> > (2) I wondered why the document said hosts MUST use port 4590. > > Certainly having a well know port is useful in many cases. But, I > > don't see why some consortium couldn't decide they were going to use > > port 4545 or whatever. Likewise, when setting up a callback it'd > > seem straightforward to give a port number, as well. > > > > I am not sure it is the biggest deal in the world, but a solution > > that leveraged late binding would strikes me as more flexible and > > hence better. > > Agreed that the restriction is too strict. However, I think it's > simpler at this point to declare the use of ports other than the > well-known port to be configurable out of band, as other RID system > and consortium configuration issues are out of scope for this document > as well.
That is what I was thinking. I.e., I was not suggesting some way to discover the port or whatnot. Just not mandating that all communication happen on the assigned port (which of course you could never do regardless of what the RFC says). Sounds like a good set of changes to me. allman
pgpwlDxSqD6Nj.pgp
Description: PGP signature
