In your previous mail you wrote:

   On Sat, 2003-02-22 at 14:54, Bill Manning wrote:
   > regarding the use of mapped addresses:
   > 
   > draft-cmetz-v6ops-v4mapped-api-harmful-00.txt
   > 
   > might be a useful ID to review before committing this draft to the
   > stds process.
   
   These issues are completely unrelated. The API issues are real but they
   are not something that can or should be considered in a protocol
   specification.
   
=> note there is no consensus at all about this ID.

   I'm not too happy about RFC2553 myself in this respect, and I strongly
   support the "Alternative solution" (fully specify IPv4-mapped behaviour)
   in the above mentioned draft.
   
=> I agree but some of us are relunctant to put the burden of a full
IPv4-mapped behavior on the shoulders of (other) implementors.

   I can attest from implementation experience that it is possible to
   create a hybrid IPv4/IPv6 stack implementation that uses around 80-90%
   shared code between IPv4 and IPv6. All IPv4 addresses are handled in
   IPv4-mapped format internally inside the stack.
   
=> IMHO this is the best thing to do where a dual stack is written from
scratch.

   Our sockets API is (mostly) version agnostic.

=> so you should adopt our local credo:
"IPv6 is not a new protocol, it is a new version of the IP protocol."

   Most applications are not
   even aware which IP version they are using. The OS is not a unix
   derivative and did not have the legacy baggage of the BSD style sockets
   API. The API that was already defined for IPv4 yielded very easily to
   support IPv6.
   
Regards

[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]
--------------------------------------------------------------------

Reply via email to