- From: Henri Sivonen <henris@clinet.fi>
- Date: Thu, 20 Dec 2001 13:34:48 +0200
- To: www-html@w3.org
>Henri Sivonen <henris@clinet.fi> writes: > >> >There is a serious roadblock for the use of MathML in web pages >> >arising from the differing expectation of user agents in regard to >> >the HTTP content-type for a web page with such content. >> >> Which user agents have which expectations? > >I was refering to a plan for portable documents that had received >discussion chez mozilla much earlier this year. That plan involved a >user agent and a third party plugin specific for MathML. For that to >work the XHTML+MathML document would require a small amount of baroque, >but legal, markup to provide hooks for the plugin and the object must be >served as text/html. AFAIK no implementation has been released. I don't think it would make sense to increase complexity in order to cater for ugly ad hockery--especially if no implementation has been released. Of the old browsers with some kind of plug-in architecture, Netscape, Opera and Mac IE, AFAIK, can't provide arbitrary document subtrees as data for plug-ins. Is this all about "supporting" the "XML data island" stuff in Windows IE? "XML data islands" inside text/html soup are already considered harmful. >> How would HTML extended by MathML be parsed? > >Probably not well. I think that's a reason enough to not endorse MathML as text/html. >A content provider following this migration path >might want also to provide TeX source or an anchor to an image (rather >than an inline image). > >But, as an example in answer of the question: see the browser lynx. How would sending MathML content to Lynx be useful? Lynx doesn't how to display MathML. The result wouldn't be accessible--just a broken rendering. >And please don't forget that many content providers are now by site >law required to provide *accessible* content. How would dumping MathML as text/html promote accessibility? If MathML source is fed to a text/html user agent, the result is a broken rendering that is useless for determining the mathematical meaning. Mozilla and Amaya work for people who can see. As I understand it, you are referring to accessibility for users who can't read math visually. Until someone produces an aural or tactile user agent for XHTML + MathML, I think the accessible solution is explaining the math as plain text and having that rendered in braille or read aloud. (If/when someone does produce an aural or tactile user agent for XHTML + MathML, I think it is safe to assume it can receive text/xml.) >The proposal is about flexibility for content providers. > >> With Mozilla and old non-math browsers, this effect can already be achieved >> using Accept header-based content negotiation to decide between >> application/xhtml+xml (or text/xml) and text/html on the server side. > >As a content provider I'm just not going to provide dual html >resources in addition to pdf. If I must choose between text/html and >application/xhtml, then it's going to be text/html until I estimate >that 75 % of my readership is fully ready for application/xhtml. I think adding complexity to the type determination would only further hinder the implementation of MathML. Requiring support for MathML as text/html would certainly be a problem for Mozilla. I think the MathML would be better promoted by helping to get Mozilla's MathML implementation into a shippable state and lobbying Netscape to ship it by default in Netscape 6.5 and by lobbying university IT depts to install MathML-enabled browsers as the default browsers on computers in university labs. >I've never been impressed with forms of robotic content negotiation, >even if they manage to penetrate caches, proxies, and firewalls, that >obstruct the relationship between content provider and user. In what way? -- Henri Sivonen henris@clinet.fi http://www.hut.fi/u/hsivonen/
Received on Thursday, 20 December 2001 06:35:48 UTC