Re: [DNSOP] ECS badly formatted ADDRESS field

[Mukund Sivaraman <[e-mail]>]

On Thu, Dec 24, 2015 at 08:01:14AM +0530, Mukund Sivaraman wrote:
> I have a related clarification. What if the ADDRESS field has fewer
> octets than than SOURCE PREFIX-LENGTH indicates? Should REFUSED or
> FORMERR be returned in this case? The draft must clarify this if it's
> requiring REFUSED.

I also suggest describing implementation behavior with FAMILY=0 clearly
in section 6 (option format). This is described all over the draft.

Example, section 7.2.1 (authoritative nameserver) says:

>   The FAMILY, SOURCE PREFIX-LENGTH and ADDRESS in the response MUST
>   match those in the query, unless the query specified only the SOURCE
>   PREFIX-LENGTH for privacy (with FAMILY and ADDRESS set to 0).

1. It is not clear how SOURCE PREFIX-LENGTH makes sense by itself
without knowing what FAMILY it is targetting. How can it be 0 with
FAMILY=0?

2. I assume ADDRESS is missing here, not set to 0? Or are ADDRESS octets
present to match SOURCE PREFIX-LENGTH and they are set to 0 bits?
ADDRESS field should be undefined if FAMILY=0. It is better to clearly
state all this under section 6 (option format).

		Mukund