Re: Follow up on charter proposal

Martin Hamilton (martin@mrrl.lut.ac.uk)
Fri, 07 Jul 1995 21:22:39 +0100


Message-Id: <199507072022.VAA11224@gizmo.lut.ac.uk>
To: uri@bunyip.com
Subject: Re: Follow up on charter proposal 
In-Reply-To: Your message of "Fri, 07 Jul 1995 13:18:46 EDT."
             <199507071718.NAA08293@beethoven.bunyip.com> 
Date: Fri, 07 Jul 1995 21:22:39 +0100
From: Martin Hamilton <martin@mrrl.lut.ac.uk>


Leslie Daigle writes:

| [Martin Hamilton writes:]
|
| > How are we doing on _implementations_ of the URI proposals ? :-)
| > 
| 
| Well, now, that would depend which URI proposals you mean... URAs have
| a prototype implementation that will be demo'ed at Stockholm.

Yep, nice piece of work!

| Oh, or did you mean URNs?  Aye, now there's the difficulty.  We can't
| have an implementation until we have an agreed-upon syntax...

Sigh... !

| But, this is part of what I've been getting at -- the progress on 
| a particular proposal in the working group is inversely proportional
| to the perceived importance.  This is normal, and good, if it implies
| that the progress is related to the amount of care in doing things
| appropriately in accordance with importance.  However, it seems
| we are just bogged down with URNs.  And, it isn't clear to me that
| we are correctly matching "perceived importance" with "actual importance"
| of proposals as a group.

Here's an idea for unjamming URNs ...

For background info, see <URL:http://inet.nttam.com/HMP/PAPER/173/>

This paper proposes a new DNS record "DX" (Directory eXchange), which 
can be simulated for the moment with "TXT" (TeXT) records:

  rs.itd.umich.edu ->
    TXT "dx: 0 ldap://terminator.rs.itd.umich.edu:6666/"
    TXT "dx: 10 ldap://vertigo.rs.itd.umich.edu:6666/umich.edu"

When presented with a name such as

  tim@rs.itd.umich.edu

your white pages software does a DNS lookup on "rs.itd.umich.edu", 
which returns the two DX records above, and is then free to use 
either (or neither!) of these to look up Tim's details.  This is 
quite neat, as the white pages _names_ which "rs.itd.umich.edu" is 
responsible for can stay the same while the servers pointed to by the 
DX records change name, protocol, port, etc

Abstract names such as "Tim Howes, Michigan, USA" are dealt with 
separately using centroids.  This is nice, as it would provide you 
with two clearly delineated approaches - lookups (DX records) versus 
searching (centroids).  Of course, centroids (aka the Common Indexing 
Protocol) are still under development.  Caveat emptor!

This approach would seem to be a promising way of progressing URNs

URCs are another matter altogether :-)

Wot you lot reckon ?

Martin

PS Just for fun - what would the degenerate case be (no DX records) ? 
 Talk HTTP to an IP address associated with the domain name ??  A la 
MX - but which port number ?!