On Sat, Jun 19, 2004 at 02:47:02PM -0500, [EMAIL PROTECTED] wrote:
> Thanks Derek.  
> 
> Good that we decided to keep that paragraph in the draft.

Well from a practical point of view it wouldn't matter if it
was removed or not.  I just thought I'd point out that there
is (at least) one implementation which does this.

The issue being that a node can not in general control which
particular router would generate such an ICMP,  or what route
the packet would follow,  so it can't reasonably rely upon
any specific source address being picked.

  [ Ignoring source routing options,  and prior knowledge ]

So frankly,  in the absense of configuration overriding any
adress selection,  the router can pick any address it owns
as the source of the ICMP packet.  The paragraph in question
was really a hint at a quality of implementation issue.

Now if the router does pick the source address so as to be "most
useful",  it's only really useful in connection with people manually
debugging specific packet trace issues.

DF

--------------------------------------------------------------------
IETF IPv6 working group mailing list
[EMAIL PROTECTED]
Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------

Reply via email to