Options:
- # Session Start: Mon Dec 31 00:00:00 2007
- # Session Ident: #html-wg
- # [00:00] <Hixie> though my kinesis professional dvorak keyboard is totally programmable and i've programmed it so that caps-lock is alt and scroll-lock is caps-lock and other silly things
- # [00:01] * Lachy notes that the Australian keyboard layout is the best ;-)
- # [00:01] <inimino> did you learn how to use the dvorak layout?
- # [00:02] <Lachy> (it's basically the same as the US layout)
- # [00:02] * inimino has always assumed it wouldn't be worth the effort
- # [00:04] * gsnedders has always assumed it is, but has never been able to afford the loss of productivity in learning a new layout
- # [00:07] <Hixie> inimino: yes
- # [00:07] <Hixie> inimino: my hands started REALLY hurting at one point
- # [00:07] <Hixie> inimino: and so i took 2 months off basically without typing, during which time i learnt dvorak
- # [00:07] <inimino> Hixie: did it help?
- # [00:07] <Hixie> yes
- # [00:08] <inimino> I'll keep that in mind, RSI is rather an occupational hazard :(
- # [00:08] <Hixie> now i alternate between three kinds of keyboards on regular basis
- # [00:08] <Hixie> to keep my hands from getting into the same pattern for too long
- # [00:08] * gsnedders realises he's had this physics textbook sitting beside him for weeks, yet hasn't actually done any revision from it with exam in mid-Jan :\
- # [00:08] <Lachy> where can you buy a dvorak keyboard from? I've never seen one in any store
- # [00:09] <inimino> Lachy: there are some online shops that specialize in ergonomics keyboards
- # [00:09] <anne-mac> just flip the keys yourself :)
- # [00:09] <Hixie> i got mine from kinesis
- # [00:10] <Lachy> I use the Microsoft ergonomic keyboard
- # [00:10] * gsnedders ought to stop using the keyboard built in to the MBP.
- # [00:10] <Hixie> i recommend http://www.kinesis-ergo.com/advantage.htm
- # [00:10] <gsnedders> Not that nice to use for a long period of time
- # [00:11] * gsnedders could never imagine using such an odd shaped keyboard
- # [00:12] <Hixie> it's really good
- # [00:12] <inimino> cool, I've looked at those but never knew anyone that used one
- # [00:13] <Lachy> that keyboard looks too weird for me
- # [00:13] <inimino> gsnedders: if you think that's odd, check out the datahand: http://www.datahand.com/
- # [00:14] <gsnedders> inimino: WTF?
- # [00:14] <gsnedders> I literally don't know what else to say.
- # [00:14] <Hixie> one of my colleagues uses the data hand
- # [00:14] <Hixie> and i know a bunch of people who use the kinesis
- # [00:15] <Lachy> how on earth does that datahand work?
- # [00:15] <inimino> gsnedders: maybe it all makes more sense when you get your first brush with RSI
- # [00:16] <gsnedders> inimino: but how does it even work?
- # [00:16] <gsnedders> inimino: what? you mean youthful naïvety doesn't last?
- # [00:16] <inimino> I haven't used it but it has something like four switches and detects finger motion in various directions
- # [00:16] <gsnedders> Just typed in IRC, then paused for a moment: "/me ought to go to bed so he can get off". Can't remember the end of the sentence.
- # [00:17] <inimino> s/four switches/four switches per finger/
- # [00:17] <gsnedders> But, I ought to go to bed so I can get off to my grandmother's funeral on time tomorrow
- # [00:18] * Quits: gsnedders (gsnedders@86.138.198.209) (Quit: Partying in teh intarwebs)
- # [00:25] <Philip> X3D says "The class attribute is reserved for future use with XML Cascading Style Sheets" - what is the point of that statement? You couldn't use it even if it wasn't "reserved" (since you can't use undefined attributes), and you can't use it now since nobody has tried to define what it means
- # [00:25] <anne-mac> what's the point of this question in this channel
- # [00:25] <anne-mac> ? :p
- # [00:26] <Hixie> Philip: it's political manoeuvering
- # [00:26] <Philip> It's related to CSS which is related to HTML :-)
- # [00:26] <Philip> (...kind of tenuously)
- # [00:26] <anne-mac> as far as X3D experts go you're probably the closest we got :)
- # [00:26] <Hixie> 53 tests to go
- # [00:27] <Philip> It was kind of a rhetorical question - I don't expect anyone to actually know the details of why that was written :-p
- # [00:50] * Quits: anne-mac (annevk@86.90.70.28) (Ping timeout)
- # [00:50] * Quits: anne (annevk@145.53.12.180) (Ping timeout)
- # [01:10] * Parts: osiris (opera@84.215.119.246)
- # [01:14] * Joins: olivier (ot@128.30.52.30)
- # [01:36] * Quits: jgraham (james@86.7.213.72) (Quit: This computer has gone to sleep)
- # [01:50] * Joins: jgraham (james@86.7.213.72)
- # [01:50] * Quits: jgraham (james@86.7.213.72) (Quit: This computer has gone to sleep)
- # [02:03] * Quits: tH (Rob@83.100.132.169) (Quit: ChatZilla 0.9.79-rdmsoft [XULRunner 1.8.0.9/2006120508])
- # [02:18] * Quits: Lachy (Lachlan@84.215.54.100) (Quit: Leaving)
- # [02:18] * Joins: Lachy (Lachlan@84.215.54.100)
- # [03:20] <Philip> Does HTML5 say that <meta name> is case-insensitive?
