Re: [pcp] WGLC: draft-ietf-pcp-base-12.txt

[Alain Durand <[e-mail]>]

On May 31, 2011, at 2:40 PM, Dan Wing wrote:
> 
> 
>> - UNAUTH_THIRD_PARTY_INTERNAL_ADDRESS could be removed, and replaced
>> with NOT_AUTHORIZED (more on that inline).
>> 
>> - There is no discussion about levels of conformance.  There's some
>> discussion about particular options being optional (e.g. THIRD_PARTY
>> not supported in certain deployment scenarios), but no overall
>> statement of what features must be supported by all conformant
>> implementations.  e.g. If I had a small embedded server that supported
>> MAP opcodes but not PEER, would it be conformant?
> 
> Nobody has initiated discussion or provided a straw-man on conformance
> levels.  What do we want?  Myself, if there is any chance of a wireless
> devices (which is approximately 100% chance), PEER needs to be supported
> because it allows the wireless device to optimize its NAT keepalive
> traffic.  But, to me, MAP doesn't need to necessarily be supported.

I think we should stay away from conformance levels. This gets very complex, very quickly,
and does not help interoperability.
The argument that MAP or PEER are only useful in a wireline or wireless scenario seams a bit weak to me.
We are now seeing fixed wireless environment... Which profile should they have?

I believe that unless there is a very strong reason against against, we should keep MAP & PEER functionality.

  - Alain.