14:02 | jeremyb | now realizes that could be confused with the line highlighting which he does remember from the last time |
| 14:03 | <Lachy> | what's confusing about it? |
| 14:03 | <jeremyb> | my description? |
| 14:03 | <jeremyb> | when you highlight a line the nicks are highlighted too |
| 14:03 | <jeremyb> | nvm :) |
| 14:06 | <hsivonen> | jeremyb: I don't recall seeing Dave Raggett here in the last couple of months |
| 14:06 | <jeremyb> | anyway, gotta run, but if anyone wants to point me to a better place for slidy that'd be great. btw, i'm planning on doing something with slidy and a wiki. prolly 1 whole slide show per wiki page. |
| 14:06 | jeremyb | wonders if anyone's done anything like that |
| 14:06 | jeremyb | waves good morning to aaron |
| 14:06 | <jeremyb> | hsivonen: thx |
| 14:07 | <aaronlev> | good morning jeremyb |
| 14:07 | <aaronlev> | hi hsivonen |
| 14:07 | <hsivonen> | jeremyb: IIRC Jacques Distler blogged about putting S5 slides on Instiki |
| 14:07 | <hsivonen> | aaronlev: hi |
| 14:08 | <jeremyb> | hsivonen: s5? |
| 14:08 | <hsivonen> | jeremyb: http://golem.ph.utexas.edu/instiki/show/S5 |
| 14:08 | <jeremyb> | thx |
| 14:08 | jeremyb | runs (busy day :() |
| 14:09 | <hsivonen> | jeremyb: S5 is a competitor of Slidy if I have the right idea of what Slidy does |
| 14:38 | <MikeSmith> | hsivonen - The GNU coding standards for error messages are documented here: |
| 14:38 | <MikeSmith> | http://www.gnu.org/prep/standards/html_node/Errors.html#Errors |
| 15:06 | <hsivonen> | MikeSmith: thanks |
| 15:42 | <Philip`_> | APNG interoperability is not good so far :-( |
| 15:48 | <hsivonen> | Philip`_: do Firefox 3 and Opera 9.5 use the same libpng extensions? |
| 15:49 | <hsivonen> | that is, is this a graphics drawing layer issue of an APNG impl. level issue? |
| 15:50 | <hsivonen> | what's today's compat situation with <m>? should I use it or go with a styled <b> or something? |
| 15:51 | <zcorpan> | hsivonen: you can't style it in ie7 |
| 15:52 | <hsivonen> | zcorpan: ok. then I'm not gonna use it |
| 15:52 | hsivonen | mumbles something about <u> |
| 15:52 | <zcorpan> | <u> works fine :) |
| 15:53 | <hsivonen> | zcorpan: it doesn't validate as HTML5 as of today |
| 15:56 | zcorpan | tries to wrap his head around datatemplates |
| 16:01 | <Philip`_> | hsivonen: From the differences, it looks like they're independent implementations - e.g. Opera thinks offset_x=2^32-1 means offset one pixel left, whereas Firefox thinks it's a fatal error |
| 16:02 | <Philip`_> | (The spec says that case is an error, so Opera I think is wrong (except error-handling doesn't seem to be strictly defined anywhere)) |
| 16:06 | <Philip`> | and e.g. Opera doesn't like zero-length fdAT chunks, whereas Firefox accepts them correctly |
| 16:06 | <hsivonen> | Philip`: ok |
| 16:25 | <virtuelv> | Philip`: filed a bug? |
| 16:32 | <Philip`> | virtuelv: Not yet - I expect there are lots more bugs, so I'm trying to develop a semi-organised way to find as many as possible |
| 16:53 | <zcorpan> | well, i have a bit trouble grasping the levenberg algorithms and thus how the whole datatemplate things actually works |
| 16:53 | <zcorpan> | can someone explain how it's supposed to work or give an example? :) |
| 18:14 | <Philip`> | zcorpan: I'm guessing it works somewhat like http://philip.html5.org/demos/datatemplate/experimental/001.html though I'm not entirely certain :-) |
| 18:19 | <Philip`> | (For backward compatibility, I guess you'd have to put the <datatemplate> and <items> parts in separate XML documents, else HTML parsers will mangle them horribly) |
| 18:43 | Philip` | wonders who Levenberg is |
| 19:24 | <hsivonen> | any better way to visualize newlines in Unicode than U+21A9? |
| 19:29 | Philip` | now remembers that HTML4 mentions the &{ script }; syntax as being reserved for script macros, and wonders if that could be used for text expansion of templates since it shouldn't conflict with existing syntax |
