Re: draft-ietf-idmr-igmp-mrdisc-10.txt

[Bill Fenner <fenner@research.att.com>]

>i am not impressed by
>
>   The following are justifications for inventing another router 
> discovery protocol:  

The real origins of this are the same as the origins of IGMP itself.
Steve originally wanted to make group membership operations part of ICMP;
Jon didn't [I'm not sure I ever knew the exact reason; perhaps because
multicast was an experiment, perhaps because he was wary of how hosts
that didn't know about multicast would behave when receiving unknown ICMP
messages sent to multicast addresses].  Using IGMP for multicast router
discovery continues this trend (and using ICMPv6 for multicast router
discovery parallels with the move of multicast functions to ICMPv6).

Perhaps it's possible to write these reasons up in a way that doesn't
sound like we're continuing an accident of history.  Or maybe it's OK
to say that it's an accident of history and that's just why it's this way.

>what could be said is that, because this discovery mechanism uses multicast,
>discovery problems will be congruent with payload problems, which is cool.

Well, it'd be multicast whether it's ICMP or IGMP (ICMP router
discovery uses multicast too where possible..)

  Bill