| 01:20 | <heycam> | what's the best way to specify fallback for mathml content in an html 5 document (say, a png of the math)? |
| 03:06 | <jcranmer> | Hixie: can you convice the Google Video/YouTube people to use <video> ? |
| 03:09 | <jcranmer> | tell them they're evil bastards until they do so |
| 03:09 | <jcgregorio> | jcranmer: http://www.youtube.com/html5 |
| 03:10 | <jcgregorio> | that was demoed at Google IO |
| 03:10 | <jcranmer> | can it load ALL videos in there? |
| 03:12 | <jcranmer> | that seems to break in FF 3.6a1pre |
| 03:15 | <kinetik> | jcranmer: it's serving h.264 |
| 03:44 | <shepazu> | http://www.google.com/squared/search?q=races |
| 04:53 | <roc> | that's cool, but check this out: |
| 04:53 | <roc> | http://www.google.com/squared/search?q=programming+languages |
| 04:53 | <roc> | why is there a "death" column? |
| 04:54 | <roc> | with very ... strange ... values |
| 05:09 | <MikeSmith> | <cite style="font-style: normal">Lincoln</cite> |
| 05:09 | <MikeSmith> | nice |
| 06:03 | <hsivonen> | jgraham: whoa. sorry. I just noticed that the test case didn't match Safari and Opera and checked that <frame> and </frame> don't set frameset-o to not ok |
| 06:12 | <hobo> | What is going to happen with the onbeforeunload and onunload events, Opera doesn't acknowledge them |
| 06:12 | <hobo> | does whatwg deal with that? |
| 06:14 | <roc> | I find it hard to believe Opera doesn't support them |
| 06:16 | <hobo> | i dunno my test keeps failing |
| 06:16 | <Hixie> | the html5 spec does define those events, yes |
| 06:17 | <hobo> | The previous specs included them as well right? |
| 06:17 | <hobo> | Opera just does whatever? |
| 06:21 | <Hixie> | i don't think they've ever been defined |
| 06:21 | <Hixie> | some spec might have mentioned them though i doubt it |
| 08:10 | <annevk42> | Opera doesn't support those events |
| 08:10 | <annevk42> | (yet) |
| 08:12 | <annevk42> | heycam, no fallback is needed for MathML; it just needs to be supported |
| 08:12 | <heycam> | annevk42, ? |
| 08:12 | <heycam> | what if i want to target browsers that do support mathml in text/html as well as those that don't? |
| 08:13 | <hsivonen> | heycam: tough luck |
| 08:14 | <hsivonen> | heycam: actually, there's a way |
| 08:15 | <hsivonen> | heycam: 1) make all your MathML text nodes into CDATA sections |
| 08:16 | <hsivonen> | heycam: 2) stick and <img> with the image of the formula into annotation-xml |
| 08:16 | <hsivonen> | heycam: 3) hide annotation-xml using a selector that has an explicit namespace so it only matches if the elements really got assigned to the MathML namespace |
| 08:17 | <heycam> | that's so crazy it just might work |
| 08:17 | <Philip`> | What if a browser implements the HTML5 parsing algorithm but not MathML rendering? |
| 08:18 | <heycam> | in reality, pubrules won't let me use mathml in text/html anyway so the question is moot (for my purpose) |
| 08:18 | <annevk42> | Philip`, browsers have bugs |
| 08:18 | <heycam> | annevk42, likes |
| 08:18 | <heycam> | erm, lies! |
| 08:18 | <hsivonen> | Philip`: oh, right. are there MathML elements that in the normal course of MathML rendering hide <annotation> or <annotation-xml>? |
| 08:18 | <annevk42> | film at eleven |
| 08:18 | <hsivonen> | Philip`: at least <annotation-xml> as child of <semantics> isn't hidden automatically, IIRC |
| 08:19 | <annevk42> | heycam, you wish :p |
| 08:19 | <hsivonen> | heycam: anyway, I recommend playing with <annotation> and <annotation-xml> |
| 08:19 | <heycam> | hsivonen, ok |
| 08:20 | <heycam> | are there any browsers that support mathml in text/html yet? |
| 08:20 | <hsivonen> | heycam: there are my tryserver builds |
| 08:20 | <heycam> | k |
| 08:20 | <hsivonen> | heycam: SVG is easier, because you can count on <desc> getting hidden by the UA when SVG is supported |
| 08:21 | <heycam> | put fallback content inside <desc>? |
| 08:21 | <hsivonen> | heycam: right |
| 08:21 | <heycam> | it is a kind of description i suppose |
| 08:42 | <annevk42> | http://sandfly.net.nz/blog/2009/05/the-html5-video-tags-fatal-flaw/#comment-463 -- very typical indeed :p |
| 08:46 | <hsivonen> | sigh. the next commenter blames Mozilla instead of blaming MPEG-LA |
| 08:46 | <heycam> | oh my goodness is that blog the shape of things to come (re @font-face usage)? |
| 08:47 | <annevk42> | isn't it awesome? |
| 08:47 | hsivonen | notes that howcome promoted Theora before Mozilla was visibly behind Theora |
| 08:47 | <annevk42> | @font-face is the new <marquee> + animated GIF |
| 08:47 | <hsivonen> | (not to suggest blaming howcome. MPEG-LA should be blamed.) |
| 08:48 | <hsivonen> | annevk42: I haven't seen any animated @font-face yet |
| 08:48 | <annevk42> | lol |
| 08:50 | <heycam> | and of course you know that svg fonts can have videos and animated paths and whatnot inside glyphs |
| 08:51 | <annevk42> | not really, but I do now... |
| 08:53 | <Philip`> | It should only take a few years for things to settle down, so that people only use @font-face when it's a sensible design decision rather than using it simply to be cutting-edge |
| 08:54 | <Philip`> | and in the meantime I hope Opera adds a 'disable font download' option |
| 08:55 | heycam | imagines an svg font whose glyphs have child foreignObjects with marquees in them |
| 08:55 | jwalden | ponders self-referential SVG fonts |
| 08:56 | <Philip`> | The advantage of SVG fonts is that they're weird and complicated and nobody knows how to use them, so they won't get abused |
| 08:57 | <Philip`> | whereas anyone can download a TTF file from some random 1001 FREE FONTS! site and stick it on their own page |
| 08:58 | <jwalden> | frankly, I don't understand the use case for generic SVG font embedding, outside of use within SVG files themselves |
| 08:58 | <jwalden> | a large amount of complexity for not that much gain over font files themselves |
| 08:59 | <jwalden> | "I've always wanted to animate my font's glyphs!" |
| 09:22 | <othermaciej> | jwalden: SVG fonts do have some capabilities that are not available in typical outlined font formats |
| 09:22 | <othermaciej> | jwalden: such as multicolored characters |
| 09:22 | <othermaciej> | whether anyone should use such capabilities is beyond me |
| 09:22 | <jwalden> | wait, do fonts have intrinsic colors? |
| 09:22 | <othermaciej> | they don't have to |
| 09:22 | <othermaciej> | but they can |
| 09:22 | <jwalden> | what does that even mean in conjunction with CSS? |
| 09:23 | <jwalden> | castles in the sky |
| 09:23 | <othermaciej> | the outline definition doesn't by default define a color, in which case the CSS color is followed |
| 09:23 | <othermaciej> | but you can draw part or all of the character with an explicit color |
| 09:23 | <jwalden> | gag gag gag gag gag |
| 09:24 | jwalden | wonders what other absurd bells and whistles exist |
| 09:24 | <othermaciej> | also, for very one-off custom letterforms for headings, it may be easier to draw them with SVG than to create a throwaway TTF |
| 09:24 | <othermaciej> | but still I think TrueType Web fonts are much more useful |
| 09:24 | <jwalden> | it still seems unlikely that conversion tools are so difficult to procure or so impossible to use to validate the concept |
| 09:25 | <Philip`> | For custom headings you might as well just use an SVG file, rather than create a font and use it in @font-face |
| 09:25 | <jwalden> | although I suppose you do need SVG for your glyphs to be able to have realtime socket access to change in response to external network inputs |
| 09:26 | <othermaciej> | I don't think that SVG fonts were necessarily created with well thought out use cases and as a meaningful alternative to TrueType |
| 09:26 | <othermaciej> | it just seemed like something that an output format for Adobe Illustrator should have |
| 09:27 | <jwalden> | ah, back to the famed Macromedia hijinks again |
| 09:28 | <Philip`> | Macromedia? |
| 09:28 | Philip` | thought it was always an Adobe thing |
| 09:30 | <othermaciej> | SVG was Adobe's pet project before they formed Macrodobe and decided SVG should die |
| 09:41 | <zcorpan> | Philip`: you can't select the text if you just use an svg file, though |
| 09:42 | <zcorpan> | although i probably wouldn't bother creating an svg font as opposed to just use an svg file |
| 09:47 | <hsivonen> | othermaciej: my impression of the history is that having Illustrator export normal hinted OpenType fonts along with SVG would have been a horrible pirate vector |
| 09:47 | <hsivonen> | othermaciej: so SVG needed something that you can't drop in your system font folder |
| 09:48 | <hsivonen> | othermaciej: and then it just ballooned from there |
| 09:48 | <othermaciej> | hsivonen: I did not at the time follow nor have I since studied the actual history |
| 09:48 | <othermaciej> | I assumed it was part of SVG's desire to do everything in SVG |
| 09:48 | <hsivonen> | othermaciej: the above is speculation from few data points |
| 09:49 | <hsivonen> | I haven't done actual mailing list archeology |
| 09:49 | <roc> | does anyone actually support the full SVG font madness with arbitrary SVG content in glyphs? |
| 09:49 | <roc> | with animations and all? |
| 09:49 | <roc> | and looking at that blog, is it too late to drop support for font-face? |
| 09:49 | <othermaciej> | WebKit supports arbitrary content in glyphs, though I don't know if declarative animations will work |
| 09:50 | <hsivonen> | othermaciej: I do note, however, that SVG (1.1 IIRC) spec has an unusual field of use prohibition for fonts |
| 09:50 | <hsivonen> | the most interesting Process question is if the PP applies if you violate the field of use prohibition |
| 09:50 | <othermaciej> | hsivonen: field of use prohibition seems like a bizzarre thing for a W3C spec to contain |
| 09:51 | <roc> | with the complete <use>-alike style inheritance, it seems like an implementation nightmare |
| 09:52 | <hsivonen> | roc: do you plan on implementing it all anyway? |
| 09:52 | <jgraham> | Hey the fonts on that blog are useful. They tell me that I have better things to do than read the blog without even having to look at the words. Such is the amazing ability of typography |
| 09:52 | <roc> | I'm no sure |
| 09:53 | <roc> | I think to do it "right" we'd have to clone the SVG content into anonymous children of the text |
| 09:53 | <roc> | or something equally crazy |
| 09:54 | <roc> | for non-HTML text this is really crazy |
