| 00:54 | <BenMillard> | hixie, from the IRC logs I see you wrote "anyone have any web pages that use mathml that aren't, like, the blogs of editors of the spec or anything like that?" to which I'd suggest the only web pages that make *good* use HTML are from a similar segment of the population |
| 01:00 | <BenMillard> | "You can learn a lot by viewing an interesting page with MathML and selecting 'View Page Source'." from http://www.pdas.com/mathml.htm suggests it exists |
| 01:02 | Philip` | finds one page using a <math> element |
| 01:03 | <Philip`> | http://www.unics.uni-hannover.de/cpr/x/rueckemann/de/rueckemann.html - "Paderborn Center for Parallel Computing (PC<MATH CLASS=INLINE><SUP>2</SUP></MATH>)" |
| 01:03 | <BenMillard> | http://www-math.mit.edu/18.013A/MathML/chapter06/section01.xhtml is from http://www-math.mit.edu/18.013A/ which is a MIT course material...since MIT is part of W3C maybe that doesn't count? |
| 01:04 | <BenMillard> | http://www.math.uah.edu/stat/ seems to use it throughout (states MathML support as a requirement on the homepage) |
| 01:04 | <BenMillard> | btw I'm finding these by Googling for "mathml on the web" |
| 01:05 | <BenMillard> | http://www.math.uah.edu/stat/foundations/Sets.xhtml - an example from the previous site |
| 01:14 | <BenMillard> | http://www.mozilla.org/projects/mathml/ was a project to develop MathML authoring tools and interoperable practices, such as using <object> |
| 01:18 | <BenMillard> | http://pear.math.pitt.edu/mathzilla/Examples/markupOftheWeek.mhtml uses MathML, the site also has this paper where MathML is used inside sentences: http://pear.math.pitt.edu/mathzilla/Examples/DynHieruni.xml |
| 01:29 | <BenMillard> | http://www1.chapman.edu/~jipsen/mathml/asciimath.html uses ASCII which is then turned into markup using JS, it states "ASCIIMathML is being used with Movable Type, WordPress, phpBB, and many wikis." but I can't find links to example sites |
| 01:30 | <BenMillard> | the Web appears to be a big place |
| 01:45 | <BenMillard> | http://www.dessci.com/en/products/webeq/ is a MathML editor which produces MathML and uses Java. that page claims "The world's leading e-learning companies, content developers and education portals are using WebEQ to create web-based learning environments that help educators engage students in math and science on the web." but I cannot find example pages of these |
| 01:57 | <Dashiva> | BenMillard: Probably a lot on intranets, but that doesn't help us |
| 02:02 | <Hixie> | webben_: html5 has <ol reversed> or some such now |
| 02:02 | <webben_> | Hixie: cool |
| 02:02 | <Hixie> | BenMillard: oh i'm not looking for pages that use mathml correctly :-) |
| 02:03 | <Hixie> | BenMillard: i'm looking for "real" pages |
| 02:06 | <webben_> | http://golem.ph.utexas.edu/category/ and the journals mentioned in the mathml thread seem relevant |
| 02:24 | <Hixie> | BenMillard, webben: thanks for the links |
| 02:39 | <BenMillard> | hixie, np. there were lots of results for searches along those lines |
| 02:39 | <Hixie> | cool |
| 02:39 | <Hixie> | i shall investigate further |
| 02:43 | <BenMillard> | webben, <ol reversed> was in Hixie's response here http://lists.w3.org/Archives/Public/public-html/2008Feb/0357.html |
| 02:43 | <BenMillard> | (search for "9 Jan 2008" to see messages which led to it) |
| 02:44 | <BenMillard> | (the forum thread it links to here http://forums.whatwg.org/viewtopic.php?t=130 has lots of use cases) |
| 02:44 | <Hixie> | it's in the spec here: http://www.whatwg.org/specs/web-apps/current-work/#reversed |
| 02:45 | <BenMillard> | and in the multipage version (which I much prefer!) here: http://www.whatwg.org/specs/web-apps/current-work/multipage/section-grouping.html#reversed |
| 02:47 | <Hixie> | :-) |
| 02:47 | <Hixie> | i hate multipage specs |
| 02:47 | <Hixie> | i can never find anything in them |
