Re: [Hipsec] WGLC: draft-ietf-hip-native-nat-traversal

[Derek Fawcus <[e-mail]>]

On Mon, Mar 07, 2016 at 02:35:07pm +0200, Gonzalo Camarillo wrote:
> First he will look into adding clarifications to the existing draft
> while still referencing the old RFC. If the group is not happy with the
> readability after the editorial pass (or our AD does not finally let us
> downref the old RFC), we can consider bringing material from the old RFC
> directly into the new one.

Sorry,  that I'm quite late in looking at these,  but have been doing
so recently...

I have to say that I find the it difficult to decode simply because
of having to refer to 3 (the draft, 5770, 5245) plus possibly the
STUN/TURN docs at once.

I'd certainly find it easier to comprehend if the text from 5770 was
incorporated (suitably modified to account for not doing STUN/TURN)
within the draft.  That way the references to the significant pieces
of 5245 text would be easier to nail down.

As it is,  I currently find it a bit like reading an Act of Parliament!

e.g. $3.8 Connectivity Checks
   refers to $4.6 of 5770 with some exceptions, $4.6 of 5770 refers to
$5.7 of 5245 and $7 of 5245,  where the exceptions (use of UPDATE instead
of STUN) have to be applied to that $7 referencing 5389,  so possibly
I don't have to read 5389, since hopefully it would just be packet formats.

> I would also like the group to comment on the following two proposals:
> 
> 1) the draft will allow implementers to use HIP native relays only. In
> addition, the use of STUN and TURN relays will be optional.

I'd suggest the draft be native only,  but say with an appendix referencing
5770 for use of STUN/TURN,  maybe indicating which bits of the 5770
to take heed of.

> 2) in addition to covering the base exchange, the draft will also cover
> the mobility readdressing exchange.

Not having read that recently,  I can't really comment.

DF