| 10:09 | <othermaciej> | I think we only allow use of SVG fonts from SVG - not 100% sure though |
| 10:09 | <jgraham> | Philip`: BTW hg rebase looks like it might be useful for your source control problem |
| 10:09 | <othermaciej> | using SVG fonts also makes you take a slower and more memory-intensive code path |
| 10:09 | jgraham | thinks he will want to solave a similar problem so is quite interested in solutions |
| 10:12 | <Philip`> | "hg: unknown command 'rebase'" |
| 10:12 | <jgraham> | Philip`: It hould be packaged with hg > 1.1 I think, but you may have to enable it |
| 10:13 | <Philip`> | Ah, looks like an extension thingy |
| 10:14 | <Philip`> | I found that pulling from an SVN-converted-to-Hg repository didn't work anyway, because it pulled everything into the root directory, whereas I wanted it moved into some subdirectory instead |
| 10:14 | <Philip`> | so I just gave up and copied the files with no history |
| 10:14 | <othermaciej> | good god, that font is truly fucking terrible |
| 10:14 | <Philip`> | and if I want to patch the originals then I guess I can figure out some way to do that in the future |
| 10:16 | <Philip`> | http://sandfly.net.nz/blog/2009/02/a-better-handwriting-font-with-yourfontscom/ - apparently he is "very happy with the result" |
| 10:17 | Philip` | doesn't think it really aids in readability much |
| 10:17 | <othermaciej> | I am curious what is really up with Google's H.264 license by the way; if they can really legitimately ship ffmpeg with H.264 support, it seems like Mozilla could too if it coughed up the money |
| 10:17 | <othermaciej> | I'm not really sure if their explanation makes sense or not though |
| 10:18 | <hsivonen> | othermaciej: it wouldn't work for downstream distributors |
| 10:19 | <othermaciej> | is Chrome (as opposed to Chromium) redistributable? |
| 10:20 | <annevk42> | I think so |
| 10:20 | <hsivonen> | othermaciej: I meant not only redistributable as binary but redistributable as in compilining your own build and shipping that |
| 10:20 | <hsivonen> | *compiling |
| 10:20 | <annevk42> | I think the point is that the patent license is for Chrome as a whole, not for usage of FFmpeg |
| 10:21 | <othermaciej> | I don't know the exact nature of Google's patent licenses, but it's unclear to me if source vs. binary would be a critical legal issue |
| 10:21 | <roc> | I'm waiting for our lawyers to reach some conclusions before I comment further |
| 10:23 | <othermaciej> | I don't think you can build your own Chrome (as opposed to Chromium) but that's mainly due to trademarks |
| 10:23 | <roc> | but even if Google's interpretation is completely right, I think it subverts the intent of the LGPL |
| 10:23 | <othermaciej> | I was surprised by their interpretation, such a thing had not occurred to me |
| 10:23 | <othermaciej> | but the legal issues are too subtle for me to have a useful opinion on whether is right |
| 10:25 | <roc> | We've fought pretty hard to keep Web standards royalty free |
| 10:25 | <roc> | I'm not in a hurry to abandon that fight now |
| 10:25 | <Philip`> | One email from a Google person made it sound like it was a loophole that was closed in the LGPL3 |
| 10:25 | <othermaciej> | I wish video standards had the same IP norms as Web standards |
| 10:26 | <hsivonen> | it seems it's that it's possible to be open-source in terms of copyright while still being freedom-deprived for downstream (e.g. by using BSD on code that infringes on a 3rd party patent) |
| 10:26 | <othermaciej> | even more, I wish algorithms and algorithmic techniques could not be patented |
| 10:26 | <roc> | absolutely |
| 10:27 | <othermaciej> | ffmpeg itself seems to be a case of open source software that is freedom-deprived upstream |
| 10:27 | <othermaciej> | since it infringes on third-party patents and doesn't grant you a license |
| 10:27 | <othermaciej> | er, downstream |
| 10:27 | <roc> | yeah but at least if you download ffmpeg from them, you get the same rights they have |
| 10:27 | <othermaciej> | and, in the US and other countries where software patents are valid |
| 10:27 | <roc> | when you download it from Google, you don't |
| 10:28 | <roc> | The open source community was up in arms when Novell appeared to have struck a patent license deal with Microsoft |
| 10:29 | <othermaciej> | GPL certainly seems less ambiguous about this than LGPL |
| 10:29 | <othermaciej> | since you can't link with non-GPL code which has a hypothetical grant of rights |
| 10:30 | <othermaciej> | I also believe the Novell move was seen as conceding Microsoft had legitimate patent claims against Linux, which was something that hadn't been convincingly demonstratd |
| 10:30 | <othermaciej> | afaik no one disputes that MPEG-LA has legitimate patent claims against ffmpeg (at least in countries where it holds patents) |
| 10:30 | <roc> | it's true that the GPL and the LGPL are different here |
| 10:31 | <roc> | I don't know why that is |
| 10:31 | <roc> | but I always thought they were in the same spirit |
| 10:31 | <roc> | modulo the virality |
| 10:31 | <othermaciej> | patents in Software and especially on open standards are kind of bullshit |
| 10:31 | hsivonen | wonders if the FSF is going to comment on this whole thing |
| 10:32 | <othermaciej> | I would be curious what they think, or a group like the SFLC |
| 10:32 | <roc> | does anyone really think that the Linux kernel does not infringe Microsoft's patents? |
| 10:32 | <roc> | the kernel is big and complicated |
| 10:32 | <roc> | Microsoft's patents are many and broad |
| 10:32 | <roc> | of course they intersect |
| 10:33 | <othermaciej> | it does make one wonder why Microsoft won't cite specifics in public |
| 10:39 | <roc> | I've never understood why Microsoft hasn't gone on an all-out patent attack on free software |
| 10:39 | <roc> | I would have |
| 10:39 | <Philip`> | Maybe they're not evil? |
| 10:40 | <othermaciej> | maybe they really don't have any patents that are all that good |
| 10:40 | <roc> | neither of those options seem credible |
| 10:40 | <jgraham> | maybe they don't want IBM to sue them? |
| 10:42 | <roc> | seems like a risk worth taking, they can always settle with IBM for the right amount of money |
| 10:43 | <othermaciej> | maybe they fear antitrust litigation |
| 10:43 | <jgraham> | Maybe they fear precipitating change in the whole patent system |
| 10:43 | <othermaciej> | though, I heard a recent story to the effect that Microsoft was threatening and pressuring WinMobile handset vendors who bundled a non-Pocket IE browser |
| 10:44 | <roc> | antitrust maybe... |
| 10:44 | <roc> | for a while I thought they might be holding fire so that the EU didn't get spooked away from formalizing software patents |
| 10:44 | <roc> | but that receded |
| 10:45 | <roc> | my brother is a patent attourney, a partner in the law firm that handles Microsoft's NZ work |
| 10:45 | <roc> | you can imagine what our family gatherings are like |
| 10:45 | <othermaciej> | that must be... interesting |
| 12:35 | <hsivonen> | hmm. it seems we may be lacking test case coverage for EOF inside doctype in each possible position |
| 12:36 | <jgraham> | hsivonen: In tree construction or in tokenizer? |
| 12:36 | <jgraham> | In tokenizer we will lack coverage for things that changed since Philip` made all those tests |
| 12:36 | <hsivonen> | jgraham: tokenizer |
| 12:37 | <jgraham> | In tree-construction we lack tests everywhere |
| 12:37 | <jgraham> | e.g. html5lib failed 0 tests due to not implementing the framesetOK flag at all |
| 12:38 | Philip` | really ought to update his tokeniser / test generator |
| 12:39 | <jgraham> | Philip`: Where is the code? I have no intentions to make changes I just want to know if it makes more sense to me than last time I looked at it... |
| 12:40 | <Philip`> | jgraham: There's some version at http://canvex.lazyilluminati.com/svn/tokeniser/ which might (or might not) be close to the latest version that exists |
| 12:41 | <Philip`> | test_gen.ml is the bit that generates tests, based on the interpreter in tokeniser_interp.ml which executes the algorithm in tokeniser_spec.ml |
| 12:45 | <Philip`> | Hmm, I'm not entirely certain that any of it makes sense to me |
| 12:46 | <hsivonen> | the tokenizer alone seems to have over 50 different errors and warnings |
| 12:47 | <Philip`> | I think test_gen basically has a list of test strings (initially [""]), then for each one it looks at the final interpreter state (just before it would consume an EOF) and finds all the 'interesting' characters for that state, and then constructs a new set of strings, and then trims out ones that end up in duplicate states of others, and then repeats a few times |
| 12:48 | <Philip`> | hsivonen: What kinds of errors? |
| 12:49 | <hsivonen> | Philip`: I mean the number of different error messages the V.nu tokenizer can emit |
| 12:50 | <Philip`> | hsivonen: Oh, right |
| 13:00 | <jgraham> | Is the live dom viewer broken in IE8-in-windows 7 for anyone else or is it just me? |
| 13:00 | <Philip`> | "this is clearly going to be a long discussion" - surely not! |
| 13:00 | <Philip`> | jgraham: Broken in what ways? |
| 13:00 | <jgraham> | Specifically using w() to write to the output |
| 13:00 | <Philip`> | jgraham: (e.g. some way other than the XSS filter breaking it?) |
| 13:00 | <Philip`> | Oh |
| 13:00 | <jgraham> | (it orks in compat view but that is rather useless) |
| 13:01 | Philip` | doesn't remember whether w() worked in IE8-in-Vista |
| 13:39 | <hsivonen> | maikmerten: does Cortado provide enough scripting surface to JScript to be able to support the HTML5 video API in IE? |
| 13:40 | <maikmerten> | hsivonen, actually right now I'm not 100% sure the IE-Java scripting bridge works at all ;-) |
| 13:40 | <maikmerten> | but anyway, right now Cortado is just exposing bare essentials somewhat HTML5-compliant |
| 13:41 | <maikmerten> | no events, no fancy stuff, mostly just play and pause and a bit of broken seeking |
| 13:42 | <maikmerten> | does IE still support ActiveX? If it does I guess the best idea may be to develop an ActiveX-thingie 100% targeted at doing HTML5-video in IE |
| 13:42 | <hsivonen> | maikmerten: is the IE-Java bridge a Cortado thing or something in IE or the Sun plug-in? |
| 13:42 | <hsivonen> | maikmerten: any plans to expose scripting API in cortado to support HTML5 emulation shims? |
| 13:43 | <hsivonen> | maikmerten: can sites easily autoinstall ActiveX anymore? |