| 19:30 | <hsivonen> | Philip`: what about XHTML5? |
| 19:30 | <Philip`> | Hmm |
| 19:30 | <Philip`> | I suppose you'd have to write &{ script }; to get the equivalent DOM |
| 19:31 | <Philip`> | (assuming the &{ script }; template expansion happens somewhere outside the parser, so you should get the & character in the DOM) |
| 19:31 | <hsivonen> | maikmerten: what's the closed captioning and closed audio description situation with the Ogg family? |
| 19:31 | <Philip`> | which isn't overly pretty |
| 19:32 | <maikmerten> | hsivonen, I'm currently trying to figure out what "closed" means in that context ;) |
| 19:32 | <maikmerten> | ah, subtitles |
| 19:32 | <hsivonen> | maikmerten: closed means opt-in |
| 19:33 | <hsivonen> | maikmerten: open captioning means that everyone sees them by default |
| 19:33 | <maikmerten> | hsivonen, I think using annodex + ogg together with Ogg Skeleton can give the semantics wanted |
| 19:34 | <maikmerten> | (Ogg Skeleton is basically carrying meta information about streams multiplexed in Ogg) |
| 19:34 | <hsivonen> | maikmerten: does annodex bring in more complexity than just timed text? |
| 19:34 | <hsivonen> | maikmerten: whatever happened to Ogg Writ? |
| 19:35 | <maikmerten> | let's simply say that Ogg Writ did fall flat onti its face |
| 19:35 | <maikmerten> | onto |
| 19:35 | <hsivonen> | ouch |
| 19:36 | <maikmerten> | the way to go is annodex, which is semantically more rich than yet-another-Ogg-only-thingie anyway |
| 19:36 | <hsivonen> | I'm a bit afraid of scope creep |
| 19:37 | <maikmerten> | well, if there's need for a "more straightforward" subtitle thingie I guess it may be possible to specify that, too (given the manpower and time) |
| 19:37 | <hsivonen> | maikmerten: I see. I was looking for an off-the-shelf replacement for 3GP Timed Text |
| 19:38 | <hsivonen> | MPEG-4 part 17, IIRC |
| 19:38 | <maikmerten> | http://trac.annodex.net/wiki/CmmlSubtitles |
| 19:38 | <hsivonen> | and also a way to have a second sound track off-by-default and flagged as audio description |
| 19:38 | <hsivonen> | thanks |
| 19:39 | <maikmerten> | http://wiki.xiph.org/index.php/Ogg_Skeleton |
| 19:39 | <maikmerten> | Skeleton is exactly for that kind of stuff AFAIK |
| 19:39 | <hsivonen> | thanks |
| 19:40 | <maikmerten> | oggplay, which is the kitchen-sink solution to play Ogg (and used by Chris Double for Firefox <video>) carries support for Annodex and IIRC Skeleton |
| 19:40 | <hsivonen> | ooh. cool. |
| 19:41 | <maikmerten> | well, liboggply is developed by the annodex guys ;) |
| 19:41 | <maikmerten> | little surprise here |
| 19:41 | <maikmerten> | but makes all of our lives much easier ;) |
| 19:41 | <hsivonen> | maikmerten: well, I am a bit surprised, because I've been awfully naive about the status of Writ |
| 19:42 | <maikmerten> | hsivonen, well, seems Writ can/does work |
| 19:42 | <hsivonen> | the Wrip page on the Xiph wiki and the one in Wikipedia give the impression of ongoing work instead of abandonware |
| 19:42 | <hsivonen> | oh |
| 19:42 | <maikmerten> | it's just that it never actually made it to some sort of "recommendation" |
| 19:43 | <maikmerten> | plus I may be misinformed about the state of things |
| 19:43 | <maikmerten> | and if there's demand a solution isn't out of reach, I guess |
| 19:43 | <maikmerten> | I'll have to investigate |
| 19:43 | <hsivonen> | in order to promote the Ogg family for <video>, the accessibility stuff needs to be sorted out at some point |
| 19:44 | <maikmerten> | - what the exact status of subtitles is in Annodex |
| 19:44 | <maikmerten> | - what the status of Writ is |
| 19:44 | <maikmerten> | (or if there's a newer approach) |
| 19:44 | <maikmerten> | I see |
| 19:45 | <hsivonen> | as a sidenote, I've so far learned that "subtitles" is a bad word in North America in the accessibility context and captioning is the preferred word. (even though en-GB uses subtitles) |
