Jari Arkko wrote:

[EMAIL PROTECTED] wrote:

Perhaps the only question that I have left is... whether
IPv6 nodes should support MLDv2 instead of MLDv1. The
node requirements document and the MLDv2 draft intro
gives the impression that the only difference is
source-specific filtering. Then the current node
reqs rules make sense. But Appendix B of the MLDv2
draft seems to list a number of other modifications,
including some enhancements to robustness. Are these
important? I'm not an expert enough to determine that.



Appendicies are non-normative, so I think that we don't need to
worry about it in the Node Requirements document, though I am by no means an MLD expert.


Appendix B of the MLDv2 draft is simply a list of changes
from MLDv1. While this list is non-normative, I think the
changes are in the body... so my question is whether there's
really a bit more to MLDv2 than source-specific multicast.
But lets stop the discussion between MLD non-experts and
ask someone who knows: Brian, are there other improvements
in MLDv2 than source-specific multicast? If yes, are those
significant enough that nodes should migrate to MLDv2
even if they don't use the source-specific multicast feature?

So, I think I answered this already in an earlier message, but let me clarify something here. The IP stack layer really does not know what kinds of multicast applications are going to be run on it. A user could arbitrarily install an SSM application just as easily as an ASM application, if the stack supports multicast at all.

With that in mind, if a stack developer knows for sure that his stack
couldn't be used for a certain flavor of multicast application (SSM
or ASM), then he can use the current guidelines in the node reqs to
determine whether MLDv1 or MLDv2 should be supported.

In all other cases, I would prefer MLDv2 for flexibility reasons.

So, if I back off my answer in the last post a little, perhaps the
wording for any-source should be something like "SHOULD support MLDv2
but MAY support MLDv1".

Brian


-------------------------------------------------------------------- IETF IPv6 working group mailing list [EMAIL PROTECTED] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------

Reply via email to