| 02:49 | <BenMillard> | as an author, I'm usually looking for a specific element or attribute, so they are OK if there's an index of those things |
| 02:50 | <Hixie> | yeah |
| 02:50 | <Hixie> | we will eventually add much better indexing |
| 02:50 | <BenMillard> | the front page is pretty good with find-in-page |
| 02:51 | <Hixie> | (gsnedders doesn't know it yet, but if he ends up writing a good preprocessor, he'll probably end up helping out a lot in that area) |
| 02:53 | <Hixie> | it's weird how people associate syntax with language |
| 02:53 | <Hixie> | (c.f. David Carlisle's e-mail) |
| 02:55 | <BenMillard> | there are so many I don't know which that is :) |
| 02:55 | <Hixie> | more recent ones |
| 02:55 | <Hixie> | it is weird that we are getting so much reluctance from the mathml group to make their syntax easier |
| 02:55 | <Hixie> | given that the mathml syntax is the number 1 problem with mathml |
| 02:56 | <BenMillard> | oh, I see |
| 02:58 | <BenMillard> | if the people using the syntax are happy with the syntax then it doesn't seem like a problem to me... |
| 02:59 | <Hixie> | the people who aren't happy with their syntax aren't using it :-) |
| 03:00 | <Hixie> | seriously though, compare the first two examples in http://wiki.whatwg.org/wiki/Equations_in_HTML |
| 03:00 | <Hixie> | how do we _not_ want people to use the second? |
| 03:00 | <BenMillard> | neither are things I'd ever want to write, personally :P |
| 03:01 | <BenMillard> | from what I've read, the tools handle the syntax and the author uses a word processor, or they use an ASCII-based syntax which is then coverted by a tool |
| 03:01 | <Hixie> | can't argue with that |
| 03:01 | <Hixie> | unfortunately "the tools with save us" has rarely worked on the web |
| 03:02 | <BenMillard> | indeed, but mathematicians aren't noobs and the MathML which exists seems pretty well authored, from the little I've seen |
| 03:03 | <Hixie> | just like english majors aren't the only audience of prose html elements, mathematicians aren't the only audience for the math parts |
| 03:03 | <Hixie> | and frankly, most mathematicians _are_ noobs when it comes to writing markup and using equation editors |
| 03:03 | <Hixie> | after all, they're experts in maths, not editors |
| 03:03 | <BenMillard> | heh |
| 03:04 | <BenMillard> | you think non-mathematicians would use any sort of math markup? |
| 03:10 | <Hixie> | i'm not a mathematician |
| 03:10 | <Hixie> | and my final year report for my physics degree had loads of mathml in it |
| 03:10 | <Hixie> | handwritten |
| 03:10 | <Hixie> | painfully |
| 03:11 | <BenMillard> | my idea of a "mathematician" is broad and would include people who publish any degree-level maths |
| 03:12 | <BenMillard> | would you expect math markup to appear around casual sums, like people adding up the cost-per-year of their TV service? |
| 03:13 | <Hixie> | no |
| 03:13 | <Hixie> | but i wouldn't be surprised to see it in a web page talking about 3d graphics |
| 03:13 | <BenMillard> | ok, so we only expect math markup from people who are fairly serious about maths? |
| 03:14 | <Hixie> | i'm not sure what "fairly serious" means really |
| 03:14 | <Hixie> | i'd expect people to use things like fractions long before they knew about integration |
| 03:16 | <Hixie> | well a brief adhoc survey of pages that use mathml suggests that none of them use both presentational mathml and content mathml, though one page did use content mathml and convert it to presentational mathml using xslt, which is pretty cool |
| 03:16 | <Hixie> | clearly i need to do a more serious study |
| 03:28 | <webben_> | "A valid browsing context name is any string that does not start with a U+005F LOW LINE character, or, a string that case-insensitively matches one of: _self, _parent, or _top. (Names starting with an underscore are reserved for special keywords.)" ... is making "_blank" invalid deliberate? |
