[RPRWG] .17 b multicast learning simplification
RPRWG,
We haven't yet reached consensus over the variety of methods
proposed for controlling whether or when multicast frames
should be learnable.
It seems to me the whole thing has become a lot more complex
than it needs to be for standardization.
I also noticed that the poll shows that for Q4 (whether MC
frames should be learnable) only two thought both a) and b)
were unacceptable. That could mean that another choice,
"a) or b)," meaning the SAS transmit could be set to either
always enable learning, or never, might be acceptable.
This simplification makes sense to me. The other options,
controlling frame by frame either by client control or a bit
in the MC address table, or both, are enhancements that
vendors can implement for added-value differentiation. They
are good candidates for that because they don't require
special action from other nodes to interwork.
So I would like to propose Q4 choice d) = "a) or b)" as a
compromise that will help get .17b done and maybe help
alleviate the perception of excess complexity in the standard.
Comments?
Gary