At 14:24 -0700 6/11/04, Michel Py wrote:
I know of other experimental uses of the TXT RR, for one,
opportunistic encryption. There is an RFC in preparation to
document that. So, MARID can't claim the TXT for it's own
- others already use it.
If they are using a well-known set of attributes, I don't see why. Part
of the spec could be acknowledging the other attributes already in use.
The issue is that, as ISC does, anyone at anytime can take up using
the TXT record for any purpose. My point, which I've done a poor job
of stating, is that if documents are relied upon for specification
enforcement is also restricted to document space. The IETF does not
have a specification enforcement capability, adherence is voluntary.
Through specification you can't guarantee no one else will squat on
the TXT records.
Once again making a case for $MARID RR, if that is defined, then the
protocol will restrict the use of the RR to the rules set forth for
it. The enforcement is then in the "on the wire protocol". The
enforcement is limited to the syntax and handling of the record in
DNS, not the use, that is true - a non MARID app can use it, but the
implementations would bound its use to one that is compatible with
the definition of the $MARID RR.
This is too strong; what I meant is that we should have a list of known
attributes for the TXT record. Anyone wanting to use an attribute should
register it. Would this make the TXT record palatable for PS if
attributes were reserved for SPFID?
To do this, there's a need to establish a registry of TXT contents,
most likely an IANA registry. This would be a change one of the
original (RFC 1035) types, so I'm sure it's undergo a thorough
review. I'm doubtful this would be doable, honestly, but one can
always try.
--
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis +1-703-227-9854
ARIN Research Engineer
"I can't go to Miami. I'm expecting calls from telemarketers." -
Grandpa Simpson.