On 15 Mar 2017, at 18:59, Brian E Carpenter
<brian(_dot_)e(_dot_)carpenter(_at_)gmail(_dot_)com> wrote:
On 16/03/2017 07:14, Joe Touch wrote:
On 3/15/2017 9:33 AM, Leddy, John wrote:
Does this mean that only middle boxes, not covered by the architecture
could insert an extension header for use within the domain?
Please see my recent post about Stefano's issue. IMO, any opaque
(distributed) system that acts like a host can follow the host (node)
requirements.
The instant that system is not opaque or fails to act like a single
host, it becomes noncompliant.
That's the point. That's why the 6man WG shot down proposals to play
intra-domain tricks with the flow label a few years ago, and they
didn't even break PMTUD or IPsec/AH.
In another form, the answer to John is that there are no protocol police,
so what consenting adults do inside their own networks simply isn't an
issue that an Internet-wide spec can or should address. And for sure, the
spec for IPvN for any value of N is an Internet-wide spec.
If Stefano and colleagues describe how private domains can perform tricks
that MUST NOT be exported to the Internet, that is fine. Whether that
becomes a standards track document or an Independent Submission RFC is
another question. But IMHO it is completely orthogonal to the rough
consensus on 2460bis.
Exactly.
Tim