| 03:28 | <webben_> | http://www.w3.org/html/wg/html5/#valid8 |
| 03:32 | <Hixie> | yeah |
| 03:32 | <Hixie> | but we'll probably change that |
| 03:33 | <Hixie> | and instead make it optional for UAs to honour it |
| 03:34 | <webben_> | Hixie: Okay, just wondering. Thanks for the info. :) |
| 03:43 | <BenMillard> | the month I took off work ends on 1st April, officially. so making websites will become my priority once again |
| 03:43 | <BenMillard> | night night |
| 10:14 | <hendry> | anyone going to Xtech? |
| 10:14 | <hsivonen> | hendry: I am |
| 10:15 | <hendry> | hsivonen: ok, great. :) |
| 10:33 | <gsnedders> | Hixie: I know now :P |
| 10:47 | <jgraham__> | http://blog.ianbicking.org/2008/03/30/python-html-parser-performance/ |
| 10:52 | <hsivonen> | jgraham__: last summer, the html5lib source didn't look performance-oriented |
| 10:52 | <jgraham__> | hsivonen: It's not |
| 10:52 | <hsivonen> | though I don't know if cPython does Scheme-like optimizations |
| 10:52 | <hsivonen> | I assumed it doesn't |
| 10:52 | <jgraham__> | although it has got better since last summer |
| 10:53 | <jgraham__> | You mean the tail call optimization stuff? |
| 10:53 | <hsivonen> | jgraham__: yes |
| 10:53 | <jgraham__> | That doesn't even feature on the radar of perf. problems for typical documents |
| 10:54 | <jgraham__> | Last I checked, we spend somewhere around half the time just reading characters |
| 10:54 | <hsivonen> | ah |
| 10:55 | <jgraham__> | (although that may actually have gone down, we had a patch that improved things and I might be remembering numbers from before that patch) |
| 11:47 | <gsnedders> | Philip`: http://hg.gsnedders.com/hgwebdir.cgi/http-parsing/file/fbfbeb6b74b7/Philip%20Taylor%27s%20Header%20Data/README.txt — comments? |
| 12:15 | <hsivonen> | book authors should read my site :-) http://lists.w3.org/Archives/Public/www-style/2008Mar/0435.html |
| 12:34 | <hsivonen> | hmm. the University of Illinois ARIA demos have bad target attribute values and unescaped spaces in URIs |
| 12:43 | <hsivonen> | WHOA! http://lists.w3.org/Archives/Public/www-style/2008Mar/0437.html |
| 12:45 | <Philip`> | gsnedders: That makes it sound really complex |
| 12:45 | <gsnedders> | Philip`: :P |
| 12:45 | <Philip`> | gsnedders: In my existing content, the uri attribute isn't necessarily syntactically valid |
| 12:46 | <gsnedders> | ergh. |
| 12:46 | <Philip`> | If I remember correctly, it's taken directly from the dmoz.org list |
| 12:46 | <Philip`> | and a few of those are broken |
| 12:46 | <Philip`> | so they're likely to end up as <error>s |
| 12:47 | <gsnedders> | from what I saw they are all valid IRIs, but not URIs |
| 12:47 | <takkaria> | hsivonen: that makes me unhappy |
| 12:47 | <hsivonen> | takkaria: yeah. versioning is bad for Web formats |
| 12:49 | <Philip`> | gsnedders: Hmm, error messages have types? |
| 12:49 | <gsnedders> | Philip`: in what you gave me they do :P |
| 12:49 | <Philip`> | Oh, so they do |
| 12:50 | <Philip`> | res.statusCode != HttpStatus.SC_OK causes type=http |
| 12:50 | <Philip`> | IOException (e.g. any HTTP error detected inside HttpClient, I think) causes type=io |
| 12:50 | <Philip`> | Any other Exception causes type=other |
| 12:51 | <hsivonen> | Validator.nu with HTML5+ARIA is now up |
| 12:51 | <hsivonen> | (where ARIA is more like ARIA with my opinion) |
| 12:51 | gsnedders | wonders if he can con his way into going to France/Swizerland for next Sunday |
| 12:52 | <hsivonen> | I'll test it some more and then proceed to announce it as an executable suggestion |
| 12:52 | <gsnedders> | LHC's last open day |
| 12:54 | <Philip`> | gsnedders: All you need to do is become a particle physicist, and then spend a couple of years analysing data from it, and then maybe you'd be able to visit it and have a good look |
| 12:54 | <gsnedders> | Philip`: Particle physics aren't really my area, though |
