[ietf-dkim] FW: An issue with DKIM reporting extensions

[Steve Atkins]

On Oct 13, 2010, at 11:25 AM, Scott Kitterman wrote:

> On Wednesday, October 13, 2010 11:55:25 am Steve Atkins wrote:
>> On Oct 13, 2010, at 8:07 AM, Rolf E. Sonneveld wrote:
>>> or
>>> a special selector (e.g. s=notifications), to identify the different
>>> nature of this mail stream?
>> 
>> No. Never do this.
>> 
>> Selectors are an operational convenience for key rotation and
>> ease of domain delegation. They have no semantics beyond
>> being used to query DNS to find the public key.
> 
> Sure.  That's the right answer from a standard POV, but if I can extract 
> statistically significant information by segregating mail streams by selector, 
> I'll do it anyway, I don't care what the standard says (no, I haven't done 
> this analysis yet, so I don't have an opinion on if one actually could do it 
> or not).


Do whatever you like as a receiver, as long as you do it in private -
I expect there will be some interesting, though probably not terribly
useful correlations there.

But don't encourage senders to break DKIM by suggesting
that there are any semantics to selectors. There aren't, and
trying to add any will break things.

Cheers,
  Steve