In your previous mail you wrote:
Just to make things clearer, are you:
- agreeing with the document, while at the same time reminding the
general principle that "the link-layer should adapt, not the
network layer", or
- disagreeing with the document based on that same principle
If the former then I'd agree with you.
=> it is the former because by applying the principle it lacks
of any real justification, i.e., it is some kind of transient
disagreement.
But if the latter then let me disagree. I don't think the document
proposes that the network layer adapt to the link layer. My
understanding is that the document is relaxing some constraints on
the *configuration* of the network layer (rather than modifying the
*protocols*), and AFAICS this is fine if it does not break operation
of the protocols themselves.
=> but there is no justification when the document is applied to
a general link layer.
In that specific case, when an interface's link layer is such that the
router never changes, it makes sense to allow for an infinite
lifetime in the configuration.
=> too large lifetimes are as stupid as too small lifetimes.
We have crazy timing from people using RAs as a beacon,
IMHO we already accepted too much.
But perhaps you have in mind a specific scenario that would be broken
by the relaxation of configuration constraints?
=> I have not to find one, the document has to be justified
(in French I have not the "charge de la preuve").
Regards
[EMAIL PROTECTED]
PS: IMHO the current default and maximum lifetimes are pretty good...
--------------------------------------------------------------------
IETF IPv6 working group mailing list
[email protected]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
