On Fri, 30 Jul 2004, Ian Hickson wrote:

> But yes, in principle, that's exactly what I'm saying. Specs should
> define all these cases.

If I understand you correctly, you are saying that HTML specifications
should define browser behavior (perhaps at the SHALL level?) for all
possible markup errors. Wouldn't this mean that we would effectively
change the language to permit everything? If you specify mandatory error
processing, aren't you actually extending the language so that the
erroneous constructs involved become correct? In practical terms, authors
could rely on the "error handling rules" to the same extent as on
rules for "correct" constructs.

And this would mean, IMHO, defining HTML as tag soup. You could write any
mixture of tags and it would have a defined meaning.

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/