| 19:45 | <hsivonen> | subtitles has a translation connotation |
| 19:46 | <maikmerten> | oh |
| 19:47 | <hsivonen> | I'll put the urls you gave into Bugzilla |
| 19:49 | <maikmerten> | http://svn.xiph.org/trunk/writ/ <-- the writ implementation, by the way |
| 19:49 | <maikmerten> | I never tried it, but it seems like it's doing something |
| 19:50 | <gsnedders> | Hixie: is it worth defining application/octect-stream sniffing at all, for the sake of HTML documents served as such? |
| 19:51 | <hsivonen> | maikmerten: thanks |
| 19:52 | <maikmerten> | hsivonen, you're welcome |
| 19:53 | <maikmerten> | I'm currently writing a mail to our advocacy mailing list forwarding your request and asking for status information |
| 19:53 | <hsivonen> | maikmerten: thanks |
| 20:59 | <Hixie> | Philip`: wow, you pretty much nailed it. |
| 20:59 | <Hixie> | Philip`: i'm glad i didn't specify the expansion algorithm, you came up with something quite interesting that i hadn't thought of |
| 20:59 | <Hixie> | i was thinking of using an extensions to CSS |
| 20:59 | <Hixie> | also, you don't need a registration mark in your example |
| 21:00 | <Hixie> | the only other thing is you'd want to have the input elements update the items |
| 21:02 | <Hixie> | e.g. onchange="dataNode.setAttribute('value', value)" value="&{this.getAttribute('value')}" |
| 21:12 | <Hixie> | othermaciej: interesting site for browser stats, though they really should show the data on a log scale |
| 21:13 | <Hixie> | they could show more browsers that way too |
| 21:13 | <othermaciej> | Hixie: that would certainly make it easier to see the trends |
| 21:14 | <Philip`> | I'm not sure why I bothered putting the registrationmark in, since it's useless without scripting |
| 21:15 | <Hixie> | Philip`: it's useful when the data regenerates and hits a different set of conditions, but you want the same element to be used for different parts of the two conditions, yeah |
| 21:15 | <Hixie> | othermaciej: these stats also don't include the iPhone browser, which i believe would be interesting to see |
| 21:16 | <othermaciej> | Hixie: if you mean these stats: <http://marketshare.hitslink.com/>;, then they lump the iPhone browser in with Safari, but you can see it in the OS share stats |
| 21:17 | <othermaciej> | (you can also see Nintendo Wii in the OS share stats) |
| 21:17 | <Hixie> | ah, so they do |
| 21:17 | <Hixie> | but those don't appear on the line graph |
| 21:17 | <Hixie> | pah |
| 21:17 | <Hixie> | do you know if there's a way to download those numbers without site scraping? |
| 21:17 | hsivonen | reads old Joel Strategy Letters over supper |
| 21:17 | <Philip`> | Oops, I missed the bit where the algorithm does useful stuff if you regenerate the structure without any registrationmarks |
| 21:18 | <hsivonen> | (they should be required reading) |
| 21:18 | <Hixie> | or without downloading each page in turn? |
| 21:18 | <Hixie> | hsivonen: uris? |
| 21:18 | <Hixie> | Philip`: that was the Levenberg breakthrough. :-) |
| 21:19 | <hsivonen> | Hixie: http://www.google.fi/search?q=site%3Ajoelonsoftware.com+strategy+letter top six hits |
| 21:19 | <Hixie> | (btw for what it's worth i can confirm that these marketshare.hitslink.com stats are remarkably high quality) |
| 21:19 | <Hixie> | hsivonen: aha |
| 21:20 | <hsivonen> | HTML5 (unlike XHTML2) gets plugging into existing network right |
| 21:22 | <hsivonen> | Hixie: btw, what ARIA has going for it over HTML5 plus XBL is that ARIA doesn't disrupt existing systems |
| 21:25 | <othermaciej> | Hixie: they do show a surprising amount of month-to-month variance which seems unlikely to reflect real changes in user base in at least some cases, but since it's based on usage, not count of unique users, I guess that could just reflect variability in browsing habits |
| 21:25 | <othermaciej> | Joel's letter # VI confuses me |
