Re: Support for 240/4 and 0/8 addresses in NetBSD

[Havard Eidnes <he%uninett.no@localhost>]

> Stepping back from optimizing code, I think there are unanswered
> questions:

I'll present my take:

>   A) Are any of the addresses that are currently not allowed by standards
>      actually in use?  [seems like obviously not]    If not, what is a
>      rational timeframe for them being issued and used?

Probably not.

Rational timeframe?  Well, this goes to the deployment considerations
for making a non-backwards-compatible change such as this.  How long
do you wait for any changes to percolate through standardization,
development and actual deployment for the myriad of IP stacks in
user's hands out there?

If you start using this "too early", interoperability problems will
appear, and I would claim that one of the important guiding principles
for the IETF is "maximal interoperability".  Therefore, I predict that
this change is going to meet resistance on the grounds of the interop
problems (and therefore cost) it will cause.

>   B) Are the code changes on the table fundamentally:
>      1) aligning to emerging standards?

That would be a stretch.  Some people have started advocating for
changing the status of these certain address blocks, and I think there
was an individual submission, but so far no WG adoption.

>      2) part of a process that argues that the standards should be
>         changed?

Pass, but see above.

>   C) We hear that the documents are not yet standards-track work items,
>      and people offlist have told me (paraphrasing vastly) that this is
>      because "IETF complicated" and "IETF more political than technical
>      these days".  If that's true, then it seems that for this to become
>      reality so that A is true, documents will need to progress through
>      standardization, to the point where registries are willing to
>      issues addresses.  Assuming that's true, why is it useful for
>      NetBSD to do anythning now, aside from stepping into IETF politics
>      via B)2) ?

I think the main source of the resistance is the interoperability
problems such a change would cause, and that people advocating for
this appear to under-estimate how long this transition phase will have
to last.  And ... the amount of addresses "released" by this change is
so small that many people make the judgement that the total system-
wide costs for such a change far outweighs the benefits.

And, I agree, we should leave the changing of the standards to the
IETF, and not become a protagonist for the proposed change by
implementing it long before there's even agreement to work on the
change within the IETF.

Regards,

- Håvard