[
Lists Home |
Date Index |
Thread Index
]
People built all sorts of systems with SGML that didn't
use DTDs before XML. That is why most of us knew that
it could be done. We also knew that even if DTDs were
optional, they were too useful to do without. Given
a contract based system, that is obvious. Relaxing the
constraint to ensure the contract isn't at the front of
every message sent is good sense, but do you publish
books without copyrights and other 'front matter'?
Nope.
Point taken but keep in mind that ASN.1 would
have to be reengineered and they need to learn the
right lessons, not the religion of markup or its
historical myths served up as lessons of history.
It's never been XML vs SGML: they are the same
thing. You just needed a bit of theatre to get
you to learn what the SGMLers knew by practice.
Now the ASN community should be asking itself
what its best practices are and if they still
apply. They may also want to scope down the
domain of the applications. Remember, "SGML
ON THE WEB". The implication was that in the
future, everything would be ON THE WEB, and
if it wasn't, it would be available when it
needed to be.
len
From: Eric van der Vlist [mailto:vdv@dyomedea.com]
Another reason why I'd compare ASN.1 to SGML rather than XML is that,
like SGML, ASN.1 seems to be schema centric: it looks like you can't do
anything in ASN.1 without designing a schema first like you can't do
much with SGML without designing a DTD first.
Since the breakthrough of XML versus SGML is the fact that schemas are
now optional, isn't ASN.1 a regression in this respect?
I'd say that this might be a reason why ASN.1 isn't binary XML, it would
rather be binary SGML!
|
