[Wikipedia-l] Parsing TeX to MathML

[Brion Vibber]

Jens Frank wrote:

>I took a deeper look at MathML today. One thing seems to be a showstopper 
>in the Moment: For Mozilla to render the MathML extensions the page must
>have Content-Type "text/xml" instead of "text/html". But in XML many of
>our pages are not valid, e.g. a line break has to be coded as <BR\> in
>XML instead of <br> in HTML. Mozilla's parser will stop and create an 
>error message as soon as it finds an unmatched tag like <br>.
>
All the more reason to wipe out HTML...

In the particular case of, eg <br>, it would be elementary to transform 
them to <br /> when we find them. Other cases could be similarly dealt 
with, to a degree (there's already limited logic to try to keep paired 
tags matched); but it's still pretty likely that as long as we allow 
HTML it'll be hard to guarantee that we can produce a well-formed XHTML 
document from it.

Since the "correct" response to a malformed XML document is a rude error 
message rather than graceful failure, that may not be someplace we want 
to delve just yet.

-- brion vibber (brion @ pobox.com)