- # [03:55] <Hixie> html5's story on case sensitivity is a mess right now
- # [03:56] * Quits: jmb (jmb@152.78.68.189) (Ping timeout)
- # [03:56] * Joins: jmb (jmb@152.78.68.189)
- # [04:05] * Philip finds that 5% +/- 6% of Wordpress pages use <u>, and thinks he needs to stop using approximations that give a positive chance of there being negative pages
- # [04:11] <Philip> Hmm, 8% +/- 0% of all pages use <u>, but only 6% +/- 0% of pages with no meta-generator use <u>, which is quite a big difference
- # [10:14] * Quits: sbuluf (nvdbzm@200.49.132.118) (Quit: sbuluf)
- # [10:20] * Joins: ROBOd (robod@89.122.216.38)
- # [10:22] * Joins: anne-mac (annevk@86.90.70.28)
- # [10:39] * Joins: jgraham (james@86.7.213.72)
- # [10:51] * Quits: jgraham (james@86.7.213.72) (Quit: This computer has gone to sleep)
- # [10:54] * Joins: edas (edaspet@82.233.238.50)
- # [11:08] * Quits: olivier (ot@128.30.52.30) (Quit: Leaving)
- # [11:28] * Quits: anne-mac (annevk@86.90.70.28) (Ping timeout)
- # [11:33] * Quits: edas (edaspet@82.233.238.50) (Quit: Quitte)
- # [11:53] * Joins: nickshanks (nickshanks@86.10.70.24)
- # [12:00] * Joins: jgraham (james@86.7.213.72)
- # [12:01] * Quits: jgraham (james@86.7.213.72) (Quit: This computer has gone to sleep)
- # [12:06] * Joins: jgraham (james@86.7.213.72)
- # [12:06] * Quits: jgraham (james@86.7.213.72) (Quit: This computer has gone to sleep)
- # [12:29] * Quits: Lachy (Lachlan@84.215.54.100) (Quit: This computer has gone to sleep)
- # [12:31] <hsivonen> Philip: cheracter fixed in local copy. will push out later. thanks
- # [12:34] * Joins: tH_ (Rob@83.100.132.169)
- # [12:34] * tH_ is now known as tH
- # [13:29] * Joins: anne-mac (annevk@86.90.70.28)
- # [13:29] <anne-mac> xhtml.se is now webbstandard.se it seems
- # [13:39] <hsivonen> :-)
- # [14:00] * Quits: anne-mac (annevk@86.90.70.28) (Client exited)
- # [14:01] * Joins: anne-mac (annevk@86.90.70.28)
- # [14:03] <anne-mac> hmm, <cite> for ship names?
- # [14:11] <anne-mac> great, xlink:href is not supported on <style> by at least Opera and Safari
- # [14:25] * Joins: Lachy (Lachlan@84.215.54.100)
- # [14:27] * Quits: Lachy (Lachlan@84.215.54.100) (Quit: Leaving)
- # [14:27] * Joins: Lachy (Lachlan@84.215.54.100)
- # [14:35] <anne-mac> and for the SVG script element the type attribute is ignored by Opera and WebKit
- # [14:36] <anne-mac> (both do support l:href on it)
- # [14:40] <hsivonen> anne-mac: what's l:href?
- # [14:56] <anne-mac> oh, l bound to the XLink namespace
- # [14:57] * anne-mac usually uses one-letter abbreviations for prefixes in code
- # [14:57] <anne-mac> as in: XSLT > t, XLink > l, SVG > g, XHTML > h, ...
