If A connects to B, and B wishes to pass the connection info
to C, B only has the address with which A accessed B. If A
and B are in the same site and used site-local communication,
and C is not in that site, B does not have enough information
to refer C to A. Hence, your suggestion does not solve this case.
I agree, but is it reasonable for B to tell C to talk to A, or is it
more reasonable for B to tell A to connect to C by the 'name blob' it
would have used to contact C?
let me rephrase the question. is it reasonable to expect that every 'blob'
that can be used to derive a set of locators for a host is tightly bound to
that host? or is it more reasonable for the network infrastructure to
provide a token, independent of any application-specific blob, that can be
tightly bound to that host?
back in the mid 1990s I had a system for looking up URLs or URNs in a
distributed database which would return blobs that could be used to find
accurate replicas of those resources. those blobs included IP addresses in
order to speed up access (requiring clients to do extra DNS lookups
drastically slowed the system down). but those blobs weren't bound to hosts,
they were bound to resources, and the mapping between resources and hosts was
(by design) arbitrary.