Date:        Mon, 19 Aug 2002 19:30:21 +0900
    From:        [EMAIL PROTECTED]
    Message-ID:  <[EMAIL PROTECTED]>

  |     i have never said we should terminate existing connections.

No, I know you didn't.

  |     i suggested we should refuse new incoming connections (TCP SYN).

Yes, that is what I am arguing against.

The line of reasoning seems to be

        if we accept a connection to a deprecated address, we may
        have to open a connection from a deprecated address later.

        we're not allowed to open a connection from a deprecated address

        hence we should not accept connections to deprecated addresses

That's wrong because the middle point is wrong, you are allowed to
open a connection from a deprecated address.   You're just not supposed
to do that unless there is a reason (existing communications).

Note "communications" != "connection".

But even if it weren't for that point, you'd just be shifting the
finish line in a race condition.

That is,
        if we accept a connection to a non-deprecated address, and that
        address is later deprecated (which any might be) we may have to
        open a connection from that address even later

        we're not allowed to open a connection from a deprecated address

        hence we should not accept connections to non-deprecated addresses

and that's absurd.    That is, even if we weren't allowed to open a new
connection from a deprecated address (which we are, but ignoring that),
that's no reason not to accept incoming connections to deprecated addresses,
or to addresses that might become deprecated before the new connection has
finished.

Just treat a deprecated address identically to any other address, except
don't choose it as the source address if there's some reasonable alternative.

Or, if you like, just follow the rules in the address selection draft,
and aside from that, ignore "deprecated" as it applies to addresses.

kre

ps: I seem to recall that all this was discussed back when deprecated
addresses were invented.

pps: that we're having this discussion does suggest that you're correct
and the docs need clarification.

--------------------------------------------------------------------
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]
--------------------------------------------------------------------

Reply via email to