- # [14:59] * Quits: Lachy (Lachlan@84.215.54.100) (Ping timeout)
- # [15:04] * Joins: edas (edaspet@82.233.238.50)
- # [15:06] * Parts: anne-mac (annevk@86.90.70.28)
- # [15:08] * Joins: Lachy (Lachlan@84.215.54.100)
- # [15:09] * Joins: anne-mac (annevk@86.90.70.28)
- # [15:22] * Philip wishes he had paid more attention during statistics lessons
- # [15:25] <krijnh> So which generator is the best? :)
- # [15:27] <Philip> krijnh: I'm trying to avoid making any judgements of good vs bad - I just want to get the numbers right, and other people can argue about the interpretation of them :-)
- # [15:28] <krijnh> Yeah, but your opinion based on those numbers?
- # [15:28] <krijnh> If goodness can be measured in terms of <u> usage :)
- # [15:29] <Philip> The only thing that's safe to say is that FrontPage is by far the most common generator :-)
- # [15:29] * Quits: Lachy (Lachlan@84.215.54.100) (Quit: Leaving)
- # [15:29] * Joins: Lachy (Lachlan@84.215.54.100)
- # [15:30] <anne-mac> statistics on "generated" pages using <em>, <strong>, <blockquote>, etc. would also be good
- # [15:30] <Philip> It's not obvious whether the <u> usage numbers come from features of the editor itself, or from the people who tend to use them
- # [15:31] <krijnh> Should pages from a CMS always have a <meta name="generator"> ?
- # [15:31] <Philip> anne-mac: If you could expand on "etc.", it should be trivial to do that
- # [15:32] <anne-mac> <xxx align> <xxx style>
- # [15:33] <Philip> That's less trivial :-p
- # [15:33] <anne-mac> mostly <p align> and <div align>
- # [15:33] <anne-mac> and <span style> I guess
- # [15:33] <Philip> (since I'm just working from tag counts at the moment)
- # [15:33] <anne-mac> oh
- # [15:33] <anne-mac> k
- # [15:33] <anne-mac> can you do <p><b> ?
- # [15:33] <anne-mac> and <br><br> ?
- # [15:34] <anne-mac> guess not if you can just count
- # [15:35] <Philip> Those things probably aren't especially hard; I just need to modify the code that parses the pages, rather than the code that processes the output
- # [15:41] <Philip> anne-mac: http://philip.html5.org/data/generators.txt has some stuff
- # [15:43] <anne-mac> cool
- # [15:45] <hsivonen> Philip: is your analyzer framework public and open source?
- # [15:47] <hsivonen> Philip: do you have stats on the percentages of pages with a given generator?
- # [15:48] <Philip> hsivonen: No, but it could be if I stopped worrying about it being ugly and hacked-together
- # [15:48] <hsivonen> (Interestingy, Dreamweaver is so good that it doesn't need to advertise itself in its output)
- # [15:48] <Philip> (and the old bits tend to break whenever I add new things to it)
- # [15:49] <hsivonen> I wonder if it is possible to detect Dreamweaver output with good confidence
- # [15:49] <anne-mac> hsivonen, being sarcastic? :)
- # [15:49] <Philip> hsivonen: http://philip.html5.org/data/underline-generators.txt shows pages per generator
- # [15:49] <hsivonen> anne-mac: no
- # [15:49] <anne-mac> if Dreamweaver still outputs some the script crap it used to...
- # [15:49] <anne-mac> it should be pretty trivial
- # [15:50] <Philip> hsivonen: I see varieties Dreamweaver on 32 pages
- # [15:50] <Philip> (in the meta-generator)
- # [15:50] <hsivonen> Philip: versions older than 4?
- # [15:51] <hsivonen> I wonder what it means to have a WebSphere generator
- # [15:51] <Philip> hsivonen: 2, 3.0, 3.01, 4, MX, MX 2004
- # [15:51] <Philip> I don't know how many of these generator strings were hand-written
- # [15:51] <hsivonen> Philip: ooh. that's surprising
- # [15:52] <Philip> (There's a "Dreamweaver 3.0, Photoshop 6.0, ImageReady 3.0, Illustrator 9.0" and a "Dreamweaver, Flash" and a "dreamweaver // flash" and a "WebExpert,Dreamweaver,Fireworks" etc)
- # [15:53] <hsivonen> hmm. apparently people write all sorts of things in their HTML
- # [15:53] <Philip> hsivonen: http://philip.html5.org/data/dreamweaver-generators.txt
- # [15:53] <Philip> (...is the result from 'grep -i dreamweaver' on my parsed data)
- # [15:54] <anne-mac> <meta> has a uri= attribute? interesting
- # [15:54] <anne-mac> anyway, got to go
- # [15:54] <Philip> anne-mac: No, that's just from my parsed data output format
- # [15:55] <Philip> which happens to look a little bit like XHTML but isn't
- # [15:57] * Quits: Lachy (Lachlan@84.215.54.100) (Quit: Leaving)
- # [15:58] * Joins: Lachy (Lachlan@84.215.54.100)
- # [15:59] * Quits: edas (edaspet@82.233.238.50) (Quit: Quitte)
- # [16:02] * Quits: anne-mac (annevk@86.90.70.28) (Ping timeout)
- # [16:04] <hsivonen> Philip: what language is your framework written in?