| 13:44 | <maikmerten> | hsivonen, I assume in worst case it would be the usual "click yes to install this security hole" thingie |
| 13:44 | <maikmerten> | hsivonen, currently Cortado just declares some methods/fields public in its main applet class |
| 13:45 | <maikmerten> | there's no bridge whatsoever to e.g. do events |
| 13:45 | <maikmerten> | and I've found Java <-> browser communication to be rather unreliable |
| 13:45 | <hsivonen> | I wonder if having a JS shim doing polling from timeout in order to emulate events would kill perf |
| 13:45 | <maikmerten> | (browsers/applets hanging and stuff) |
| 13:46 | <hsivonen> | Well, I hope someone writes a Theora ActiveX control for IE with an accompanying JS shim |
| 13:46 | <maikmerten> | hsivonen, here's a demo demonstrating one use case for the (limited) HTML5 media API support in Cortado: http://people.xiph.org/~oggk/elephants_dream/elephantsdream.html |
| 13:47 | <maikmerten> | hmmm... actually, no, doesn't demonstrate that really |
| 13:48 | <maikmerten> | (I supposed it would read the time to display a fitting subtitle with DOM-manipulation, but it's just "all Java subtitles") |
| 13:49 | <maikmerten> | hsivonen, another possibility would perhaps be to do a Vorbis/Theora player in Flash, but I have no idea how performant that would be |
| 13:50 | <othermaciej> | probably not very |
| 13:50 | <othermaciej> | the Flash H.264 codec is native code, and even that uses shocking amounts of CPU |
| 13:50 | <maikmerten> | right. The interesting question would be "how much slower than Java is it" ;-) |
| 13:51 | <othermaciej> | I'd expect ActionScript is a lot slower than Java |
| 13:51 | <maikmerten> | (after all Java *is* fast enough to decode video, audio and do all the scaling for ~ SD video on common rigs) |
| 13:51 | <othermaciej> | it's actually slower than the fastest current JS implementations, for many kinds of code |
| 13:51 | <maikmerten> | well, it depends on how effective their JIT is - but getting that fast is easier on statically typed languages... |
| 13:52 | <hsivonen> | cool. the applet loads reasonably fast on the elephants dream demo |
| 13:52 | <hsivonen> | much faster than a random JavaFX demo |
| 13:53 | <maikmerten> | http://people.xiph.org/~maikmerten/demos/bigbuckbunny-applet.html <-- another demo with a higher resolution video |
| 13:53 | <maikmerten> | (no fancy subtitles, though) |
| 13:53 | <MikeSmith> | I think a significant blocking issue with codecs will always remain what is implemented in hardware level on mobile devices |
| 13:54 | <maikmerten> | that indeed is an issue - and to make matters worse this is a moving target (quite literally, too) |
| 13:54 | <jgraham> | MikeSmith: Alays is a long time |
| 13:54 | <jgraham> | *Always |
| 13:54 | <MikeSmith> | well, in general, implementing XYZ in flash does not solve problems for mobile handsets |
| 13:54 | <hsivonen> | maikmerten: your demo showed an unverifiable signature warning in IE8/Java6 |
| 13:55 | <MikeSmith> | for browsing from mobile devices, I mean |
| 13:55 | <maikmerten> | hsivonen, aye, it's a signed applet with a Xiph.org certificate without official blessing |
| 13:55 | <maikmerten> | downside: User nag, upside "does play remote video" |
| 13:56 | <MikeSmith> | jgraham: "always" meaning until CPU and RAM resources on mobile handsets approach same level as PCs |
| 13:56 | <hsivonen> | maikmerten: yeah, the actual video performs well |
| 13:57 | <maikmerten> | MikeSmith, one would have to have a close look on what mobile devices are relevant, what hardware capabilities they have (hardwired decoders like in 1997 or programmable DSP stuff) and (in case DSPs are common) get DSP implementations done |
| 13:57 | <hsivonen> | this stuff really needs to be packaged for dummies so that one can use one script in head and then <video src=foo.ogg> in body and have things Just Work |
| 13:57 | <maikmerten> | hsivonen, that would be http://people.xiph.org/~maikmerten/demos/bigbuckbunny.html ;-) |
| 13:58 | <maikmerten> | hsivonen, (advanced version of this at http://www.celt-codec.org/presentations/ ) |
| 14:01 | <maikmerten> | anyway, as for making-IE-do-HTML5-media I guess something slammed together using liboggplay and and an additional DirectShow hook (for additional codec coverage) may be best |
| 14:02 | <maikmerten> | I develop Java software every day, I'm pretty fond of Java - but *applets* are something always giving me a "don't use unless necessary"-feeling |
| 14:03 | <maikmerten> | after all the Java plugin used to be a great annoyance to basically everyone (slow, crashy) and the from-scratch-new plugin isn't thaaat common, I'd guess |
| 14:03 | <hsivonen> | maikmerten: ooh. excellent. re: celt-codec video |
| 14:04 | <hsivonen> | I like Java, too, but I browse the Web without applets |
| 14:04 | <hsivonen> | maikmerten: you should do more marketing for the script |
| 14:05 | <hsivonen> | maikmerten: thanks. I need to study the script some more |
| 14:05 | <maikmerten> | hsivonen, I hope we can put author's instructions of how to use that sort of scripting into the support resources of HTML5-media supporting browser vendors |
| 14:06 | <maikmerten> | hsivonen, be warned, *I* initially wrote that script, so parts are crude and DOM-API usage at worst |
| 14:07 | <maikmerten> | hsivonen, (gmaxwell from Wikimedia cleaned it somewhat, so it still may give pain but the bleeding is gone) |
| 14:11 | <maikmerten> | hmmm... anybody from Chrome/Chromium here? |
| 14:11 | <maikmerten> | apparently Chrome (dev channel build) exactly duplicates this Firefox bug: https://bugzilla.mozilla.org/show_bug.cgi?id=480058 |
| 14:14 | <gsnedders> | Yay! Swedish accommodation contracts that mean nothing to me! |
| 14:14 | <jgraham> | gsnedders: Me neither. And I signed two |
| 14:15 | <gsnedders> | "Since I don't know how much swedish you know…" |
| 14:15 | <hsivonen> | duplicating a bug exactly is called interop :-) |
| 14:17 | <gsnedders> | jgraham: one of you and zcorpan lives west from the office, right? |
| 14:17 | <MikeSmith> | bugwards compatibility |
| 14:19 | <gsnedders> | 1.6km more or less |
| 14:20 | <MikeSmith> | hmm, is "bugwards compatibility" not a familiar term? |
| 14:20 | <MikeSmith> | (google seems to suggest it's not) |
| 14:20 | <maikmerten> | hsivonen, okay, I just installed an ActiveX control in IE (Bitdefender online virus scanner) to see what the experience is like (the wonders of having Windows caged in a virtual machine) |
| 14:21 | <gsnedders> | Also, apparently the flat comes with a bunch of things with Swedish names :P |
| 14:21 | <maikmerten> | hsivonen, it involves clicking "yup, install that thing" twice, that's it |
| 14:21 | <maikmerten> | (first is a yellow "web page wants to install bla bla bla" in the web-browser - not unlike the Mozilla pop-up blocker) |
| 14:22 | <hsivonen> | maikmerten: ok |
| 14:22 | <maikmerten> | (when allowing this it'll ask again on installation) |
| 14:23 | <maikmerten> | the program keeps installed even after reboots, so I guess this *would* be a two-click solution for IE to install basically anything |
| 14:26 | <maikmerten> | however, no applet (that for sure) or ActiveX control (well, that one could exchange the rendering engine ;-) ) will be able to e.g. properly layer HTML content on top of the video, do transparency and whatnot |
| 14:26 | <maikmerten> | but then again there's a chance most people want "just video, thank you" |
| 14:29 | <Philip`> | I think most people really do want their videos to be translucent and embedded in a rotating SVG triangle |
| 14:30 | <hsivonen> | can laying an iframe on top of an ActiveX control punch a rectangular hole in the rendering of the ActiveX control? |
| 14:30 | <Philip`> | otherwise they'd still be happy with Flash |
| 14:30 | <maikmerten> | (hmmm... this whole pimp-your-browser talk makes me think writing a (signed) applet downloading and installing XiphQT on Safari may be something workable) |
| 14:33 | <hsivonen> | maikmerten: except Mac users run with Java off |
| 14:33 | <maikmerten> | hsivonen, by default? |
| 14:33 | <hsivonen> | maikmerten: by experience of instability and slowness and now due to unpatched security holes |
| 14:33 | <maikmerten> | hsivonen, I don't believe in the casual Mac user noticing how badly their Java is maintained ;-) |
| 14:35 | <maikmerten> | (this may be a vastly inaccurate opinion based on no numbers whatsoever) |
| 15:00 | <gsnedders> | ezyang: ping |
| 15:00 | <ezyang> | pong |
| 15:01 | <gsnedders> | ezyang: Trying to get a single emitToken in Tokenizer, what do you think of adding a callback function and calling call_user_func in Tok? |
| 15:01 | <ezyang> | That'll be slower. Why are you trying to get a single emitToken? |
| 15:02 | <gsnedders> | ezyang: So we don't need handling like setting PCDATA on end tag in every emitToken we have, nor removing duplicate attributes, etc. |
| 15:04 | <ezyang> | I was under the impression that's why it was a function call |
| 15:04 | <gsnedders> | I mean for the tokenizer tests we have an entirely separate emitToken, quite possibly with different bugs to the real tokenizer. |
| 15:04 | <gsnedders> | Which is kinda bad. |
| 15:05 | <ezyang> | Oh, I see. |
| 15:06 | <gsnedders> | I want to have a single emitToken in Tokenizer which calls another function, normally the emitToken method in TreeConstructor, but could be something else for the tests etc. |
| 15:06 | <ezyang> | So, with SimpleTest, you don't have to write any extra code: you just partially mock emitToken in TreeConstructer and do the wraparound |
| 15:07 | <gsnedders> | But that impls TreeConstructor class which makes it impossible to run the whole test suite as then TreeConstructor is already defined. |
| 15:07 | <ezyang> | No, you make a new TreeConstructer, and then you use our $builder parameter to inject the mock |
| 15:08 | <Lachy> | sigh. The same old arguments are being rehashed again about the summary attribute on public-html. :-( |
| 15:09 | gsnedders | wonders how that avoids defining two classes of the same name |
| 15:10 | <ezyang> | SimpleTest does all the hairy stuff for you. |
| 15:10 | <ezyang> | It's just a line or two of code |
| 15:11 | <Philip`> | Lachy: Maybe somebody should write them all down for easy reference when the discussion comes up again in another few months |
| 15:14 | <hsivonen> | aaargh! Astral non-characters! |
| 15:14 | <ezyang> | Hooray! |
