Roger Moser wrote:
v=spf1 sc=mfrom,pra a mx
sc=mfrom exists:%{l}._ses.example.com -all
That would be a new kind of positional modifier, not the
variant documented in SPF protocol-03. But all existing
implementations ignore unknown modifiers. Your proposal
could result in SPF1 forced on sc=pra, which is not much
better than forcing PRA on all existing v=spf1 policies.
It's an idea for v=spf2, but it doesn't work with v=spf1.
For v=spf1 you'd need an "opt-in" modifier like pra=yes.
With my "properties" resp. "predicates" list idea you
could also define p=pra as in say p=helo,pra,hardpass
I still think that we might need more than one possible
"property" resp. "predicate" for v=spf1 in the future.
Bye, Frank