- From: Justin James <j_james@mindspring.com>
- Date: Mon, 31 Mar 2008 12:27:11 -0400
- To: "'Bruce Miller'" <bruce.miller@nist.gov>, "'Henri Sivonen'" <hsivonen@iki.fi>
- Cc: "'David Carlisle'" <davidc@nag.co.uk>, <ian@hixie.ch>, <public-html@w3.org>, <www-math@w3.org>
> Can someone please fill in some of the gaps, here? > I get the feeling there's a stage(s) where "Magic Happens"... I could not agree more. Right now, we are only accounting for MathML. Even when we discuss this issue as not MathML-specific, we are thinking only in that context. What's to happen if I come up with my own DOM and want to embed it? Do I also have to account for, in my DOM, these kinds of issues? Do I submit it to the W3C and hope that the handling rules make it into HTML 6? I propose a much simpler solution that is properly within the spirit of HTML. Allow the HTML author to embed an OBJECT reference to a parser that handles their format. Problem solved. Just because MathML (or *ML) happens to use the XML syntax, what right do their authors have to expect that an *HTML renderer* will be able to handle it? Why should everyone who writes an HTML consuming library or application have to account for every DOM that we either hardwire in at this time, or allow to be dynamically used? Why not treat non-HTML DOMs the same way we would treat any other non-HTML file format, as either something that can be embedded with OBJECT, or something that the user agent can download and let the OS handle? In other words, what makes MathML (or *ML) so special that we feel the need to cater to it in HTML 5? It is *impossible* to properly handle this without a "magic happens" style spec, or creating some giant kludge of an embedded *ML handling system that allows the authors of "foreign" DOMs to use some standardized scripting to control the HTML user agent. In other words, we would need to re-create something like ActiveX, Silverlight, Java applets, Flash, etc. to allow DOM authors to provide the full handling implementation to an HTML user agent in a standardized way. Or we can just use the existing, working, and standardized plugin system. I vote for the latter. J.Ja
Received on Monday, 31 March 2008 16:33:19 UTC