Thomas Narten wrote:
> [...]
> The current ID defines some ranges of address for special purposes,
> but says the rest should be treated (from an implementation
> perspective) as global unicast. I.e., treat them all the same.
>
OK. Your clarification is useful, since this is really what I had in
mind.
Perhaps the text in the draft could be made more clear - something
closer to
yours.
> [..]
> As Randy has pointed out, the IANA considerations section in the ID
> makes it clear that only 1/8th of the available space is being made
> available for usage and assignment. So we still have lots of space to
> use differently, should someone make the case that we need to.
>
Two comments relative to this - already made earlier.
a.
Since I missed the earlier opportunity, I am bringing this feedback now,
sorry.
I would have been happier if the clarification was up front, like having
a section on address space partitioning, like I suggested in some text
sent earlier, or some text at
the beginning in the IANA section, rather than some text in a
"foot-note".
b.
The IANA considerations text is in Appendix C, after 2 informational
appendices.
Shouldn't the section be "normative", and part of the main document,
rather than an appendix ?
> > Carving out such a space ahead of time, like in IPv4, worked well. It
> > provides the benefit that if people experiment with a number of
> > addresses, they know ahead of time, what they have available, and will
> > not conflict with global addresses, or other addresses, already
> > in use.
>
> Actually, reserving an experimental range in advance, is not the best
> way to do it. You might reserve too much or too little. Hard to tell
> in advance what the future will bring.
I am not talking about a large chunk. A tiny space can always be
extended, if someone needs bigger space, for a new thing, which seems
more than just an experiment, or it can ask for a separate new
assignment. Having that tiny space ahead of time, saves the time of
going and asking for it, in particular for temporary experiments, kind
of like what D, and E, where in IPv4.... The effort of getting an
assignment should take place only if the assignment is
(quasi-)permanent. This is my opinion, based on past and current
experience.
> [..]
> Thomas
Alex
S/MIME Cryptographic Signature
