Re: NHRP/MARS version number location.

[James Luciani <[e-mail]>]

> > > However, as you show below, the MARS header doesn't really have a
> > > version number...
> > 
> > What is ar$op.version then? 
> > 	[...]
> 
> As you say below, it identifies a protocol, not a version of a protocol.
I am honestly mystified as to why you feel that we cannot have versions
of address resolution protocols.  ar$op.version=2 is not defined yet but
it could be the next NHRP or the next MARS or it could be an address resolution
mechanism that solves a separate address resolution problem but does not 
want to inflict the writing of new parsers, state machines, etc. unnecessarily
on some poor defenseless programmer.  If the variation from the existing
address resolution mechanisms is so great that the existing machinery can't
be used then it will have a separate LLC/SNAP code point.

> I don't believe that NHRP and MARS are versions in the sense that one
> in intended to replace the other.
No!  They are intended to merge, IMHO, at some point in the future but
their respective aims at present are too diverse to be included in 
the same document.  Basically we need to know a bit more about both 
technologies before we can realistically expect them to work and play well
with others.

> (Do we need to solicit opinions from the broader IETF community about
> the appropriate definition of the term "version?")
I assume your kidding right?  I assume you are not suggesting
that we waste a couple of terebytes of disk space with email over the 
sophistry of the deeper, truer meaning of the word "version".
Sounds to me like I need to brush the dust off "Being and Nothingness"
over New Years :-)

> > > So, it appears that what you have been calling a "version" number is
> > > really a protocol identifier.  In other words, there is no protocol
> > > version number within the MARS header, only a [mislabeled] field which
> > > indicates the protocol.
> > 
> > Again, you missed the point of the harmonization of the NHRP and MARS.
> > The version number identifies the instance of the address resolution
> > protocol.
> 
> It sounds like you just affirmed my assertion that the MARS "version"
> number is actually a protocol identifier.
Again... It is the version of the address resolution protocol being used.

> As I mentioned earlier, you might find it instructive to examine the
> use of the version field other Internet protocols.
To be perfectly honest, you might find it instructive to note that (as I 
pointed out in a previous note) THE Internet Protocol has a 
differenct PID between v4 (PID=0x0800) and v6 (PID=0x86DD). And, you'll note
that the fixed part (the IP header) is so different that any attempt at parsing
them in the same way would be futile.