| 12:54 | <hsivonen> | hmm. I forgot to add IDREF checking for ARIA... |
| 12:55 | <gsnedders> | s/aren't/isn't. |
| 12:55 | <gsnedders> | s#.#/# |
| 12:55 | <gsnedders> | you can tell I'm a scientist, my English is bad :P |
| 12:56 | Philip` | starts reading the maths thread again, then sees the size of the scrollbars and stops |
| 12:56 | gsnedders | has just started too |
| 12:56 | <Philip`> | s/starts/accidentally starts/ |
| 13:00 | <takkaria> | gsnedders: ~£155 it looks like |
| 13:00 | <hsivonen> | how do I test in XPath than a node returned by id() is a descendant of the context node? |
| 13:06 | hsivonen | guesses test='id(@aria-activedescendant) = descendant::*' |
| 13:07 | <Philip`> | gsnedders: It would be useful to say stuff like whether the data for a given uri might be repeated, and like which redirect URI matches the other data's uri |
| 13:08 | <Philip`> | Also this is a rubbish format for storing HTTP headers |
| 13:08 | <Philip`> | like how it's just a load of elements mashed together |
| 13:08 | <Philip`> | It'd seem much better to design a more useful format, then convert the old XML files into that |
| 13:09 | <Philip`> | (particularly since any similar files I produce in the future are likely to be slightly different formats anyway) |
| 13:10 | <hsivonen> | yay. correct guess |
| 13:26 | <hsivonen> | hmm. what's the point of putting a W3C Validation button on this page: http://test.cita.uiuc.edu/aria/alert/view_inline.php?title=Alert%20Example%201:%20Number%20Guessing%20Game&ginc=includes/alert1_inline.inc&gcss=css/alert1_class.css&gjs=../js/globals.js,../js/widgets_inline.js,js/alert1_class.js |
| 13:27 | <hsivonen> | (aside: the URL looks like an invitation for XSS) |
| 13:33 | <Philip`> | hsivonen: Not just XSS |
| 13:33 | <Philip`> | http://test.cita.uiuc.edu/aria/alert/view_inline.php?ginc=/etc/passwd |
| 13:35 | <Philip`> | http://test.cita.uiuc.edu/aria/alert/view_inline.php?ginc=http://google.com - still with the valid HTML 4.01 button |
| 13:39 | hsivonen | alerted the admin |
| 13:39 | <Philip`> | You are too kind :-( |
| 13:44 | <hsivonen> | XSS vulnerabilites are alarmingly common |
| 13:45 | <Philip`> | Printing arbitrary local files seems to be much less common, fortunately |
| 13:45 | <hsivonen> | this morning, a tinyurl embedding the Pirate Bay on the site of a mandatory royalty collection society circulated on IRC |
| 13:56 | <gsnedders> | Philip`: I'm moving all the data to SQLite anyway |
| 13:56 | <gsnedders> | Philip`: the URI after a redirect is which? |
| 13:57 | <gsnedders> | Philip`: And please don't have even more incompatible formats. I already have two to deal with :P |
| 13:58 | <Philip`> | gsnedders: I think I always print the source URI, i.e. the one that came from dmoz, i.e. before any redirects |
| 13:58 | <gsnedders> | Philip`: it seems so. I think I'll keep that undefined :P |
| 13:59 | <Philip`> | gsnedders: I'll make as many formats as are convenient for my survey process :-p |
| 13:59 | <gsnedders> | Philip`: damn you :P |
| 14:01 | <Philip`> | It's much easier to transform things into the desired format after it's all been condensed down to only a few hundred megabytes of easily-parsed (by regexps) XML |
| 14:03 | <gsnedders> | Philip`: Yeah, I just need more and more converters :P |
| 14:12 | <gsnedders> | Philip`: just keep the vague syntax of the header element the same, so that the same attributes exist on it :P |
| 14:12 | <gsnedders> | Philip`: even if you change the uri attribute to represent the actual URI where it came from |
| 15:11 | hsivonen | learns an important Schematron detail I should have learned two years ago |
| 15:34 | <hsivonen> | zcorpan: ARIA-aware Validator.nu deployed |
| 15:36 | <zcorpan> | hsivonen: sweet! |
| 15:37 | <zcorpan> | hsivonen: i'll have a look later |
| 15:39 | <hsivonen> | I'll write some documentation and findings |
| 17:06 | <annevk> | anyone ideas for tomorrow? |
