On Wed, 19 Aug 1998, Keith Moore wrote:
One certainly can't re-use the text/html label for such a profile as there
are deployed clients which deliberately don't display text/html to the user
because it's almost always unreadable (I'm using one such client).
Surely this is configurable?
Not only is it not configurable in my client, but it's quite clear that
any new text subtype will be treated as an attachment and not displayed by
a significant number of deployed clients. That means if we want a
chance at fixing the text/plain problem, the fix will also have to be
called text/plain. Quite unfortunate.
The impression I'm getting is that HTML has suffered from what I think
of as "ASN.1 disease" (though it could equally apply to a number of other
things): If a technology has a lot more features than is needed to do a
particular job, those features get used whether they're needed or not.
ASN.1 has fundamental architectural flaws which go well beyond just being
vastly over complicated. Oddly enough, XML fixes most of the major flaws
in ASN.1 without introducing too many new flaws.
The results tend to be so ugly that people are biased against any use
of the technology, even when such use is reasonable and appropriate.
If a simple profile of HTML were used in an interactive protocol where a
profile violation would be immediately rejected with an error, I'd have no
problem with it. The problem is that a profile of HTML is pointless
without strict enforcement of the profile because people will use any HTML
feature they can get away with.
- Chris