[manet] DYMO Address Prefix length: discrepancy between DYMO and PacketBB drafts

DYMO in all drafts states that if the prefix field is missing the prefix 
length is assumed to be 0.
PacketBB on the other hand states that the prefix length, if absent, is 
considered to be equal to the address length. Certainly, this makes more 
sense and saves implementations from having to handle this case differently.

My guess is that DYMO has adopted its default from AODV where 31 is the 
maximum for this field since it has only 5 bits. IPv4's address length of 32 
cannot be represented. Although DYMO in draft 03 already reserved 6-7 bits 
for this field(depends if you read page 7 or 12).

In my opinion DYMO and PacketBB should have the same prefix length semantics.
However, if that is not desired, I'd suggest to state this fact in upcoming 
DYMO drafts to avoid confusion.

Relevant sections:
------------------------------------
draft-ietf-manet-dymo-04.txt
Page 10:

RM should contain a Prefix for each address that is not a host
         address.  If a prefix is not included in conjunction with an
         address, it is assumed zero (host address only).  For more
         information on advertising a Prefix see Section 4.7.

------------------------------------
draft-ietf-manet-packetbb-00.txt
Page 13:

An address SHALL NOT appear more than once in the same message
      with the same prefix length (an address without a PREFIX-LENGTH
      TLV is considered to have a prefix length equal to the address
      length);

------------------------------------

Thanks,

Uli

_______________________________________________
manet mailing list
manet@ietf.org
https://www1.ietf.org/mailman/listinfo/manet