| 17:11 | <annevk> | as in, for the blog |
| 17:11 | <annevk> | we drop support for <a> this year? |
| 17:12 | <virtuelv> | annevk: no, we can re-add the <hype> element |
| 17:12 | <annevk> | and suggest everything be done using onclick=location=... instead |
| 17:12 | <annevk> | to more clearly indicate the move towards applications |
| 17:12 | <virtuelv> | context, http://jwz.livejournal.com/856745.html |
| 17:14 | <hsivonen> | annevk: you mean like href on any element when a is gone? |
| 17:16 | <annevk> | heh |
| 17:16 | <zcorpan> | we need something more drastic |
| 17:16 | <annevk> | i meant having them do it with event listeners |
| 17:17 | <zcorpan> | like... html5 is cancelled |
| 17:17 | <aroben> | stopping development in favor of XHTML2? |
| 17:17 | <zcorpan> | or something else |
| 17:17 | <billmason> | In honor of Netscape history, version 5 is cancelled and version 6 will be written from scratch. |
| 17:18 | <annevk> | hehe |
| 17:19 | <annevk> | http://blog.whatwg.org/html6-plan was last year fwiw |
| 17:20 | <virtuelv> | mandate python as scripting language? |
| 17:20 | <virtuelv> | because python whitespace rules are such a good match for the web |
| 17:20 | <hsivonen> | Google Python conspiracy |
| 17:22 | <annevk> | dropping HTML in favor of wiki syntax |
| 17:22 | <annevk> | all 42 flavors |
| 17:25 | <zcorpan> | the whatwg is starting to get unresponsive and is moving in the wrong direction, so we'll start a new body |
| 17:26 | <annevk> | lol, "much simpler" |
| 17:26 | <annevk> | -- http://www.w3.org/mid/035401c8934c$12dd26e0$389774a0$@com |
| 17:41 | Philip` | wonders if someone will point out that the text in "Definition of the style attribute in HTML 5" is only defining document conformance and nothing about processing requirements and so there is no effect on parsing |
| 19:13 | <annevk> | there's now http://webforms2.org/ |
| 19:13 | <annevk> | content welcome |
| 19:17 | <hober> | annevk: sorry we didn't get a chance to talk much the other night (I was the one in the 5>2 shirt @ the µf dinner) |
| 19:17 | <annevk> | ah :) |
| 19:18 | <annevk> | mostly my fault I guess, I wasn't there too long |
| 19:19 | <hober> | well, it was a sufficiently long/narrow table such that I didn't really talk to anyone at the other end |
| 20:16 | <zcorpan_> | hsivonen: i think aria attributes that would otherwise be specified on root are specified on body when the document is (x)html as implemented in firefox |
| 20:19 | <zcorpan_> | (because i and aaronlev thought that it would be more intiutive to authors) |
| 20:26 | <hsivonen> | zcorpan_: ok |
| 20:26 | <hsivonen> | thanks |
| 20:32 | <hsivonen> | zcorpan_: deployed |
| 20:58 | <xShad0w_> | does anyone know about embedding xml, or adding xml into a html file |
| 20:58 | <xShad0w_> | anything about html5 supporting it? |
| 20:59 | <annevk> | currently no, what problem are you trying to solve? |
| 21:00 | <xShad0w_> | well im trying to embed some data |
| 21:00 | <xShad0w_> | i just need a way to store some data thats accessible with javascript |
| 21:00 | <annevk> | ah ok |
| 21:01 | <annevk> | Hixie is working on making that possible: http://wiki.whatwg.org/wiki/CustomData |
| 21:01 | <xShad0w_> | i dont want add too much syntax with div's or such, any suggestions? |
| 21:01 | <xShad0w_> | would that pass in current browsers? |
| 21:01 | <annevk> | well, .dataset wouldn't work |
| 21:01 | <annevk> | but you could use set/getAttribute to do that |
| 21:02 | <xShad0w_> | i should set the display to none right |
| 21:02 | <annevk> | do data-* attributes work for you? |
| 21:03 | <xShad0w_> | umm yea that should work |
| 21:04 | <xShad0w_> | so all the attributes have to start with data- ? |
| 21:04 | <annevk> | that's the proposal |
| 21:05 | <xShad0w_> | looks like it should work thanks |
| 21:06 | <annevk> | please note that all this is pre-alpha :) |