| 21:26 | <othermaciej> | afaik Google has at least two systems that compile languages down to JavaScript (with the source languages being Java and JavaScript respectively) |
| 21:26 | <hsivonen> | othermaciej: yeah, I noticed that he didn't mention GWT |
| 21:27 | <othermaciej> | there's also less reason on the web for a single library-level SDK to dominate, since there is no bundling/integration issue |
| 21:40 | <hsivonen> | btw, http://marketshare.hitslink.com/report.aspx?qprid=2 must be heavily U.S.-biased |
| 21:40 | <hsivonen> | there's no way the iPhone could have more market share than S60 globally |
| 21:41 | <hsivonen> | http://vowe.net/archives/008814.html |
| 21:42 | <othermaciej> | hsivonen: it's usage share, so it's possible if iPhone users browse a lot more per user |
| 21:45 | <othermaciej> | I guess WinCE shows up pretty high too though |
| 21:46 | <Hixie> | iPhone users use the web way more than any other cell phone |
| 21:47 | <Hixie> | hsivonen: as i understand it it's not HTML5+XBL vs <div>+ARIA, it's HTML5 vs <div>+ARIA |
| 21:48 | <Hixie> | hsivonen: the XBL is only required if you want to start making custom new widgets, which isn't anywhere near as important in this case afaict |
| 21:49 | <hsivonen> | Hixie: ARIA is about retrofitting custom JS+div widgets with accessibility if I've understood correctly |
| 21:52 | <hsivonen> | Hixie: XBL is also required if you want to make native HTML5 widgets looks kewl the way your art dept required |
| 21:52 | <Philip`> | http://philip.html5.org/demos/datatemplate/experimental/002.html - now somewhat more elaborate and even more likely to be wrong |
| 21:52 | <Philip`> | What'd be really nice is a scripted implementation to test things with :-) |
| 21:53 | <Hixie> | hsivonen: yeah |
| 21:53 | <Philip`> | Oops, I think the <rule condition="categories"><nest filter="category"></rule> is unnecessary |
| 21:54 | <Philip`> | (but the <rule condition="items" mode="summary"><nest id="item_filter" mode="summary"></rule> is still necessary) |
| 21:54 | <Hixie> | Philip`: dude, you kick ass. this is exactly what we need to check the spec makes sense. |
| 21:54 | <Hixie> | 002 has two item_filter IDs |
| 21:55 | <Hixie> | also, holy crap, i hadn't thought of modifying the <nest> dynamically |
| 21:55 | <Hixie> | that's awesome |
| 21:55 | <Hixie> | the spec will have to say the UA must monitor the template, too |
| 21:55 | <hsivonen> | Hixie: as far as I can tell, if someone replaces HTML 4.01 buttons, radio buttons and checkboxes with divitis, they'll going to want to do the same with sliders even when those are natively available |
| 21:56 | <Philip`> | Fixed item_filters |
| 21:56 | <Hixie> | hsivonen: yeah, can't argue with that |
| 21:56 | <Hixie> | hsivonen: let's hope xbl2 gets implemented soon :-/ |
| 21:57 | <othermaciej> | some of the kewlness can be achieved just through CSS styling of standard controls |
| 21:57 | <othermaciej> | although browsers vary in how much of that they allow |
| 21:57 | <othermaciej> | it's doable for buttons and text fields, but not really for checkboxes or radio buttons afaict |
| 21:58 | <Hixie> | yeah |
| 22:00 | <othermaciej> | you can style them in Safari 3 by setting "-webkit-appearance: none" but I don't think there is any way in IE |
| 22:06 | <Hixie> | did i get unsubscribed from public-html or something |
| 22:06 | <Hixie> | there's been no e-mail to it for hours |
| 22:10 | <jgraham> | Hixie: It's been surprisingly quiet |
| 23:42 | <copper> | Hi |
| 23:42 | <copper> | "Please leave your sense of logic at the door"? |
| 23:43 | <Hixie> | hi copper |
| 23:43 | <Hixie> | Philip`: so my original idea was to have bla{input::attr(value)}bla sort of stuff |
| 23:43 | <Hixie> | Philip`: but i'm not a big fan of that either |
| 23:47 | <Hixie> | i do kind of like the JS idea |
| 23:47 | <Hixie> | maybe we can say that any mutations of the DOM during expansion will cause the entire thing to abort |
