[Standards-JIG] the word "multicast" is not appropriate

Carlo v. Loesch CvL at mail.symlynX.com
Thu Jun 1 11:27:26 UTC 2006


your definition of multicast wouldn't make it through the ietf
even if the ietf itself in its early days published a very lax
definition. the latest definition according to the ietf reads
like this:

multicast (http://www.ietf.org/rfc/rfc1983.txt)
      A packet with a special destination address which multiple nodes
      on the network may be willing to receive.  See also: broadcast,
      unicast.

i miss the efficiency aspect which was the original reason why
somebody would implement multicast in the first place - sending
something to multiple nodes doesn't need a word, but being able
to do it in a topologically efficient way so far has only deserved
the word 'multicast.' but that's history, time goes fast and vocabulary
gets banalized.

now if we look at JEP-0033, like smtp it provides for multiple addresses
in a packet, not for a "special destination address" .. so it is just
as multicast as any email. if you want to define multicast that way,
it's like calling yellow golden. you're in a free world, i can't stop
you from doing that.

in the case of our smart distribution proposals, we do indeed specify
a "special destination address" and we do indeed distribute to
"multiple nodes on the network" using our context lists. so you can
argue that the proto-jep already provides for multicasting, no matter
if we use your <iq/> or our <presence/> syntax populate the lists.

i'd say it's still a poor man's multicast if no elaborate distribution
is happening, but okay - you want a poor man's multicast? you can have
a poor man's multicast.

Richard Dobson typeth:
| continuing to try and make out that this is the case is just going to 
| make you look like a fool, please let this be the end of this thread for 

keep your argumentation style below your own belt and stay away from mine.




More information about the Standards mailing list