On 9/11/2004 1:59 PM, Andrew Newton wrote:
2) On the issue of compliance with the use of the TXT record, the
working group has at least rough consensus that TXT usage is acceptable
for compliance and should not be specified as a configuration that will
be non-compliant. However, there is at least rough consensus that the
use of the SPF-specific record type is more desirable than the use of a
TXT record type. It is the opinion of the co-chairs that the -protocol
document clearly state that the usage of TXT records will most likely
be deprecated by future protocol definition.
I don't think that's good enough and would offer an alternative:
* specs sent to last call MUST include a new RRType code
* an addendum or a sibling spec MAY describe coexistence mechs for
a TXT RR for backwards compatibilty
This model ensures that future implementations cannot claim "compliance"
until they fix whatever bugs are keeping them from using the new RRType,
while also providing a support channel for legacy implementations.
--
Eric A. Hall http://www.ehsco.com/
Internet Core Protocols http://www.oreilly.com/catalog/coreprot/