| 23:50 | <Hixie> | Philip`: your 002 has an error -- your onchange should not be in &{} |
| 23:50 | <Hixie> | interesting error though. might well be common. |
| 23:50 | <Philip`> | Would implementors find it easier to detect and abort from DOM mutations than to make the algorithm interruptible on a timeout? |
| 23:51 | <copper> | People can't really serve their documents as XHTML because of MSIE... Since Microsoft doesn't support the development of (X)HTML5, does it mean that XHTML is definitely out of the window? |
| 23:51 | <Hixie> | Philip`: implementors have to detect DOM mutations anyway for other parts of the algorithm |
| 23:51 | <Hixie> | also, your data tree is non-compliant -- <div>s can't contain <shop>s |
| 23:51 | <Hixie> | copper: not definitely, but for now, certainly, i wouldn't recommend using XHTML |
| 23:51 | <Philip`> | (They'd have to support &{while(1);} too, so it doesn't seem obvious that they couldn't recover from infinite loops elsewhere) |
| 23:52 | <copper> | Hixie: on practical ground, not technical grounds, right? |
| 23:52 | <Philip`> | (or at least it doesn't seem obvious to me, but then I've never written a web browser and don't intend to :-) ) |
| 23:52 | <Hixie> | copper: on practical technical grounds, right, as opposed to theoretical ones |
| 23:52 | <copper> | :( |
| 23:53 | <Hixie> | copper: but don't worry, we're working on html5, so html won't be left behind |
| 23:53 | <Hixie> | Philip`: catching an infinite loop in your own code is a lot harder than catching it in code you are running in a sandbox |
| 23:53 | <copper> | Good for you, but I was hoping XHTML would win eventually |
| 23:53 | <Hixie> | copper: well, it might eventually, if IE loses more market share or implements XHTML |
| 23:54 | <Hixie> | copper: HTML5 and XHTML5 are being developed in lockstep |
| 23:54 | <Philip`> | Hixie: I assume I'd have to use X(HT)ML to make it not non-conformant HTML, so I'm happy to neglect conformance for convenience as long as it works :-) |
| 23:54 | <Hixie> | copper: so we're agnostic as to which one succeeds |
| 23:54 | <Hixie> | Philip`: well, you'll probably have to use XML for the datatemplate anyway |
| 23:54 | <Hixie> | Philip`: e.g. your <rule condition="category"><option ...></rule> rule will probably not parse as you'd expect |
| 23:55 | <Hixie> | same as your <select> <nest/> </select> |
| 23:55 | <copper> | That's good, but it doesn't give people the choice to serve XHTML, because we are still dependent on browser vendors... |
| 23:55 | <Philip`> | Hmm, I guess that means I'll have to give up the convenience :-( |
| 23:55 | <Hixie> | copper: yup, not much we can do about that from the specs side of things :-) |
| 23:56 | <copper> | *sigh* |
| 23:56 | <Philip`> | except this is a non-working demonstration, so I can keep the convenience even though it wouldn't actually work in reality :-) |
| 23:56 | <Hixie> | Philip`: yeah, i don't expect to have <datatemplate> ever work in HTML5 without external files for the template at least (and the data, if the data isn't in HTML) |
| 23:56 | <Hixie> | (though the data could easily be in HTML, e.g. using <ol>, maybe marked up as a microformat) |
| 23:56 | <Philip`> | (Fixed the onchange) |
| 23:57 | <copper> | Hixie: do you know from experience if an X(HT)ML-only browser would be faster than an HTML/XHTML one? |
| 23:57 | <Hixie> | copper: wouldn't make any difference in speed |
| 23:58 | <copper> | just code complexity? |
| 23:58 | <copper> | or not even that? |
| 23:58 | <copper> | (on the browser's side I mean) |
| 23:58 | <Hixie> | copper: it's almost entirely the same code, except for the parser, which is just different, not especially more or less complex or slower or faster, and not even that different really |
| 23:58 | <copper> | I would have thought a tag-soup parser would be much more complex? |
| 23:59 | <Hixie> | not really |
| 23:59 | <Hixie> | and xml is pretty complex to parse itself anyway |