OK . . . lots of input
From: David Frascone (davefrascone.com)
Date: 24 Oct 2001 19:55:47 -0000
First off, for symantic representation of the ABNF, I think we just need to
include the AVPs that are REQUIRED for an individual command code.  I'm 
invisioning something like:

<command  blah>
<RequiredAvps>
List of avps.
</RequiredAvps>
</command>

If you can come up with a better way so that the avps can be validated by
the DTD, that would be nice.

Here at the bake off, I've had some input from the guys here, and they would
like to see multiple ABNF representations in the dictionary.  Let me explain:

In the case where a Mobile=Ip agent is denied access, there would be many
fewer required avps than on a successful attempt.  (The ABNF in the draft
has to be changed too).  But, we would still like to be able to enforce the
mandatory AVPs at the protocol layer with the dictionary.  (Protocol as 
opposed to application).

So, I proposed having named lists of the ABNF required AVPs, with a particular
name being the default (the one used at the protocol layer).  That way, an
application could read in the other lists, and do more application layer
checking.


<command  blah>
<RequiredAvps name="default">
ResultCode
</RequiredAvps>
<RequiredAvps name="Mobile-Ip Success">
ResultCode
Keys
etc.
</RequiredAvps>
</command>


What do you think?

Results generated by Tiger Technologies using MHonArc.