| 15:14 | <hsivonen> | UTF-16 sucks |
| 15:15 | hsivonen | goes resurrect code from SVN |
| 15:16 | <jgraham> | Philip`: You would have to find someone who understood all the issues but was not percieved as biased |
| 15:41 | <hsivonen> | yay. down to one tokenizer test failure and that's only because I get a parse error between characters |
| 15:49 | <hsivonen> | sigh. it's diffecult to get the right character and error ordering when CR causes an early flush |
| 16:05 | <mpilgrim> | http://www.google.com/squared/search?q=html+5+elements |
| 16:06 | <annevk42> | http://www.google.com/squared/search?q=w3c |
| 16:07 | <Philip`> | mpilgrim: http://krijnhoetmer.nl/irc-logs/whatwg/20090604#l-8 - old news :-p |
| 16:07 | <annevk42> | hmm, http://www.google.com/squared/search?q=WHATWG and http://www.google.com/squared/search?q=whatwg are different |
| 16:07 | <mpilgrim> | damn it |
| 16:10 | <Philip`> | http://www.google.com/squared/search?q=polygons |
| 16:10 | <Philip`> | Apparently the internal angle of a circle is 20 degrees |
| 16:10 | <hsivonen> | LOL. add column "Status" to squaring "W3C" |
| 16:12 | <hsivonen> | (XForms is "unsupported") |
| 16:12 | <hsivonen> | (but then HTML is "closed") |
| 16:12 | <hsivonen> | (Semantic Web is "resolved: (list of) uris") |
| 16:13 | Philip` | tries adding the column "color" |
| 16:13 | <Philip`> | Apparently HTML is green, and CSS is red |
| 16:14 | <hsivonen> | I get "purple" for HTML |
| 16:14 | <mpilgrim> | http://www.google.com/squared/search?q=w3c+standards |
| 16:15 | <hsivonen> | and CSS is "black;" for me |
| 16:15 | <mpilgrim> | google has recreated the Celestial Emporium of Benevolent Knowledge! |
| 16:16 | <mpilgrim> | ( cf http://www.themodernword.com/borges/borges_quotes.html ) |
| 16:17 | <Philip`> | It's like Wolfram Alpha, except instead of giving you either useful answers or "I don't know", it always gives you complete rubbish |
| 16:18 | <mpilgrim> | it's powered by FACTS! from the INTERNET! |
| 16:18 | <mpilgrim> | (bugzilla bug #1: there are no facts on the internet) |
| 16:19 | <annevk42> | HTML is '#000"/>' for me |
| 16:19 | <annevk42> | in the 'color' column |
| 16:19 | <annevk42> | lol, DOM's Type is a "String" |
| 16:19 | <hsivonen> | annevk42: apparently the facts are different for everyone |
| 16:19 | <annevk42> | and for HTML it's "Security" |
| 16:20 | <Philip`> | That's the great thing about facts |
| 16:20 | <hsivonen> | perhaps google adapts them to the profile of your past searches |
| 16:20 | <annevk42> | I'm not signed in |
| 16:20 | <mpilgrim> | http://www.google.com/squared/search?q=natural+language+processing |
| 16:20 | <hsivonen> | annevk42: Google knows you even when you aren't |
| 16:21 | <hsivonen> | maybe Google Squared could be more accurate if it mined RDF |
| 16:21 | <annevk42> | when I search for uppercase planets the Pluto image is from Disney :) |
| 16:22 | <mpilgrim> | hsivonen: http://www.google.com/squared/search?q=rdf |
| 16:22 | <Philip`> | http://www.google.com/squared/search?q=programming+languages - apparently Pascal died in Paris, France |
| 16:22 | <hsivonen> | Philip`: and Prolog died in 2000 |
| 16:23 | <Philip`> | hsivonen: Hmm, I get "no value found" there |
| 16:23 | <Philip`> | but JavaScript died in "Nissan 300zx z32 ConZult Diagnostics OBDII Fault Codes Engine ECU Computer Software" |
| 16:23 | <hsivonen> | Philip`: it's the 'Year of Death' column |
| 16:23 | <mpilgrim> | http://www.google.com/squared/search?q=operating+systems <-- won't linus be surprised that O'Reilly Media wrote Linux |
| 16:24 | <jgraham> | mpilgrim: I was just looking at that. I this MS will be surprised windos is Free |
| 16:24 | <mpilgrim> | "Author: DOS" |
| 16:25 | <mpilgrim> | there's a kernel of truth in that somewhere |
| 16:25 | <Philip`> | I think one might conclude that parsing unstructured data is very much an unsolved problem |
| 16:28 | <mpilgrim> | http://www.google.com/squared/search?q=things+to+do+on+a+saturday yields exactly 1 result: "The Circus Starring Britney Spears" |
| 16:28 | <mpilgrim> | well, i know what *I'M* doing this saturday |
| 16:30 | <annevk42> | $128 |
| 16:30 | annevk42 | added a column cost that came back with some random value from a seemingly unrelated site |
| 16:31 | <annevk42> | squared is a lot of fun |
| 16:31 | <annevk42> | cost of HTML is apparently $49.95 |
| 16:31 | <hsivonen> | more fun that the pre-scripted fun on Wolfram|Alpha |
| 16:32 | <annevk42> | "css properties" is sort of good |
| 16:33 | <annevk42> | it copies the applies to/inherited data |
| 16:33 | <annevk42> | to compare, the cost of <frame> is $9,881,575 |
| 16:33 | <annevk42> | and class is 137,802,000 JPY |
| 16:33 | <annevk42> | table is "FREE!" |
| 16:34 | <mpilgrim> | http://www.google.com/squared/search?q=microformats says VoteLinks is $1250 |
| 16:34 | <mpilgrim> | also, rel-nofollow is depicted by scantily clad women |
| 16:35 | <annevk42> | all the good things are forbidden :p |
| 16:36 | <mpilgrim> | it's not even like the quality degrades linearly |
| 16:37 | <mpilgrim> | http://www.google.com/squared/search?q=dog+breeds is one of the front-page suggestions |
| 16:37 | <mpilgrim> | but i change it to http://www.google.com/squared/search?q=cat+breeds and it tells me that President George Bush is Persian |
| 16:37 | <tantek> | mpilgrim - wow that is wacky |
| 16:37 | <tantek> | (the microformats "square") |
| 16:37 | <mpilgrim> | hey tantek |
| 16:37 | <mpilgrim> | i'm guessing your irc client flashed on "microformats" |
| 16:38 | <tantek> | actually on VoteLinks |
| 16:38 | <mpilgrim> | ha |
| 16:38 | <mpilgrim> | you should make them less expensive if you want people to adopt them |
| 16:38 | <tantek> | LOL |
| 16:38 | <tantek> | as this channel knows, using the "rev" attribute is quite expensive |
| 16:39 | <tantek> | what's the going pay-for-play rate on HTML5 features again? |
| 16:39 | <annevk42> | tens of millions? |
| 16:40 | <mpilgrim> | apparently $600: http://www.google.com/squared/search?q=html+5+features |
| 16:40 | annevk42 | forgot |
| 16:40 | <annevk42> | regardless, you can buy HTML for 49.95USD |
| 16:42 | <Philip`> | http://googleblog.blogspot.com/2009/06/square-your-search-results-with-google.html - "This technology is by no means perfect." - gosh |
| 16:44 | <mpilgrim> | i submitted a bunch of bugs when it was strictly internal |
| 16:44 | <jgraham> | Philip`: Well at least they haven't gone all Wolfram on us proclaiming that they have invented world-changing technology only for it to be revealed as something rather less |
| 16:45 | <mpilgrim> | they don't seem to be fixed, but it's hard to tell |
| 16:45 | <Philip`> | jgraham: Indeed, but it does still seem surprisingly useless |
| 17:46 | <rubys1> | Hixie: ping? |
| 18:50 | <mpilgrim> | shelleyp: did you want a serious answer to your question on twitter? |
| 18:50 | rubys | would like a serious answer to that question |
| 18:50 | <shelleyp> | Yes, I did. |
| 18:51 | <mpilgrim> | ok. tip: if you want to directly engage me, try not to lead off with "yanked" next time |
| 18:51 | <mpilgrim> | but anyway |
| 18:51 | <mpilgrim> | my opinion on accessible markup has changed since i wrote "dive into accessibility" |
| 18:51 | <shelleyp> | Why? |
| 18:52 | <mpilgrim> | partly having seen usability studies |
| 18:52 | <mpilgrim> | partly having worked directly with screenreader vendors (while i was at ibm) |
| 18:52 | <shelleyp> | Do you have a link to the studies? |
| 18:52 | <mpilgrim> | partly from having access to real-world data on what the web looks like (from google's cache) |
| 18:52 | <mpilgrim> | they were mentioned on the list a while back |
| 18:53 | <mpilgrim> | lemme check... |
| 18:53 | <shelleyp> | Thanks |
| 18:53 | <shelleyp> | I don't know that's a good idea to base anything on Google cahe |
| 18:53 | <shelleyp> | sorry, cache |
| 18:54 | <mpilgrim> | google's cache gives a view into how markup scales |
| 18:54 | <mpilgrim> | on a global scale |
| 18:54 | <mpilgrim> | which mistakes are common |
| 18:54 | <mpilgrim> | etc. |
| 18:54 | <shelleyp> | But it's a snapshot of a point in time, not necessarily a directive for the future |
| 18:54 | <rubys> | mpilgrim: your opinion has changed, got it; care to state what your opinion now is? |
| 18:54 | <mpilgrim> | you can't get that from anecdotes or accessibility tutorials |
| 18:55 | <mpilgrim> | shelleyp: that is absolutely true |
| 18:55 | <mpilgrim> | it's only a cache of the now |
| 18:55 | <shelleyp> | Sam's impatient -- what is your view now? |
| 18:57 | <mpilgrim> | i can't find the link to the usability studies |
| 18:57 | <mpilgrim> | perhaps someone can jump in? |
| 18:57 | <mpilgrim> | it was mentioned here: http://lists.w3.org/Archives/Public/public-html/2008Feb/0059.html |
| 18:58 | <mpilgrim> | it was just 1 study, IIRC |
| 18:58 | <mpilgrim> | as for my view, i thought i spelled it out pretty well in http://diveintomark.org/archives/2009/03/21/accessibility-is-a-harsh-mistress |
| 18:58 | <mpilgrim> | "The accessibility orthodoxy does not permit people to question the value of features that are rarely useful and rarely used." |
| 18:58 | <Philip`> | Do you mean the http://lists.w3.org/Archives/Public/public-html/2007Sep/0103.html videos? |
| 18:58 | <mpilgrim> | indeed, thanks Philip` |
| 18:59 | gsnedders | is somewhat scared by how quickly Philip` can find any email on the list |
| 18:59 | <mpilgrim> | i think that there has to be some cost-benefit analysis of proposed accessibility features |
| 18:59 | <mpilgrim> | in fact, of existing accessibility features |
| 18:59 | <mpilgrim> | and there just isn't |
| 18:59 | <Philip`> | gsnedders: It's not exactly hard - I just Bing some words and click the links |
| 18:59 | <mpilgrim> | (that i know of) |
| 19:00 | <mpilgrim> | Philip`++ for using Bing (or at least claiming to) |
| 19:00 | <mpilgrim> | and you can say, "oh, if we can allow just 1 blind person to use the internet, how can you put a price on that?" |
| 19:00 | Philip` | just wanted to verb its name |
| 19:00 | <mpilgrim> | well that's crap |
| 19:00 | <mpilgrim> | we (as a society) can put a price on anything |
| 19:00 | <gsnedders> | £42. |
| 19:00 | <mpilgrim> | we put a price on a child's life |
| 19:01 | <mpilgrim> | (seatbelt laws) |
| 19:01 | <rubys> | ok, so mpilgrim doesn't care for the "accessibility orthodoxy", got it; still curious as to what mpilgrim's opinion now is... |