- # [16:16] <Philip> hsivonen: Java for downloading and parsing pages and extracting the relevant bits of data, and Perl for summarising the data
- # [16:17] <hsivonen> Philip: ok
- # [16:17] <hsivonen> Philip: have you been analyzing parse errors?
- # [16:18] <hsivonen> the Validator.nu parser has a specific analysis mode that allows the doctype error to be suppressed
- # [16:18] <Philip> hsivonen: I've been ignoring errors for now, since I didn't know I could catch them and do anything useful
- # [16:19] * Joins: gsnedders (gsnedders@86.138.198.209)
- # [16:22] <hsivonen> Philip: well, depends on whether finding out what pages parse without errors is useful
- # [16:23] * Joins: billmason (billmason@69.30.57.156)
- # [16:30] <Philip> hsivonen: The error message "This document is not mappable to XML 1.0 without data loss to “--” in a comment." should probably say "...loss due to..."
- # [16:30] <Philip> or maybe it shouldn't, since it sort of makes sense anyway
- # [16:30] <hsivonen> Philip: thanks
- # [16:30] <Philip> but it sounds a bit odd
- # [16:30] <hsivonen> Philip: I will fix in svn but won't spin a release for this
- # [16:32] <hsivonen> fixed in svn
- # [16:35] * Joins: dbaron (dbaron@72.94.185.124)
- # [16:47] <Philip> http://philip.html5.org/data/errors.html
- # [17:11] <hsivonen> Philip: wow. great to see 30% of pages examined are free of parse errors
- # [17:11] <hsivonen> Philip: that's *way* more than I expected
- # [17:11] <hsivonen> and doesn't corroborate Hixie's data
- # [17:12] <Philip> hsivonen: This is ignoring ~50% of pages that have no doctype, which ought to count as an error
- # [17:12] <hsivonen> Philip: surely Hixie didn't count those as errors?
- # [17:13] <Philip> Hmm, no idea
- # [17:13] <hsivonen> Interesting that </embed> is so common
- # [17:13] <Philip> http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/index#parse-errors says 55% have no errors when just running the tokeniser
- # [17:14] <hsivonen> yeah, most of the errors in your latest data are from the tree builder
- # [17:15] * Philip updates http://philip.html5.org/data/errors.html so it's not just the top 20
- # [17:15] <Philip> (since I had forgotten I had that limit in there...)
- # [17:22] * Joins: edas (edaspet@82.233.238.50)
- # [18:06] * Quits: billmason (billmason@69.30.57.156) (Connection reset by peer)
- # [18:10] * Joins: billmason (billmason@69.30.57.156)
- # [18:12] * Quits: nickshanks (nickshanks@86.10.70.24) (Ping timeout)
- # [18:12] * Joins: Sander (svl@86.87.68.167)
- # [18:21] * Quits: dbaron (dbaron@72.94.185.124) (Ping timeout)
- # [18:38] * Joins: dbaron (dbaron@72.94.185.124)
- # [18:41] * Quits: edas (edaspet@82.233.238.50) (Quit: Quitte)
- # [18:42] * Joins: Fabulari (fabulari@66.195.32.2)
- # [18:42] * Quits: Fabulari (fabulari@66.195.32.2) (Quit: Leaving)
- # [19:10] * Joins: nickshanks (nickshanks@86.10.70.24)
- # [19:11] * Quits: Sander (svl@86.87.68.167) (Quit: And back he spurred like a madman, shrieking a curse to the sky.)
- # [20:20] * Quits: billmason (billmason@69.30.57.156) (Quit: .)
