Re: RFC 2460 issue

[Markku Savela <msa@burp.tkv.asdf.org>]

> Off the top of my head I know that RFC3493 needs to be updated since
> the IPV6_UNICAST_HOPS socket option now accepts 0 as a valid hop
> count. I really do not understand what a hop count of 0 implies and
> why we should bother updating the RFCs.

Heh, yes. I too wondered about what I should do if application sets
TTL = 0. There are two choices

 a) Packets go to /dev/null (perhaps some obscure testing feature?)

 b) Just let packets with TTL=0 go out.

I chose (b), because

 - TTL is naturally checked only on fowarding, not when sending
   own packets out. Thus, any TTL just gets accepted and sent.

If packet with TTL=0 is for this node, it is accepted (again, because
TTL test is only for forwarding).

Forwarding decrements TTL and if result is 0 or < 0, packet dropped
(with appropriate ICMP if needed).

I'm happy with above semantics. I don't see any need to worry about it
too much.


--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------