| 19:01 | <mpilgrim> | actuaries can put a price on anything |
| 19:01 | <mpilgrim> | but nobody ever has |
| 19:01 | <mpilgrim> | the only effort has been in defining markup (usually, accessibility-specific markup) that purports to solve certain problems |
| 19:02 | <mpilgrim> | but there's no followup to determine if those particular solutions actually DO solve those particular problems |
| 19:02 | <mpilgrim> | long descriptions are a real problem |
| 19:02 | <rubys> | http://esw.w3.org/topic/HTML/SummaryForTABLE/SummarySpecification |
| 19:02 | <mpilgrim> | navigating complex tables is a real problem |
| 19:03 | <mpilgrim> | rubys: i've read all that |
| 19:03 | <mpilgrim> | and i agree with everything except the conclusion |
| 19:03 | <mpilgrim> | because they're missing something |
| 19:03 | <mpilgrim> | the something is "does it actually fucking work?" |
| 19:04 | <mpilgrim> | nobody checks |
| 19:05 | <mpilgrim> | every time somebody does check, the answer is "no" |
| 19:05 | <mpilgrim> | nobody likes that answer, so they ignore it |
| 19:05 | <mpilgrim> | and we go round and round |
| 19:05 | <rubys> | help me prepare the right question, and I'll make sure that it is not ignored. |
| 19:05 | <mpilgrim> | it's been asked |
| 19:05 | <mpilgrim> | it's been explained |
| 19:06 | <rubys> | url? |
| 19:06 | <shelleyp> | philip: All of the links in that one site are dead - I guess hypertext links don't work, either |
| 19:06 | <mpilgrim> | um, most recently, http://lists.w3.org/Archives/Public/public-html/2009Jun/0173.html |
| 19:06 | <mpilgrim> | yeah, shelleyp: that's exactly the kind of rhetorical attitude that other people make fun of you for |
| 19:07 | <mpilgrim> | if you want a real discussion, stick around |
| 19:07 | <mpilgrim> | if you just want to take cheap shots, tell me now |
| 19:07 | <mpilgrim> | i can do that, too |
| 19:07 | <shelleyp> | But that wasn't a snarky comment -- if you look at things at specific points in time, you can find that nothing really works |
| 19:08 | <mpilgrim> | i'm not sure what that means |
| 19:08 | <shelleyp> | What I'm trying to say is that maybe the real problem with accessible markup is we haven't given it enough time to work |
| 19:09 | <shelleyp> | Our sampling size, in this case based on time, is too small -- same as judging links by one page |
| 19:10 | <mpilgrim> | html 4 is almost 10 years old now |
| 19:10 | <mpilgrim> | it was published when google had 40 employees |
| 19:10 | <mpilgrim> | how much more time do you suggest we give it? |
| 19:10 | <shelleyp> | If i remember correctly, it took decades for wheelchair ramps to be added to commercial buildings |
| 19:11 | <rubys> | mpilgrim: would you agree that the key question from Ian's note is """ Could you elaborate on why you believe summary="" serves its purpose well?" """, and acceptable answers would be stated in a form like http://diveintoaccessibility.org/day_20_providing_a_summary_for_tables.html does ? |
| 19:11 | <mpilgrim> | there's a difference: every single physical building you've ever entered in your lifetime was built by a professional and inspected/regulated by a government |
| 19:12 | <mpilgrim> | unless you're suggesting we apply that same standard to building a web site, there will always be infinitely more sites built by amateurs than by professionals |
| 19:12 | <shelleyp> | So? We can't control what amateurs do, but the government is stepping up to commercial site development |
| 19:12 | <mpilgrim> | esp. professionals who give a shit (or are forced by law to give a shit) about small niche audiences like accessibility |
| 19:13 | <shelleyp> | People don't have to have ramps for their homes, either |
| 19:13 | <mpilgrim> | i heard a "joke" (in rather poor taste, but whatever) at an accessibility conference i went to a few years ago |
| 19:13 | <rubys> | Isn't """ When JAWS encounters your calendar, Jackie hears "Summary: Monthly calendar with links to each day's posts." """ a reasonable response to "does it actually fucking work?" ? |
| 19:13 | <mpilgrim> | "if you want accessibility to be taken seriously, stab 100 million people in the eyes" |
| 19:14 | <mpilgrim> | no, the "does it actually fucking work" question needs to apply to the entire web |
| 19:14 | <mpilgrim> | that's the point |
| 19:14 | <mpilgrim> | does it work on a web scale? |
| 19:14 | <rubys> | what sort of answer are you looking for? |
| 19:14 | <mpilgrim> | of course you can design a special-purpose system within a general-purpose system that "works" |
| 19:14 | <shelleyp> | mpilgirm: you're setting a "standard" that none of us have agreed to -- has to have web scale? |
| 19:14 | <mpilgrim> | when you author it correctly |
| 19:15 | <mpilgrim> | and pay vendors to support it correctly |
| 19:15 | <mpilgrim> | and educate the user about it before you turn on the camera in the usability study |
| 19:15 | <mpilgrim> | but it doesn't fucking scale |
| 19:15 | <shelleyp> | mpilgrim: I would imagine there is more use of @summary than there is local database storage or web workers |
| 19:15 | <mpilgrim> | another category error |
| 19:15 | <hober> | shelleyp: Of course it has to have web scale--we're talking about the language of the web. |
| 19:16 | <hober> | if html features don't work at web scale, they're not features at all, they're bugs. |
| 19:16 | <mpilgrim> | if people try to use local storage and they're stupid, it won't work |
| 19:16 | <mpilgrim> | nothing is lost; it just doesn't work |
| 19:16 | <shelleyp> | mpilgrim: But who determines what web scape is? Something like SVG isn't used as widely as a div element |
| 19:16 | <mpilgrim> | maybe they give up and publish their site without oflfine capabilities |
| 19:16 | <shelleyp> | Does that mean SVG is failed? |
| 19:17 | <mpilgrim> | SVG is wildly successful |
| 19:17 | <mpilgrim> | i believe both gnome and KDE use it now |
| 19:17 | <mpilgrim> | to provide scalable icons and graphics on linux desktops |
| 19:17 | <mpilgrim> | on the web, not so much usage |
| 19:17 | <mpilgrim> | no |
| 19:17 | <hober> | Hopefully, the outcome of the SVG-in-HTML effort allows SVG to succeed *on the Web*. It's succeeded elsewhere, as markp just said. |
| 19:17 | <shelleyp> | mpilgrim: but it doesn't have the same scale on the web as div |
| 19:17 | <mpilgrim> | although you may have heard that we (google) pre-announced our make-SVG-work-in-IE javascript library |
| 19:18 | <rubys> | ... built on flash |
| 19:18 | <mpilgrim> | yes |
| 19:18 | <mpilgrim> | as a shim for a browser that will likely never support it natively |
| 19:18 | <shelleyp> | @hober: I don't agree with you. You're saying that unless an arbitrary percentage of sites use @summary, it's "failed" |
| 19:18 | <mpilgrim> | no |
| 19:18 | <mpilgrim> | that is not what we're saying |
| 19:19 | <mpilgrim> | you have to look at the percentage of people who use it |
| 19:19 | <mpilgrim> | the percentage of people who use it properly |
| 19:19 | <mpilgrim> | and the percentage of people who should use it, but aren't |
| 19:19 | <mpilgrim> | that's the problem with @summary and @longdesc |
| 19:19 | <mpilgrim> | it's not useful for most images |
| 19:19 | <mpilgrim> | it's frequently abused (i.e. wikipedia abused it for many years) |
| 19:19 | <mpilgrim> | it's frequently misused |
| 19:19 | <shelleyp> | But people are capable of learning, both from mistakes and from other people's successful efforts |
| 19:20 | <mpilgrim> | yes |
| 19:20 | <mpilgrim> | individuals are capable of learning |
| 19:20 | <mpilgrim> | groups are not |
| 19:20 | <mpilgrim> | (much) |
| 19:21 | <mpilgrim> | and (finishing my thought on @summary) most of the images that should be using it, don't |
| 19:21 | <shelleyp> | Returning to the topic, specifically @summary, where is the "harm" in its use? |
| 19:21 | <mpilgrim> | even if you educated every single blind person (the alleged target audience for the feature), they wouldn't ever run into a proper use of it |
| 19:21 | <mpilgrim> | except in accessibility tutorials |
| 19:22 | <mpilgrim> | the harm has also been explained |
| 19:22 | <mpilgrim> | i'll ignore the cost to implementers and other non-blind people for now |
| 19:22 | <mpilgrim> | focusing strictly on blind web users |
| 19:23 | <rubys> | summarizing:" summary has been around for a number of years, by this point whether it works or not should not be a matter of assertion from first principles, but from actual measurements. Key questions to be answered: who should use summary? are they? If not, why not? what should user agents do with summary? are they? If not, why not?" Is that a fair summary? |
| 19:23 | <mpilgrim> | if the feature is frequently abused or misused, then blind people are just being bombarded with crap |
| 19:23 | <mpilgrim> | accessi-spam, as it were |
| 19:24 | <shelleyp> | Unfortunately, blind people are most likely bombarded with crap on a regular basis -- including tables for layout |
| 19:24 | <mpilgrim> | if there is a real problem (navigating complex tables), and if @summary is put forth as the ONE TRUE WAY to solve that problem, |
| 19:24 | <mpilgrim> | then blind people are being harmed by the missed opportunity of a better solution |
| 19:24 | <shelleyp> | So what's your solution? |
| 19:24 | <rubys> | I am serious about wanting to find the right question, in today's HTML WG conference call, it looks like it is time to create a vote... |
| 19:24 | <mpilgrim> | i'm not saying i have an easy solution |
| 19:24 | <hober> | rubys: basically, yes. after ten years, we can be empiricists. as tantek said, http://twitter.com/t/status/318565892 |
| 19:24 | <mpilgrim> | i'm still trying to get people to understand the problem |
| 19:25 | rubys | thanks hober |
| 19:25 | <mpilgrim> | read the last sentence of http://blog.whatwg.org/the-longdesc-lottery |
| 19:25 | <shelleyp> | But if you don't have an alternative, can't we at least let those most committed have the tools they need? |
| 19:25 | <mpilgrim> | "So can we please get past the grandstanding and start talking about a better solution?" |
| 19:26 | <rubys> | forgive me, mpilgrim, but different people have different perceptions as to exactly who is doing the grandstanding. I want to get beyond that question. |
| 19:26 | <mpilgrim> | that would actively dissuade people from attempting a better solution |
| 19:26 | <mpilgrim> | my point is, i acknowledge that some images are complex enough to require further explanation |
