>>>>> On Tue, 11 Jun 2002 16:38:07 -0700,
>>>>> Alain Durand <[EMAIL PROTECTED]> said:
> I would agree there. If an application stores the IP address of its peer
> for future use,
> on a multi-interface node, if link-local addresses are used and the
> application
(in theory, the "multi-interface node" should be "multi-link node".
but anyway)
> did not kept track of the incoming interface, there would be some issues
> when
> it would like to reconnect to its peer.
That's true, but IMO this is a bit different issue from address
selection, particularly the issue of source address selection. Even
if an application avoids using link-local addresses as source address,
it cannot assume the source address of an incoming packet is always a
non-link-local address. And, if such an application needs to keep the
peer's address of an incoming packet, it will definitely need to keep
the corresponding scope zone. This is irrelevant to how the
application chooses the source address for outgoing packets.
I agree that scoped addresses introduce additional costs to
application in general, though.
JINMEI, Tatuya
Communication Platform Lab.
Corporate R&D Center, Toshiba Corp.
[EMAIL PROTECTED]
--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page: http://playground.sun.com/ipng
FTP archive: ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------
