Re: new version of ESP ID

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

>I've always assumed that a single controller or coordinated set of 
>controllers assigned SPIs for a given multicast address anyway.

SSM changes the semantics of the multicast address space; it is now
per-sender (i.e. source=S1 dest=G and source=S2 dest=G are two different
"channels") in the SSM range.  Thus, there can be no single coordinator
for G and SPI conflicts can occur if S is not used for demultiplexing.

  Bill