| 19:26 | <mpilgrim> | beyond a short phrase |
| 19:27 | <mpilgrim> | i acknowledge that some tables are complex enough to warrant a description of their structure |
| 19:27 | <rubys> | the problem is that there is a growing perception that no matter how good the answers given are, they will never be "good enough" as the goalposts are always being raised. |
| 19:27 | <mpilgrim> | but for many years, the only people trying to solve these problems were only interested in solving them for themselves |
| 19:27 | <mpilgrim> | with accessibility-specific markup |
| 19:27 | <mpilgrim> | which doesn't scale |
| 19:27 | <mpilgrim> | across the world wide web of authors |
| 19:28 | <mpilgrim> | (unless you're serious about stabbing 100 million people in the eyes) |
| 19:28 | <shelleyp> | @mpilgrim, forgive me (and no snark) but that is a very circular argument |
| 19:28 | <mpilgrim> | that mindset is the problem |
| 19:28 | <mpilgrim> | it's why we're talking past each other |
| 19:28 | <mpilgrim> | one side says "if the image is so complex, wouldn't everyone benefit from a longer description of it?" |
| 19:29 | <mpilgrim> | and the other side says "YOU STOLE MY @LONGDESC!" |
| 19:29 | <mpilgrim> | "YOU HATE BLIND PEOPLE!" |
| 19:29 | <shelleyp> | If we're talking past each other, isn't there a chance that you might be wrong? |
| 19:29 | <mpilgrim> | of course |
| 19:29 | <mpilgrim> | i have been wrong about many things |
| 19:29 | <mpilgrim> | but you asked how my opinion had changed |
| 19:29 | <mpilgrim> | and that's it |
| 19:29 | <mpilgrim> | when i wrote "dive into accessibility", i believed (as the w3c believed) that accessibility-specific markup could ever be a solution to accessibility problems |
| 19:30 | <shelleyp> | But in the case of HTML5, "being wrong" and removing attributes that have existed in the past could have far reaching consequences |
| 19:30 | <mpilgrim> | and now i don't believe that |
| 19:30 | <mpilgrim> | in fact, i believe that accessibility-specific markup is actively harmful to accessibility |
| 19:30 | <rubys> | ah, now I finally see an answer to the question shelley actually asked |
| 19:30 | <mpilgrim> | and the more specific the use case, the more harmful it gets |
| 19:30 | <shelleyp> | rubys raised a point: there is a growing perception that no matter how many answers are provided, they'll never enough |
| 19:31 | <mpilgrim> | no one has demonstrated why accessibility-specific markup is the right solution |
| 19:31 | <mpilgrim> | they just keep repeating it |
| 19:31 | <mpilgrim> | they take it on faith that it is |
| 19:31 | <hober> | I think something slightly weaker is true: accessibility-specific markup whose contents aren't shown to users not using AT, that's the harmful kind of markup |
| 19:31 | <inimino> | would that include @alt? |
| 19:31 | <hober> | E.g., seems like @aria-describedby would be much better than @summary |
| 19:31 | <shelleyp> | But you're taking it on faith that enough time has passed that we can determine that accessibility markup has "failed" |
| 19:31 | <mpilgrim> | @alt is the most successful of a very weak lot |
| 19:31 | <hober> | (assuming the element pointed at is visible to everyone) |
| 19:32 | <mpilgrim> | measured in absolute terms, it has been a miserable failure |
| 19:32 | <shelleyp> | What absolute terms? |
| 19:32 | <mpilgrim> | despite being easy to grasp, and REQUIRED FOR VALIDATION |
| 19:32 | <mpilgrim> | and short |
| 19:32 | <hober> | It's been a decade! Half of the lifetime of the Web! How has not enough time passed? |
| 19:32 | <mpilgrim> | and, in many cases, easy for tool vendors to build in (at least images within templates) |
| 19:32 | <shelleyp> | No, I just got through saying it took decades just to get ramps for commercial buildings |
| 19:32 | <shelleyp> | It takes time to do things right |
| 19:32 | <mpilgrim> | (these are the cases i targeted in "dive into accessibility") |
| 19:33 | <shelleyp> | But if you change the game, frequently, nothing ever has a chance to succeed |
| 19:33 | <mpilgrim> | that may be true |
| 19:33 | <mpilgrim> | but that's a weak argument for sticking with solutions that aren't working |
| 19:33 | <inimino> | it seems reasonable to expect that the next decade will see more new HTML authors than the previous decade |
| 19:33 | <shelleyp> | A better approach would to leave the markup in, increase education, and then look at alternatives |
| 19:33 | <mpilgrim> | ("stay the course", etc.) |
| 19:34 | <shelleyp> | But now, we'll have nothing. Nothing at all. How can that be the best solution? |
| 19:34 | <mpilgrim> | that's not true |
| 19:34 | <mpilgrim> | there have been proposals put forth for how to describe structurally complex tables for everyone |
| 19:34 | <mpilgrim> | and the response from the accessibility community is always the same |
| 19:35 | <mpilgrim> | "YOU STOLE OUR @SUMMARY!" |
| 19:35 | <mpilgrim> | "YOU HATE BLIND PEOPLE!" |
| 19:35 | <mpilgrim> | "don't take away what little we have" |
| 19:35 | <shelleyp> | But I'm not saying that mpilgrim |
| 19:35 | <mpilgrim> | (even if the little we have isn't working) |
| 19:35 | <mpilgrim> | what they're really saying is "we're not thinkers interested in solving problems, we're just fans" |
| 19:35 | <mpilgrim> | "go team" |
| 19:35 | <mpilgrim> | "yay mets, boo yankies" |
| 19:35 | <mpilgrim> | whatever |
| 19:36 | <shelleyp> | mpilgrim, but what you're response is, is more of the same "grandstanding" you seem to deplore |
| 19:37 | <hober> | shelleyp: what did you think of my proposal to use @aria-describedby instead of @summary? |
| 19:37 | <rubys> | what's the status of aria in html5? |
| 19:37 | <mpilgrim> | i acknowledge that you are not saying that |
| 19:37 | <mpilgrim> | shelleyp |
| 19:37 | <hober> | It avoids the "<caption> isn't the same as @summary" rathole, and it ensures (to the extent reasonable) that the summary is visible to all users... |
| 19:37 | <mpilgrim> | so, did i answer your question? |
| 19:38 | <Hixie> | rubys: here now |
| 19:39 | <shelleyp> | mpilgrim: Yes, I understand where you're coming from, which helps. Thank you. |
| 19:39 | <mpilgrim> | i acknowledge that i tend to grandstand when talking about so-called accessibility experts |
| 19:40 | <mpilgrim> | i also acknowledge that they have put a lot of work into their answers on @summary and related features |
| 19:40 | <mpilgrim> | i still think they're not asking vital questions ("does it fucking work?") |
| 19:41 | <rubys> | hixie: http://krijnhoetmer.nl/irc-logs/html-wg/20090604#l-452 |
| 19:41 | <mpilgrim> | and i still think that all of their work starts from a fundamentally incorrect premise ("let's add accessibility-specific features to solve accessibility-related problems") |
| 19:41 | <tantek> | mpilgrim, but is calling someone out for grandstanding itself grandstanding? Just as pointing out an ad hominem attack is not itself ad hominem, neither is pointing out grandstanding itself grandstanding. Both are a criticisms of the level of dialog taking place, with the (perhaps optimistic) hope that the level of the dialog can be improved. |
| 19:41 | <rubys> | http://krijnhoetmer.nl/irc-logs/html-wg/20090604#l-468 |
| 19:42 | <mpilgrim> | tantek: i don't want to get into a meta-discussion about it |
| 19:42 | <mpilgrim> | everyone has strong opinions |
| 19:42 | <mpilgrim> | and strong personalities |
| 19:42 | <Hixie> | rubys: what would the vote be? "Should we include a technology that has been proven to not work repeatedly and to actually harm accessibility, or should we include a technology that is recommended by the US gov't and might actually solve the problem?" |
| 19:43 | <shelleyp> | mpilgrim: You're putting forth a solid argument. I don't agree, but it's an argument one can work with? |
| 19:43 | <rubys> | I'd like help in formulating the question. I don't think that's it :-) |
| 19:43 | <shelleyp> | Sorry, forget the question mark -- I'm not especially good at IRC |
| 19:43 | <tantek> | mpilgrim, I've found it difficult to escape conversational loops without raising exceptions to suboptimal conversational behaviors. YMMV. |
| 19:44 | <mpilgrim> | tantek: if i did that, i'd never finish a conversation |
| 19:44 | <Hixie> | rubys: why are we even asking a question? this isn't something where there are two valid views. There's one group asking for something that is technically inferior, and there's one group who has suggested something (that's also suggested by the very gov't agencies the former group is saying are important!) that actually addresses key questions like "does it actually improve accessibility" |
| 19:44 | <Hixie> | rubys: isn't the right technical choice obvious? |
| 19:45 | <Hixie> | rubys: i'm happy to have a vote, assuming the vote fairly outlines the actual options |
| 19:45 | <Hixie> | rubys: but i don't think it's possible to fairly outline the possible options without summary="" advocates thinking they are being shafted |
| 19:45 | <rubys> | No. hearing that " When JAWS encounters your calendar, Jackie hears "Summary: Monthly calendar with links to each day's posts"" is something that I consider compelling. |
| 19:46 | <hober> | I think the same would happen were that text in <caption>, rubys |
| 19:46 | <tantek> | rubys, which version of JAWS, which platform, with which example URL on the Web, and do you have an MP3 of the audio recording? |
| 19:47 | <rubys> | tantek: EXCELLENT questions. those are the ones I'm looking for. |
| 19:47 | <mpilgrim> | he's talking about http://diveintoaccessibility.org/20 |
| 19:47 | <shelleyp> | hober: but isn't caption supposed to be visible? In fact, isn't caption more of a table title? |
| 19:47 | <Hixie> | rubys: for every web page that you can find that actually uses summary="" in a way that actually helps surse of screen readers, i can find you a page that actually _harms_ accessibility of the page. |
| 19:47 | <Hixie> | rubys: and i can also find you two more that harm screen-reader users specifically |
| 19:47 | <Hixie> | rubys: support for the attribute is disabled by expert users because it's so badly used |
| 19:48 | <Hixie> | rubys: and the information should just be included for all users most of the time anyway |
| 19:48 | <tantek> | rubys, those questions MUST be answered for us to actually form a scientific hypothesis that can be tested and corroborated with reproducible results by someone other than those that formed the hypothesis. |
