On Fri, Aug 10, 2007 at 11:28:51AM -0400, Andrew Sullivan wrote: > Sure. They're both on slide 8 from the Chicago meeting. The first > was whether we ought to include references to RFC 1788, noting that > there were alternative proposals for how to do this. Nobody came > forward in support of this idea, and some people mentioned explicitly > that they thought it irrelevant. (My own opinion is also that it's > irrelevant, but this is a WG draft and I'm happy to be overruled.)
I think that experiment is old enough (and still experimental) that
to provide guidance in that direction may be a bum steer.
So, I say leave it out.
> The other was a bit of text lifted from section 3.6 of the -anderson-
> draft, which says this:
>
> Delegate all addresses in block. Do not assume that everyone
> uses ethernet.
>
> I suggested that this was a clearer and starker statement of the
> principle than was anywhere in the current WG draft, and so it might
> be included. There is language in the WG draft to this effect,
> however. Nobody came forward in support of this proposal either.
The problem is possibly more complex than 'ethernet vs everything
else'. Even with ethernet, someone may want ot name their network
or broadcast.
I'm thinking of 9 years ago, when I encountered a problem with using
the first addresses in a network delegated as a dynamic pool to dial
access equipment (so, PPP). I'm getting old now, but I vaguely
remember;
x.y.z.0/20-or-so was routed to null0 on the routers so that they
would produce ICMP unreachables (the dial access gear
wouldn't, they'd pass it back to the default router!).
These statics were redistributed into the IGP.
x.y.z.0/32 was advertised by dial access equipment via RIP, which
carried static assignments as well as dynamic.
We discovered (during my 5am shift) that, with the routers we were
using, we could not hold a more specific RIP route on the same
division as a statically held route redistributed into our IGP.
Something about having the same "name", but I really am getting too
old to remember this point with any clarity.
My rather long winded point here is, there are a lot of strange
things in networks we can't and shouldn't presuppose. To get into
them is a hairball. To overly simplify it to the point of admonishing
assumptions about ethernet is disingenuous, or at least incomplete.
So, I say keep or simplify the existing text.
--
Ash bugud-gul durbatuluk agh burzum-ishi krimpatul.
--
David W. Hankins "If you don't do it right the first time,
Software Engineer you'll just have to do it again."
Internet Systems Consortium, Inc. -- Jack T. Hankins
pgplhTLPXn46n.pgp
Description: PGP signature
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www1.ietf.org/mailman/listinfo/dnsop
