Cullen,
Having the same NodeId size on the wire as in computing memory makes
programming much easier. Also, defining NodeID the same way as
ResourceId makes this draft consistent, extensible, and again easy to
implement:
typedef opaque NodeId<0..2^8-1>
typedef opaque ResourceId<0..2^8-1>
I know that having a variable NodeId may be confusing with the topology
notation "chord-128-2-16+", but the draft could say that
"chord-128-2-16+" means peers must guarantee the most significant 128
bits of the SHA1 hash, or simply use the "chord-sha1-2-16+" notation.
Thanks
--Michael
Cullen Jennings wrote:
> In discussing this with people, they often say that they would like
> the NodeID to be the same size of the 20-byte hash but I really don't
> understand the logic of that. We need the NodeID to be large enough
> that we don't run out of NodeID but 2^128 seems more than enough for
> nodes in a single overlay. Is there something I am missing here?
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip
