Re: application/xml in ietf-xml-guidelines

> > What I have a really hard time understanding is the belief that everything
> > should run over HTTP, or that everything should be named using HTTP URLs,
> > when it's abundantly clear that HTTP and HTTP URIs are a poor choice for
> > many applications - even if HTTP is only used as a mechanism to negotiate
> > another protocol.
> 
> I have no fondness for HTTP identifiers.  At the same time, I have no
> fondness for opaque identifiers, especially when used to identify
> vocabularies.

opacity of URNs is a design feature - URNs need to be opaque to do their jobs.

OTOH, I have no fondness for use of URNs to identify vocabularies either.

We've known essentially how we were going to resolve URNs since 1 November 
1995.  The biggest problem we've had in finishing the specifications is that 
URNs were so widely misunderstood that we were continually having to argue with 
people who were insisting that they have properties that would make them 
unusable for their intended purpose.

Keith