RE: How to use DTDs, or not to (was: RE: ACL and lockdiscovery)

OK,

here's an attempt to summarize... Almost all participants in this discussion
have been trying to "solve" the issue in different ways.

1) Remove the DTD fragments and replace them by english language
descriptions. This actually triggered this thread (see title) because in
rfc2528bis, some of the information present in rfc2518 was lost.

2) Extend/fix the DTD: we can do better with DTD notation but not much
better. However there's a readability cost associated with these changes,
and it's unclear wether that cost justifies the small increase in precision.

3) Move to XML Schema: only solves few of the problems that we have with the
DTD notation, yet has a big problem with readability in the spec.

4) Move to something else: I was proposing RelaxNG (Compact Notation)
because it solves all technical issues that we have. Drawback: some think
that the fact that it's not as popular as DTD or XML Schema makes it
unsuited for usage in a RFC.

So it seems there's no support at all for a major change. Let's fix just the
issues that we have with the DTD fragments. I'd suggest to re-read what the
ordering spec currently says and, if necessary, to clarify it. Then we can
use that in all other specs.


Julian


--
<green/>bytes GmbH -- http://www.greenbytes.de -- tel:+492512807760

Received on Friday, 17 October 2003 16:44:54 UTC