On 18/05/2017 06:04, Michael Richardson wrote:
>
> Brian E Carpenter <[email protected]> wrote:
> > We need to resolve this issue quickly, to get a new draft out in the
> next
> > few days, before the IESG meeting next week.
>
> > So, since I haven't seen any opinions on the issue below, I propose to
> > delete the UDP paragraph as suggested below. Note that this does not
> > prevent us adding a UDP mode later, but it removes the confusion that we
> > have in the text today.
>
> So, we will (initially) support UDP for multicasted Link-Layer M_FLOOD and
> M_DISCOVER, but that's all. Everything else will be TCP.
Until we write a profile for GRASP-over-UDP or GRASP-over-DECnet or whatever
we want. The GRASP protocol model doesn't really care.
> (In my opinion, inside the point to point mesh ACP, one might as well keep a
> single TCP connection up for every M_DISCOVER and M_FLOOD, as UDP has no
> advantage there)
As long as it behaves like a multicast, it really doesn't matter how the
ACP chooses to implement it. For a stand-alone specification of GRASP,
I don't see how else to specify it, though.
Brian
_______________________________________________
Anima mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/anima
