IPv6 address/port format

[Mark Thompson]

On 08:41(GMT) 18-01-00, Michael H. Warfield wrote:
> > we don't need to...
> > When we write IPv4 compatible address it can look:
> > :k.l.m.n:o.p.q.r
> > but it can't look:
> > :k.l.m.n:o.p.q.r.s
> 
> 	Ok...  Pop quiz.
> 
> 	::130.205.1.143
> 
> 	Is that port 143 on system 130.205.1.0 (host 1.0 on network 130.205)
> or is that host 130.205.1.143?  It's a rare case, but legitimate.  In
> IPv4 notation they allow 130.20.1 to be the equivalent of 130.205.1.0 in
> full.  This is BUTT UGLY, IMHO, but apparently legal and I've actually
> encountered some people using it and pointing it out to me.  So this one
> particular instance is ambiguous, unless we specify, up front, that all four
> of the dotted quads are specified in full for IPv4 compat notation.

OK, so why not borrow some semantics from prefixlen?

::152.78.65.197/32:23

IPv4 compatible address for IPv4 host 152.78.65.197, which is host 65.197
on network 152.78 (32 bit prefix length), port 23

Fairly succint in that it is not introducing any new concepts (IPv4 uses
colons to delimit address from ports). It means that if we're not always
expressing the prefix length, we end up with ::152.78.65.197/:23

It gets around the problems highlighted so far, but perhaps isn't as
eligant as what people are expecting. Needs more thought :-)

Mark/
-- 
details at: http://www.ecs.soton.ac.uk/info/people/mkt