Draft: draft-ietf-seamoby-card-protocol-07
Reviewer: Joel Halpern
Date: 7/1/2004

This draft is on the right track but has open issues, described in the review.

It should not be hard to fix the problems, but it appears to me that the 
current sub-option structure ought to be a show stopper.   Sorry.


Significant concern:
The specification indicates that the TLV encoding for the sub-options uses 
a length whose units depend upon the type of the sub-option.  This means 
that we can never (okay, maybe "only with great care, good luck, and 
difficulty") introduce new optional sub options, since if they arrive at an 
un-upgraded system the system will not be able to determine the length, and 
so will not be able to parse the packet.
I presume this was done because some options will not fit in 255 bytes.
Will they all fit in 1020 bytes?  If so, all sub-option lengths could be in 
multiples of 4 (sub-options are required to be multiples of 32 bits by 
other wording in the spec.)
Sub-options could be required to be a multiple of 8 bytes, and lengths in 
units of 8 bytes.
Alternatively, a bit or two could be taken out of the sub-option type to 
explicitly indicate the resolution of the sub-option length.
Or sub-option lengths could be 12 bits, with sub-option types only 4 bits 
(or 11 and 5).
I am sure that there are other alternatives.

Section 5.1.4 defines the Capability AVP.  The diagram explicitly shows a 1 
octet AVP length, and one Reserved octet.  The text indicates that the AVP 
Length is two octets.  Which is it?