- # [20:33] <Hixie> Philip: i had two numbers, one ignored doctype and /> errors, and was in the 70s, and one didn't, and was in the 90s
- # [20:34] <Hixie> (percentages of pages with errors)
- # [20:36] <Philip> Hixie: Ah, thanks - that "70s" sounds close enough to what I saw, which is good
- # [20:53] * Quits: trackbot-ng (trackbot-n@128.30.52.30) (Client exited)
- # [20:57] * Joins: trackbot-ng (trackbot-n@128.30.52.30)
- # [20:57] * trackbot-ng HTML Issue Tracking http://www.w3.org/html/wg/tracker/
- # [21:30] <nickshanks> Philip: how many people use <image> when they mean <img> ?
- # [21:32] <Philip> nickshanks: http://canvex.lazyilluminati.com/survey/2007-07-17/analyse.cgi/pages/tag/image shows it on 15 out of 7739 pages
- # [21:32] <Philip> (All of those occurrences are <image src> so they're trying to use it like <img>)
- # [21:33] <nickshanks> i would like image to be introduced as a container for fallback which is restricted to image/* MIME types (like the audio and video elements), but am repeatedly told that "browsers treat image the same as img". nobody has provided any evidence that sites out there are using it on pages in standards mode (image could still be treated as img in quirks mode)
- # [21:34] <nickshanks> hmm, that makes ~0.2%
- # [21:35] <Philip> nickshanks: It seems most browser developers want to minimise the differences between quirks and standards (and almost standards and IE8-standards), so they're not going to add new differences
- # [21:35] <nickshanks> the IMDB is three of those. if they fix their /name/ pages, that would be 0.15%
- # [21:36] <nickshanks> even though img alt="..." is acknowledged as bad?
- # [21:38] <Philip> http://www.powderspringsfumc.org/ is almost-standards mode and uses <image>
- # [21:39] <nickshanks> well the odd page here and there can be fixed
- # [21:39] <nickshanks> 1700 isn't really a big enough sample
- # [21:40] <Philip> (7739, not 1700)
- # [21:40] <nickshanks> yeah, not thinking
- # [21:41] <Philip> http://www.playmobil.de/on/demandware.store/Sites-DE-Site/de_DE/Link-Page?id=START_ZIRNDORF is another (not from that list)
- # [21:41] <Philip> http://www.tvguide.com/movies/b-monkey/132232
- # [21:42] <nickshanks> active sites are not the problem, it's ones that are no longer updated which are not likely to be fixed
- # [21:44] <Philip> If it's around 0.01% that are using <image> in standards mode (which roughly fits the available data), then that's millions of pages across the whole web
- # [21:44] <Philip> so it seems quite expensive to fix them
- # [21:44] <nickshanks> well i'm going to start y emailing IMDB
- # [21:44] <nickshanks> *by
- # [21:46] <Philip> (and it's tens of millions when counting quirks-mode pages too)
- # [21:47] <Philip> (though obviously some sites (like IMDB) make up a large number of those pages, so a lot can be fixed at once)
- # [21:49] <nickshanks> apart from the problem of extant usage, would you say an element specifically for image/*, which can contain <source> elements and/or fallback, as a good thing?
- # [21:50] <Philip> What would the <source> elements be useful for?
- # [21:51] <Philip> Video codecs are a horrible mess, but images seem to work well enough already - you use PNG or JPEG depending on the image's content, and that's it, so you don't need alternate source formats
- # [21:52] <nickshanks> choosing between MIME types. people seem to agree sending every type you support is a non‐workable idea these days, but choosing between SVG Animation and GIF89a for an animated image might be possible
- # [21:53] <nickshanks> mostly i want it for better accessibility reasons though
- # [21:55] <Philip> <object><img></object> works for that particular case already - there is no scripted API for images so you don't get problems like how to work out which element to call functions on (whereas you do get those problems with <video> since it's often scripted)
- # [21:55] * Joins: kingryan (kingryan@66.92.2.56)
- # [21:56] <nickshanks> yeah, but object is rather general. the point was to create a specific element (so that image‐related actions can be performed on it, like they are for img), things that object wouldn't provide, like set to desktop image
- # [21:57] <Hixie> why can't <object> provide that?
- # [21:59] <Philip> hsivonen: http://validator.nu/?doc=http://www.powderspringsfumc.org/ says "Bad value JavaScript for attribute language on element script: Subtype missing." for <script language=JavaScript>, which seems wrong since it's not meant to have type/subtype in there
- # [22:23] * Joins: jane (j@76.170.65.146)
- # [23:13] * Joins: sbuluf (vtlvn@200.49.132.110)
- # Session Close: Tue Jan 01 00:00:00 2008
The end :)