| 21:08 | <xShad0w_> | well i guess it won't pass w3c validation for now but hopefully it should work |
| 21:08 | <Hixie> | wow, whatever lxml does, html5lib should do :-) |
| 21:08 | <annevk> | having a native C implementation? |
| 21:28 | <Hixie> | nthat works :-) |
| 21:58 | <jgraham> | Hixie: Patches welcome :) |
| 22:04 | <jgraham> | (I did start working on a C implementation btw, but work on that has totally stalled) |
| 22:21 | <annevk> | does Mozilla do full MathML or just presentational MathML? |
| 22:23 | <Hixie> | p |
| 22:23 | <annevk> | right, that's what I thought |
| 22:23 | <Pavlov> | not sure, lots of mathml work has been going on lately |
| 22:23 | <Pavlov> | but not sure to what extent |
| 22:24 | <annevk> | I don't get the comments about "full MathML" and all then referring to Mozilla |
| 22:24 | <annevk> | Pavlov, fixing the breakage? :) |
| 22:24 | <Pavlov> | annevk: certain some of it, but fixing a lot of of things beyond that as i understand it |
| 22:24 | <Pavlov> | "full mathml" probably ~= "enough mathml to do what I want" |
| 22:35 | <Hixie> | i really don't see a way to go foward with the math stuff |
| 22:36 | <othermaciej> | math markup seems like a messy situation |
| 22:36 | <Pavlov> | it is |
| 22:36 | <hsivonen> | Hixie: why not. all we need is a new insertion mode that triggers on <math> |
| 22:36 | <othermaciej> | MathML seems painful as a direct authoring format |
| 22:36 | <Hixie> | hsivonen: new insertion modes don't work |
| 22:36 | <hsivonen> | Hixie: getting the mess into <math> is left to iTeX4MML, TeX4HT, OpenOffice.org 4, etc. |
| 22:36 | <othermaciej> | but majorly diverging from it on surface syntax also seems problematic |
| 22:37 | <Hixie> | hsivonen: "the tools will save us"? |
| 22:37 | <hsivonen> | Hixie: yes |
| 22:37 | <roc_> | we only do pres. mathML |
| 22:38 | <Hixie> | hsivonen: they won't :-) |
| 22:38 | <hsivonen> | Hixie: as for Content MathML, I'd allow it when Presentation MathML is there |
| 22:38 | <hsivonen> | Hixie: clipboard export of Content MathML is low-cost for browsers |
| 22:38 | <Hixie> | hsivonen: having two formats doesn't work either, you know that |
| 22:38 | <annevk> | if nobody supports content MathML why support it |
| 22:39 | <Hixie> | also very few authors seem to actually provide it |
| 22:39 | <hsivonen> | Hixie: so in case Wolfram and Waterloo come up with something wonderful, we wouldn't prevent them |
| 22:39 | <Hixie> | hsivonen: there's no way we're including redundant markup. |
| 22:39 | <Hixie> | hsivonen: that's a known anti-pattern. |
| 22:39 | <hsivonen> | Hixie: I think browsers, Google and most authors will only care about Presentation MathML |
| 22:40 | <hsivonen> | Hixie: ok. |
| 22:40 | <hsivonen> | Hixie: how do you prevent it without putting traps in the tree builder on purpose? |
| 22:40 | <Hixie> | hsivonen: we have to hardcode all the tag names anyway, it won't work unless we explicitly allow it |
| 22:41 | <hsivonen> | Hixie: why can't you scope on <math>? |
| 22:41 | <Hixie> | hsivonen: because there are documents that have <math> tags in them today where those tags are just ignored |
| 22:42 | <hsivonen> | Hixie: are those docs for IE and Google misdiagnoses them? |
| 22:42 | <Hixie> | hsivonen: and if we switch insertion modes and do something dramatically different to today until </math>, those pages will break |
| 22:42 | <hsivonen> | :-( |
| 22:42 | <Hixie> | hsivonen: some surely are, but there's all kinds of crap out on the web |
| 22:43 | <Hixie> | sam's forwarding of jacque |
| 22:43 | <Hixie> | 's comments was very useful |
| 22:44 | <annevk> | if pages have <math> doing special parsing inside <math> based on punctation and numbers won't work either |
| 22:44 | <Hixie> | annevk: adding <mo>, <mi>, and <mn> elements won't affect most pages |
| 22:45 | <Hixie> | annevk: it would only affect pages that did really weird things with script and/or css |
