The problem is that
accept: text/html
doesn't indicate which version/profile/release/level of "html" you
want. This is a problem for html now, but it's also a problem for
application/postscript in the face of Postscript level 2, for
image/gif in the face of embedded JPEG in GIF94, or even in the
negotiation over color, black & white, embedded images, etc.
I urge everyone to avoid releasing documents that use non-standard
extensions to HTML *as* HTML. If people want to extend their browsers
to accept some other document type, that would be lovely.
I urge Mosaic Communication Corp to 'fix' Mozilla to include a header:
accept: text/x-mozilla-html
and to fix their server to label their extended documents in that
form; this would allow servers to conditionally release different
documents depending on the content type, just as the web was
originally designed.
This would is a responsible and compatible way of releasing
experimental and non-standard extensions to HTML in any browser.
Don't call it 'html' until it actually *is* HTML.
If you want to infer that file:///localhost/foo.html is really
content-type text/x-mozilla-html, that's fine, too.
You know, there's good reason to get new features out for experimental
use! It's just that it's really easy to do in a way that is actually
compatible with the standards that are already released.