(WG chair hat off.)
On the occasion of the WGLC, I have done a detailed review of ND-13.
I think the specification is mostly solid, but there are a couple of open
issues.
My main points are:
-- The interaction of the C-bit and the lifetime of 6COs is not clearly defined
(or, if it is, it is nonsensical). Section 7.2 defines 6LBR behavior and
cannot be used to disambiguate 6LR or host behavior.
-- It is not clear whether hosts make use of the sequence number in an ABRO.
This compounds the previous point, in particular with indefinite lifetime
settings, possibly leading to inconsistency.
-- Some of the language around optional/mandatory is unclear (this is a
superset of Jongsoo Jeong's point). Saying that message X is optional is
meaningless. The specification must be clear for each kind of node whether (or
in what circumstances) that node is expected to be able to generate message X
and/or understand/act on message X.
Simple example: 6CO cannot be "optional" if there is context the hosts are
expected to act on and which they haven't received through some
pre-configuration.
-- The language around NCE types is not entirely clear, possibly leading to
some of the recent discussion about NC table collisions.
Since my detailed, line-by-line comments are too big to send through the
mailing list, I have archived them at the 6lowpan repository. Please see
http://svn.tools.ietf.org/svn/wg/6lowpan/nd/draft-ietf-6lowpan-nd-13-cabo.txt
(Everything that is not indented is a comment.)
Gruesse, Carsten
_______________________________________________
6lowpan mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/6lowpan