| 19:48 | <shelleyp> | Hixie: I believe the question on the floor is, will you support a vote and abide by the vote decision? |
| 19:49 | <rubys> | tantek: agreed. My (unexpert impression) is that Hixie simply is making alternate assertions. |
| 19:49 | <Hixie> | shelleyp: as an htmlwg member i will support any correct use of process, yes, of course |
| 19:49 | <Hixie> | rubys: when this issue first came up (and second, and third, and fourth, and fifth came up) i included detailed analyses including referencing which studies i was talking about and what the results were |
| 19:49 | <tantek> | Hixie, I don't think you can "fairly outline the actual options" without answering questions such as I posed for each of the options. And any such vote on options without the answers to such empirical questions is no better than randomly voting on theoretical options. |
| 19:50 | <hober> | shelleyp: re: caption and visibility, yes, it is visible, which is the whole point. @summary suffers from metacrap partially because it isn't visible. |
| 19:50 | <Hixie> | rubys: for cases 6 through 139 i stopped doing that because i am tired of repeating myself in the face of "experts" who ignore research and just demand particular solutions that have been proven to not work |
| 19:50 | <rubys> | I'm new to this, but a URL to the analysis (or to your link to the analysis) would be helpful. |
| 19:50 | <Hixie> | look through the archives |
| 19:50 | <Hixie> | i really don't have the inclination to go through this again |
| 19:51 | <rubys> | I plan to put forward a cogent question and a process to resolving this. Anybody who has the inclination to participate will be welcome to do so. |
| 19:52 | <shelleyp> | rubys: I'm no longer part of the HTML WG, but I'd like to participate |
| 19:52 | <rubys> | rejoin! |
| 19:52 | <shelleyp> | Maybe I should |
| 19:52 | <shelleyp> | I can't stop myself from getting involved, anyway |
| 19:52 | <rubys> | there is no "has to participate n hours a month" requirement. |
| 19:52 | <rubys> | well, there is, but it is really, really low. |
| 19:53 | <mpilgrim> | some data: http://lists.w3.org/Archives/Public/public-html/2009Apr/0243.html |
| 19:53 | <shelleyp> | It's not the requirement, it's the fact that I just can't seem to pull away |
| 19:53 | <mpilgrim> | some questions: http://lists.w3.org/Archives/Public/public-html/2009Jun/0173.html |
| 19:54 | <rubys> | data is anecdotal; questions aren't to tantek's level of precision |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0610.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0738.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0735.html |
| 19:54 | <shelleyp> | rubys has a good point, the data is anecdotal |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0732.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0727.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0726.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0723.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0722.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0721.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0720.html |
| 19:54 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0695.html |
| 19:55 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0693.html |
| 19:55 | <Hixie> | rubys: there is actually no minimal requirement for participation at all |
| 19:55 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0690.html |
| 19:55 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2009Feb/0680.html |
| 19:55 | <mpilgrim> | important one: http://lists.w3.org/Archives/Public/public-html/2009Feb/0601.html |
| 19:55 | <mpilgrim> | (that last url contains comments on various examples of "good summary usage" put forth by others) |
| 19:55 | <shelleyp> | mpilgrim: thanks for the links. Does anyone know if these are recorded in a wiki somewhere? Luckily logs are kept of the page |
| 19:56 | <rubys> | ok, I like the three questions in the last link mpilgrim posted |
| 19:56 | <rubys> | still not scientifically testable, but relevant questions nevertheless |
| 19:57 | <tantek> | This issue appears to provide an overview with many more links: http://www.w3.org/html/wg/tracker/issues/32 |
| 19:57 | <tantek> | "table-summary / how to provide a summary of a table, e.g. for unsighted navigation?" |
| 19:58 | <mpilgrim> | http://lists.w3.org/Archives/Public/public-html/2008Aug/0505.html |
| 19:58 | <rubys> | tantek: the goal is to get http://esw.w3.org/topic/HTML/SummaryForTABLE/SummarySpecification to the point where it contains the right question. |
| 20:00 | <shelleyp> | rubys, if I've read the wiki entries correctly, the action going forward is to frame the options into a question and then put it to a vote, correct? |
| 20:00 | <shelleyp> | Sorry, IRC entries |
| 20:01 | <tantek> | rubys, IMHO Hixie's analysis in http://lists.w3.org/Archives/Public/public-html/2009Feb/0601.html is the most thorough/scientific so far, and asks many of the tough questions that rigor demands we ask (his 1,2,3 under "several things to consider when evaluating possible solutions") |
| 20:01 | <rubys> | I try really, really, really hard to avoid votes, but unfortunately, it looks like it can't be avoided this time. |
| 20:02 | <shelleyp> | rubys: not unless you want the issue to arise, over and over again |
| 20:02 | <mpilgrim> | here's hixie giving his opinion after watching that usability study: http://lists.w3.org/Archives/Public/public-html/2008Feb/0058.html |
| 20:02 | <mpilgrim> | i don't think there are any earlier links that give hixie's opinion |
| 20:02 | <mpilgrim> | hth |
| 20:03 | <tantek> | rubys, IMHO voting on this issue won't get you a better scientific (ergo technical) result, but merely a representation of who's dogma has better propagated or who can rally the most sports fans to their side. |
| 20:03 | <Hixie> | indeed |
| 20:03 | <shelleyp> | mpilgrim: does anyone know where working links to the videos can be found? |
| 20:03 | <mpilgrim> | i do not know |
| 20:04 | <tantek> | s/who's/whose - ahem. |
| 20:04 | <mpilgrim> | i don't believe i still have them |
| 20:04 | <rubys> | The two sides I see are "I'm not listening" and "I'm not listening". |
| 20:04 | <shelleyp> | mpilgrim: I'll see if I can find them somehow |
| 20:04 | <Hixie> | woah |
| 20:04 | <Hixie> | woah |
| 20:04 | <Hixie> | woah |
| 20:04 | <Hixie> | i'm listening |
| 20:04 | <Hixie> | i've done nothing but listen |
| 20:04 | <Hixie> | did you see the e-mail i sent yesterday? |
| 20:05 | <Hixie> | i even read there dozens-of-pages-long advocacy material |
| 20:05 | <Philip`> | shelleyp: Probably http://www.cfit.ie/html5_video/ |
| 20:05 | <Hixie> | their |
| 20:05 | <rubys> | I want to see a concrete question that, if answered, would get you to change your mind. Not an abstract or untestable theory, but something tangible. |
| 20:05 | <Hixie> | i've looked at their videos |
| 20:05 | <Hixie> | rubys: anything showing that summary="" actually improves accessibility more than it hurts it |
| 20:05 | <rubys> | untestable. try again. |
| 20:05 | <tantek> | rubys, "two sides" is a false dichotomy IMHO that does not actually help the discussion move forward, and it is inaccurate to characterize requests for answers to scientific questions as "I'm not listening" (unclear if you intended that or not). |
| 20:06 | <Hixie> | rubys: no, it's not untestable |
| 20:06 | <rubys> | propose a test? |
| 20:06 | <Hixie> | rubys: it's exactly what my opinion is based on and it is the ONLY thing that really matters! |
| 20:06 | <Hixie> | rubys: did you read the e-mail mark just cited? |
| 20:07 | <rubys> | this one? http://lists.w3.org/Archives/Public/public-html/2009Feb/0601.html |
| 20:07 | <shelleyp> | Hixie: Are you saying then that it doesn't matter what's said, asked, or a vote-you won't change your mind, or include @summary |
| 20:07 | <Hixie> | rubys: we looked at how summary="" is used on the large samples of data, how it's used on small samples representing pages by people who are compelled by law to use it correctly, we looked at videos of blind users using screen readers, we looked at gov't papers on the topic, what more do you want? |
| 20:07 | <Philip`> | I guess you could test how it directly affects users (ignoring costs to authors and implementors) by finding a group of people who use the web with a screen reader and who know how to enable/disable summary support in it, and then see how many choose to enable it and how many disable it |
| 20:07 | <Hixie> | shelleyp: no, i'm saying the exact opposite |
| 20:07 | <rubys> | hixie: which email? |
| 20:07 | <Hixie> | all the ones mark just pasted into irc |
| 20:08 | <Hixie> | rubys: what would it take to change the summary="" advocates' minds? |
| 20:08 | <rubys> | at various times, yes. |
| 20:08 | <shelleyp> | Hixie, are you responding to me publicly or privately? |
| 20:08 | <tantek> | rubys, IMHO Hixie's request for "anything showing that summary="" actually improves accessibility more than it hurts it" provides you with a far wider array of options than your request for a concrete question from him. |
| 20:08 | <mpilgrim> | here's an early discussion about accessibility-specific markup vs. universal markup solutions: http://lists.w3.org/Archives/Public/public-html/2007Jun/0962.html |
| 20:08 | <Hixie> | rubys: have the summary="" advocates indicated whether they would stop asking for summary="" if the vote supported <caption> rather than summary=""? |
| 20:09 | <mpilgrim> | (read the whole thing, or at least search for "This argues for having accessibility features be implicit") |
| 20:09 | <Hixie> | rubys: have the summary="" advocates said what it would take to change _their_ minds? |
| 20:09 | <shelleyp> | Wait a sec -- folks are setting conditions on the vote, that could undermine the integrity of the vote |
| 20:09 | <shelleyp> | A vote is a vote |
| 20:09 | <Hixie> | i'm merely asking of the "other side" as rubys put it the same questions as have just been asked of me |
| 20:09 | <rubys> | tantek: the peception I want to squash is "go Mets". Where any answer that doesn't match the orthodoxy is discounted. From my perception, that applies to "summary adocates" as well as "summary detractors". |
| 20:09 | <rubys> | hixie: that's a given |
| 20:09 | <rubys> | I will follow through on that. |
| 20:09 | <Hixie> | then why are you asking me the same question? |
