Re: FW: Application for port-number (system-klensin)

[ned.freed@mrochek.com]

> > > I don't know what the rules are, but if we can delay it, I think we
> > > should.  Building a new protocol based on this view of message
> > > sequences at this point just doesn't seem like a good idea.
> > > I'd like to see the draft and talk to John before we approve it.

> I am having a hard time understanding why a remote routine would be used to
> perform this function.  I understand that Paul Hoffman wrote an
> implementation in perl so that people could play with it and determine if
> it is really going to be useful.  It is a command line hack that takes a
> port number as an argument, so it can be used without an IANA
> assignment.  I would like to hear the results of this experiment before a
> port number is assigned.

I think the logic is that the necessary table needed to perform such 
conversions can be large and/or require regular updates, so it may make sense
to perform the operation remotely for some memory-constrained and/or
hardcoded devices.

Irrespective of whether or not such argument are credibie, however, I am
concerned that the way the protocol is parameterized is inadequate. Fore
example, take a look at the work Chris Newman is doing on comparators: There
seem to be more knobs needed that this protocol provides.

This leads me to the conclusion that experimenting with this is fine but
assigning a system port number to it is premature.

				Ned