Salut and Suspend/Resume issues

John Gilmore gnu at toad.com
Thu Feb 21 07:01:21 EST 2008


> Can you explain how is this not mooted by considering mDNS, which XO's
> all run and listen for?

No, I cannot; someone who understands mDNS should describe the
protocol and its implications.  I could go off and read the RFC
eventually (is it spec'd by Informational RFC 4795?  Or does the code
implement some different variation?).  Please, if you know, be
specific in why you think it moots anything.  Most things that listen
for multicast traffic only awaken if they can actually contribute to
the conversation.  (If not, the protocol is broken.)

Hmm, I've skimmed part half of the RFC.  It indeed looks broken.  It's
just like IPv4 ARP: every machine wakes up, looks at every query, and
99% of the time throws them away and goes back to sleep.  Stupid!
Requesters should be hashing the name it's looking up, into the
multicast address, so that servers who can't possibly match the name
don't have to even see the packet, nor awaken.  No wonder it didn't
make it as an IETF draft or proposed standard.

	John




More information about the Devel mailing list