| 20:10 | <mpilgrim> | and an early thread about solving problems vs. rooting for your favorite solution: http://lists.w3.org/Archives/Public/public-html/2007Jun/1052.html |
| 20:10 | <tantek> | rubys, if you are viewing http://lists.w3.org/Archives/Public/public-html/2009Feb/0601.html as a "summary detractor", then I have yet to see anything even close to that level of scientific questioning/rigor from any other position (whether "summary advocate" or some other position) |
| 20:11 | <rubys> | I want everybody to agree on some sort of scientific test by which we can determine the answer. Yes, Hixie: I'd like your input on that. |
| 20:11 | Hixie | waves to the people on twitter enjoying the discussion :-) |
| 20:11 | <Hixie> | rubys: i have already given you my input on that |
| 20:11 | <Hixie> | repeatedly |
| 20:11 | <tantek> | rubys, perhaps you should put a call out for scientific tests first, maybe on the mailing list. |
| 20:12 | <shelleyp> | rubys now I have to disagree: I don't think "scientific test" is necessarily the appropriate approach |
| 20:12 | <tantek> | perhaps providing http://lists.w3.org/Archives/Public/public-html/2009Feb/0601.html as a starting point - in other words, any scientific tests proposed should be at least as good as the analysis in that email to be worthy of consideration. |
| 20:12 | <rubys> | I do plan to put out a question on the mailing list, I just wanted to gather some input on IRC first. |
| 20:12 | <shelleyp> | After all, we can't put the web in a petri dish, and turn on a microscope |
| 20:13 | <rubys> | tantek: agreed. |
| 20:16 | <shelleyp> | Then I think it's only fair to say that every last bit of the HTML5 spec be subjected to the same form of scientific tests, to be fair |
| 20:16 | <Hixie> | it has been |
| 20:16 | <Hixie> | that's been one of the cornerstones of html5's development |
| 20:16 | Philip` | doesn't remember having seen suggested tests for summary that are actually scientific (like with numbers) rather than being subjective judgements on the relative importance of various benefits and costs |
| 20:16 | <shelleyp> | Hixie: Most have been subjected to your interpretation, and a couple of test apps -- that's not a 'scientific test' |
| 20:17 | <Philip`> | (...like the cost of bad summary text vs the value of good summary text) |
| 20:17 | <Hixie> | Philip`: yup, just like the rest of the spec :-) |
| 20:17 | <Hixie> | Philip`: (it's been subject to the same form of scientific tests) |
| 20:17 | <shelleyp> | Philip: Thank you for clarification |
| 20:18 | <Philip`> | Hixie: You mean the unscientific form of scientific tests? |
| 20:18 | <Hixie> | shelleyp: it's language design. interpretation is the name of the game. |
| 20:18 | <Hixie> | Philip`: yup |
| 20:18 | <shelleyp> | In other words, not based on empirical measurements, but the reason and logic of the arguments |
| 20:18 | <Hixie> | based on both |
| 20:18 | <Hixie> | empirical measurements isn't science |
| 20:18 | <Hixie> | but it's still better than nothing |
| 20:18 | <shelleyp> | Hixie you're not designing a language, you're improving a language |
| 20:19 | <Hixie> | why thank you |
| 20:19 | <tantek> | Hixie, empirical measurements are used in some science. And yes, better than nothing. |
| 20:19 | <Hixie> | tantek: empirical measurements is necessary but insufficent to have science |
| 20:19 | <Hixie> | er, s/is/are/ |
| 20:20 | <shelleyp> | Philip: can you suggest a specific scientific question/test on a specific component of the HTML spec I can look at in order to ensure my interpretation of "scientific test"? |
| 20:21 | tantek | provides "Science is a process" http://evolution.berkeley.edu/evolibrary/article/0_0_0/nature_05 as hopefully a good overview. |
| 20:22 | <Hixie> | (just to clarify, i don't think i ever said that we needed scientific tests. i said we need to base our judgements on research rather than on unqualified opinions.) |
| 20:22 | <shelleyp> | tantek: thank you, but I meant specifically as has been applied in the past to HTML5 |
| 20:22 | <shelleyp> | I work best with given examples of past success |
| 20:22 | <Hixie> | (research shows that summary="" isn't actually solving the problem that needs solving here; therefore we need something better. various people, including some from the US gov't's accessibility teams, have concluded that <caption> is an overall better solution.) |
| 20:23 | <tantek> | Hixie, basing judgments on research rather than unqualified opinions *is* a scientific method. |
| 20:24 | <Hixie> | tantek: to some extent, yes; but some people (e.g. rubys earlier) believe that it means you can come up with a single test the result of which can directly drive language design without a subjective component |
| 20:24 | <Hixie> | which is clearly not the case |
| 20:24 | <Hixie> | you have to interpret the results of any study |
| 20:24 | <tantek> | shelleyp, Hixie certainly applied "scientific thinking ... observations, questions, multiple hypotheses, and more observations." to the table summary issue, per the cited 601.html email. |
| 20:24 | <tantek> | (quote excerpted from "Science is a progress" link above) |
| 20:25 | <shelleyp> | tantek, thank you, I was looking for an example of a specific test where the result was to add to the HTML5 spec. For instance, what scientific test was used with the microdata section? |
| 20:25 | <tantek> | Hixie, I'm not sure of the "single test" hypothesis either - that's why I simply asked for alternative scientific analyses, at least as rigorous as your 601.html |
| 20:25 | <Hixie> | tantek: right |
| 20:26 | <Hixie> | shelleyp: no one test could be used for something that big, the microdata section involved hundreds of individual decisions each of which had to be based on research, implementation experience, discussions with authors, etc |
| 20:27 | <Hixie> | shelleyp: for example, to test whether the microdata features as described actually solved the problems they set out to solve, i explicitly used them to do so |
| 20:27 | <Hixie> | shelleyp: (most of that isn't documented in the list, i just did it in scratch files) |
| 20:27 | <Hixie> | shelleyp: (some of it became the examples in the spec) |
| 20:27 | <tantek> | shelleyp, it may be that the evaluation process for adding new features (e.g. microdata) is different than the evaluation process for dropping existing features, since existing features can be better evaluated based on observations in the wild. |
| 20:28 | <shelleyp> | Hixie: So an instance of implementation is a successful test of a given component of the HTML5? |
| 20:28 | <Hixie> | tantek: certainly the evaluation process for features for which we have a decade of experience can be far more rigorous |
| 20:28 | <Hixie> | shelleyp: it's a data point, of course. certainly not the only one. it is necessary but not sufficient. |
| 20:29 | <shelleyp> | OK, good, that's something tangible I can work with |
| 20:29 | Hixie | thought it was pretty obvious that you had to evaluate whether proposals actually solved the problems they set out to solve |
| 20:29 | <Hixie> | shelleyp: for example one use case was to take the events on livebrum.co.uk and make them into paragraphs that can be copied and pasted into blogs while still being drag-and-droppable to calendars |
| 20:29 | <Hixie> | shelleyp: so i worked with the guy who runs that site to come up with something that would satisfy his need |
| 20:29 | <shelleyp> | Yes, but interpretation of whether a problem is "solved" are frequently based on biased, subjective viewpoints |
| 20:30 | <Hixie> | you can see this example in the spec, it's one of those that made it in |
| 20:30 | <Hixie> | shelleyp: if so, then your problem description is far too vague |
| 20:30 | <Hixie> | shelleyp: problem descriptions should be objectively testable |
| 20:32 | <shelleyp> | Hixie: Of course, there's another issue: that you seem to be the one person making a determination of what is, or is not, a viable test |
| 20:32 | <Hixie> | the vast majority of the problem descriptions for microdata were in fact written by other people |
| 20:33 | rubys | agrees with Philip` |
| 20:33 | <shelleyp> | Hixie: yeeesss, but as I pointed out in a later document, you were incorrect in your assessment of the some of the problem descriptions |
| 20:33 | <shelleyp> | Your interpretation was in error |
| 20:34 | <Hixie> | shelleyp: i am not infallible, hopefully you raised such errors to my attention on the list |
| 20:34 | <shelleyp> | Hixie: Yes, I did. Unfortunately, you've been quite busy since. I still am reviewing your work, for a longer writing |
| 20:35 | <rubys> | hixie: what's the status/plans for aria? |
| 20:36 | <Hixie> | rubys: unchanged since my last "input on the agenda" e-mail |
| 21:16 | <Hixie> | gsnedders: yt? |
| 21:16 | gsnedders | wonders what now… |
| 21:16 | Hixie | looks for the sample anolis doc |
| 21:16 | <Hixie> | do you have the link to that again? |
| 21:16 | <Hixie> | we should put it on pimpmyspec.net |
| 21:17 | <gsnedders> | <http://hg.gsnedders.com/anolis/raw-file/tip/example.src.html>; |
| 21:17 | <gsnedders> | <http://hg.gsnedders.com/anolis/raw-file/tip/example.html>; |
| 21:17 | <gsnedders> | Message-Id: <699948DD-6166-4F55-8171-44D81108A898⊙gc> |
| 21:20 | <Hixie> | awesome thanks |
| 21:32 | gsnedders | just searched his sent box for to:hixie |
| 21:32 | <gsnedders> | :P |
| 22:11 | <jgraham> | Hixie: Added links to those docs to pms.net |
| 22:33 | <Hixie> | jgraham: sweet thanks |
| 22:34 | <gsnedders> | Weee… |
| 22:35 | <gsnedders> | just got a SPARQL query down from using 1GB RAM in rdflib to around 12MB RAM. |
| 22:35 | <gsnedders> | (It is really, really, really inefficient on some queries) |
| 22:36 | <Philip`> | I think you should rewrite the query to just use grep |
| 22:36 | <gsnedders> | That would be very hard. |
| 22:55 | <Philip`> | gsnedders: Nonsense! Everything is easy with grep |
| 22:59 | <Philip`> | Am I missing some place where <foo xmlns:bar=""/> is clearly a namespace-well-formedness error? |
| 22:59 | <Philip`> | All I see is "the namespace name may not be empty", which isn't a very convincingly accurate conformance requirement |
| 23:04 | <Philip`> | Oh, it's clarified in an errate |
| 23:04 | <Philip`> | s/e/a/ |
| 23:06 | <Hixie> | the average rate of new feedback has increased |
| 23:06 | <Hixie> | interesting |
| 23:53 | <Hixie> | hsivonen: yt? |