[ietf-dkim] Re: Re: NEW ISSUE: Simplify SSP decision tree

[Frank Ellermann]

Michael Thomas wrote:

> FWIW, I don't think that t=testing is at all helpful either.

+1  It also muddies the water for the Authentication-Results,
if they map e.g. "strict deny" to "hardfail" and "all process"
to "softfail" there's nothing left for any "strict testing".

> It's that sort of subjective state that we should both learn
> from SPF and avoid.

Yes, IMO the SOFTFAIL "testing" wasn't a good idea, one of the
many SPF battles I lost.  Some folks really like it, a FAIL is
a difficult decision for the known reasons wrt SPF or PRA.

A "strict deny" SSP is also a difficult decision.  But this
"testing" loohole doesn't help receivers, a sender policy (or
a published signing practice) must make sense for *receivers*.

 Frank