Document: draft-holbrook-idmr-igmpv3-ssm-07.txt
Reviewer: Spencer Dawkins
Date: 29 augusti 2004  


This specification, taken on its own, is mostly ready for publication
as a proposed standard. I could implement the protocol from the
specification, and it's reasonably clear what's going on.

My only question is (echoing Margaret's question in the ID Tracker),
"why was this specification not written as updates to RFC 3376 and
MLDv2, especially since MLDv2 wasn't an RFC yet?"

Call me sensitive, but we have an entire working group figuring out
what TCP really is beneath all the independent specifications that
change TCP behavior as specified in RFC 793. Do we want to go here
again?

MLDv2 has been published as RFC 3810 (in June), so the reference needs
to be updated.

I guess I'm also curious about why hosts SHOULD log errors, but
routers MAY log errors. I'm guessing that the idea was that the hosts
are the ones causing the problems, so they should work harder to make
sure someone notices logged errors, but this seems like a throwback to
a simpler time - we can't get people to stop clicking on .pif files,
or .scr files, or .zip files, so shouldn't we try harder to make sure
network operators know there's a problem? But I thought SHOULD was "in
most cases, MUST, but this doesn't make sense in all cases", and am
trying to figure out what the corner cases might be.

Spencer