| 22:46 | <hsivonen> | http://www.ibm.com/developerworks/blogs/page/woolf?entry=secure_mashups_with_smash?ca=drs-bl |
| 22:46 | <hsivonen> | is that for real with existing browser implementations? |
| 22:48 | <annevk> | Hixie, true |
| 22:49 | <hsivonen> | Is Raman disagreeing with something I didn't actually say or is he making a more general point about TeX and role=math? |
| 22:55 | <annevk> | hmm, http://meyerweb.com/eric/thoughts/2008/03/27/acid-redux/#comment-344298 |
| 23:03 | <jgraham> | Hixie: "The tools will save us" could work if the tool in question is the browser itself |
| 23:04 | <annevk> | for some things tools do work, such as PNG, JPEG, and GIF |
| 23:05 | <annevk> | (though when I need simple graphics I find myself using <canvas>, to be fair) |
| 23:05 | <jgraham> | specifically I think having byzantine rules for start tag inference in a format that is partially XML is a bad idea |
| 23:05 | <annevk> | HTML is "partially" XML |
| 23:05 | <jgraham> | On the other hand treating math like a particularly complex microformat might work OK |
| 23:06 | <jgraham> | annevk: There is really nothing in HTML to compare to the idea of converting <mrow>1 + 2</mrow> to three elements |
| 23:07 | annevk | is satisfied with the <tbody> sample |
| 23:07 | <jgraham> | Do you think authors even know that <tbody> exists? |
| 23:08 | <annevk> | I don't think they do and I don't think it's much of a problem |
| 23:08 | <annevk> | It started being a problem when people started using XHTML |
| 23:08 | <annevk> | but then XHTML is a lot more complex anyway |
| 23:09 | <hsivonen> | annevk: tbody is a problem with child selectors and DOM scripting, too, if you don't know it is there |
| 23:09 | <jgraham> | Right so it's not really comparable, unless you can come up with a maths format where e.g. <mo> is _never_ needed |
| 23:09 | <annevk> | I don't think authors will know or care that <mo> is there |
| 23:09 | <annevk> | they just want to write down the Math and have it look decently |
| 23:09 | <Dashiva> | hsivonen: I can't find a single page saying anything about what smash actually is, just how awesome it's going to be. |
| 23:09 | <hsivonen> | I'd just go with MathML elements without start tag inference and leave it to authors to choose their favorite tools to save them |
| 23:10 | <annevk> | I'd like my text editor to save me :) |
| 23:10 | <hsivonen> | what does OpenAjax Alliance actually do? |
| 23:10 | <annevk> | i've no idea, but they discuss stuff |
| 23:10 | <Dashiva> | It sounds like something MS started :) |
| 23:11 | annevk | wonders if anyone else just got an e-mail he didn't |
| 23:11 | <annevk> | s/anyone/everyone/ |
| 23:12 | <hsivonen> | annevk: I pasted a URL a while ago: http://www.ibm.com/developerworks/blogs/page/woolf?entry=secure_mashups_with_smash?ca=drs-bl |
| 23:13 | annevk | finds http://www.openajax.org/member/wiki/Steering_Committee |
| 23:14 | <annevk> | "NEXAWEB/Coach Wei" was pushing XAL at some point in the W3C WAF WG... |
| 23:19 | <hsivonen> | XAL? |
| 23:21 | <annevk> | http://www.w3.org/TR/dfaui/ |
| 23:21 | <othermaciej> | Dashiva: IBM started it I think |
| 23:24 | <hsivonen> | annevk: thanks. |
| 23:24 | hsivonen | thinks HTML5 possibly with the XUL flex model in CSS is the right way forward |
| 23:24 | <sayrer> | yeah, need some box model |
| 23:25 | <annevk> | yeah, who's going to do that? |
| 23:25 | <sayrer> | I think someone at Mozilla is going to |
| 23:25 | <annevk> | awesomeness |
| 23:25 | <sayrer> | but I forget who that is |
| 23:25 | <annevk> | there are some alternate proposals for grid layouts but i think i prefer flexbox |
| 23:26 | <annevk> | others being http://www.w3.org/TR/2007/WD-css3-layout-20070809/ and http://www.w3.org/TR/2007/WD-css3-grid-20070905/ |
| 23:27 | <hsivonen> | nn |
| 23:27 | <Hixie> | who's the guy dan got to add the issues to the html issues list again? |
