Bill Mills wmills@invisibleworlds.com
Fri, 15 Jun 2001 16:12:02 -0700

I do not then see how the dotted notation proposed in one of the earlier
messages on this thread is any better.  If I read 2396 right the scheme name
is parsed complete, and would not itself be tokenized.  I think using
beep.<proto>:<specifier> still implies a combinatorial explosion of schemes
on the left had side?  

It would seem cleaner then to leave the protocol to be looked up with an SRV
lookup i.e.


Is this more palatable?

-----Original Message-----
From: Eamon O'Tuathail [mailto:eamon.otuathail@clipcode.com]
Sent: Friday, June 15, 2001 2:27 PM
To: bxxpwg@lists.invisibleworlds.com
Cc: wmills@invisibleworlds.com; wharold@tivoli.com
Subject: [BXXPwg] SOAP over BEEP

>> The vector that I thought needed to be dealt with is the underlying
>> transport mapping.  Let's leave it at TCP and SCTP and ignore SSL for
>> now then.  The question stands.

If we are only ever going to have two transport mappings, I would agree,
but what happens when (not if) we have twenty transport mappings (over the
ten years). RFC 3080 is careful to separate out the transport mapping from
the BEEP core, we really need to follow this good practice with the URI.