Re: [dccp] dccp draft | additional note to section 5.8.

[Arnaldo Carvalho de Melo <acme at ghostprotocols.net>]

On 2/8/06, Hagen Paul Pfeifer <hagen at jauu.net> wrote:
> * Eddie Kohler | 2006-02-08 09:07:41 [-0800]:
>
> >I would say that implementations MUST process all options sequentially, and
> >unless otherwise specified (as in Ack Vector and Change/Confirm), in the
> >case of multiple options, the last takes precedence.
>
> ACK
>
> >From me point of view this formulation is much clearer.
> Additional: for a dccp implementation it is also easier to take the last one
> (if multiple options occur).
>
> I also think this doesn't restrict the spec not to much, in contrast it make
> it clearer! ;-)

:-) Cool, the Linux implementation works just like Eddie described,

Ah, as I hadn't mentioned this here: linux 2.6.17 (David Miller tree,
to be merged
when 2.6.17 opens) has ack vector records, feature negotiation and an
initial CCID2
implementation, courtesy of Andrea Bittau :-)

- Arnaldo