| 23:28 | <annevk> | i think he's more busy now |
| 23:28 | <Hixie> | shawn? |
| 23:28 | <annevk> | yeah |
| 23:28 | <annevk> | medero or so |
| 23:28 | <annevk> | yeah, that's it |
| 23:28 | <annevk> | (fwiw, the style attr issue was added by jgraham already) |
| 23:30 | <Hixie> | ah ok |
| 23:30 | <Hixie> | nevermind then |
| 23:38 | <jgraham> | Hixie: You can add issues to the issues list fwiw |
| 23:38 | <Hixie> | yes |
| 23:38 | <Hixie> | i don't especially like doing so though :-) |
| 23:38 | <jgraham> | OK |
| 23:38 | <Hixie> | <mprescripts> is a weird element |
| 23:40 | <annevk> | where our you locked with the MathML stuff btw? |
| 23:40 | <annevk> | or did you find a way to move forward? |
| 23:42 | <Hixie> | i have the following problems: |
| 23:43 | <Hixie> | * mathml content + presentation involves redundancy that i want to avoid, due to the known issues of the antipattern of including data twice |
| 23:43 | <Hixie> | * presentational mathml is way too verbose too hand-author sanely |
| 23:43 | <Hixie> | * some people want both mathml content + presentation to be included in html5 |
| 23:44 | <Hixie> | * some people want no inferred tags in anything we add to html5 |
| 23:45 | <Philip`> | (PNG is way too hard to hand-author too) |
| 23:45 | <annevk> | PNG is pretty easy with <canvas> nowadays :) |
| 23:45 | <Philip`> | Okay, JPEG then :-p |
| 23:45 | <annevk> | Hixie, it seems only the first two are real problems |
| 23:45 | <Hixie> | * non-mathml-syntax solutions, e.g. ASCIIMathML, are incomplete compared to MathML, and would require really hard work to support in teh html5 parser |
| 23:46 | <annevk> | oh, that too :) |
| 23:46 | jgraham | invokes the priority of constituencies :) |
| 23:46 | <Hixie> | so i don't know what to do |
| 23:47 | <Hixie> | jgraham: ASCIIMathML is pretty non-intuitive for anything non-trivial |
| 23:51 | <jgraham> | Hixie: it doesn't look so bad to me; at least from the point of view of people who already know LaTeX |
| 23:53 | <Hixie> | the issues with doing something like ASCIIMathML or LaTeX are: |
| 23:53 | <Hixie> | * very complicated parsing rules, far more so than anything so far in html |
| 23:53 | <Hixie> | * isn't MathML, so we lose any kind of even remotely sane round-tripping |
| 23:53 | <Hixie> | * very unclear how to style it |
| 23:54 | <Hixie> | * incomplete compared to MathML |
| 23:54 | <jgraham> | Hixie: It has to map into a MathML DOM internally |
| 23:54 | <Hixie> | right, that's assumed |
| 23:54 | <Hixie> | that's actually where most of the problems come from |
| 23:55 | <jgraham> | I think that makes the round-tripping no worse than the almost-MathML-as-XML case |
| 23:56 | <Hixie> | if we use mathml, you can do this: |
| 23:56 | <annevk> | would the processing be on the tokenizer level or post processing during the parser stage? |
| 23:56 | <Philip`> | Hixie: I'd guess http://www.whatwg.org/charter.pl should be updated |
| 23:56 | <jgraham> | The styling thing isn't so critical for Maths but it can be solved using Firebug |
| 23:56 | <Hixie> | use a tool, paste into text/html, right-click and choose mathml source, paste into mathml tool, repeat |
| 23:57 | <Hixie> | if we use a mathml incompatible syntax, we lose that |
| 23:57 | <Hixie> | i really don't think we should be relying on tools to solve any of these problems |
| 23:57 | <Hixie> | we have historically seen that that simply doesn't work |
| 23:57 | <Hixie> | Philip`: thanks |
| 23:58 | <jgraham> | I tend to agree but I think MathML is so bad to hand author that you need to have a tool either for the authoring or for some other aspect of the process |
| 23:59 | <Hixie> | i've hand-written mathml |
| 23:59 | <Hixie> | if we could simplify it a bit, it would be ok |
| 23:59 | <jgraham> | Me too. That's what I base my position on :) |
| 23:59 | <Hixie> | but people are against simplifying it |