Mark Lentczner wrote:
On Sep 3, 2004, at 2:33 PM, Andrew Newton wrote:
Despite lack of agreement on the outcome, there is agreement on the
stated positions. In the words of Olafur:
1) Specify everyone to be compliant at all times.
2) Specify clearly full complaint state and tolerate non-compliant
state during phase-in.
My preference is for position 2.
I am also strongly against other options with explicit phase-in periods
and sunset-dates. As has been eloquently argued by Doug Crocker, such
language in RFCs has a bad track record of working out, and so is best
dispensed with.
Options which drop the new SPF2 RR type altogether are not likely to sit
well with the rest of the IETF community, specifically the DNS-EXT
folks. The will reasonably argue that DNS has a well designed extension
mechanism in the allocation of RR types, and it should be used, rather
than overloading the TXT type.
Options which drop the TXT RR type are not practical as there are some
significant barriers to using a new RR type in some software environments.
Even within option two, we can specify that the RR type be looked at
first, and the TXT second, if the RR is not found. This would clearly
show that the RR type is prefered over TXT, but TXT is tolerated.
Yakov