Re: 6MAN WG Last Call: draft-ietf-6man-predictable-fragment-id-01

[Brian Haberman <[e-mail]>]

On 12/9/14 12:55 PM, Fernando Gont wrote:
> On 12/09/2014 11:58 AM, Brian Haberman wrote:
>> Fernando,
>>
>> On 12/9/14 3:29 AM, Fernando Gont wrote:
>>>
>>> * Requiring that the IP ID be unpredictable
>>
>> Let's suppose that the WG decides to publish this document and
>> makes the above a MUST.
>>
>> When the WG decides to advance this PS RFC to IS, how do we
>> determine if this critical component is implemented and
>> interoperable?
> 
> (Me thinking out loud):
> 
> 1) In the same way we do for RFC6528 or RFC6056
> 
> 2) With something similar to
> (<http://www.si6networks.com/tools/ipv6toolkit>):
> 
> # frag6 -d TARGET --frag-id-poliy

That didn't answer the question.  In order to advance something on the
standards track, there needs to be at least two interoperable
implementations.  How an implementation initializes a field like the
fragment ID, TCP sequence number, or ephemeral port number has no affect
on interoperability.

What you are describing in those documents is implementation guidance,
not a protocol specification.  Given that, I would suggest that this
document strike any normative text, drop the Updates tag, and move to
Informational.

Regards,
Brian