[ietf-dkim] ISSUE: Revise wildcard discussion

[Frank Ellermann]

Eliot Lear wrote:

>> The various ADSP drafts fail to provide the essential version tag
>> for their identification, this is no general TXT wildcard problem.

>> The general TXT problem is the *size* of all RXT records combined,
>> and the various ADSP drafts refuse to share SPF records for their
>> limited size needs.
 
> While I myself am leaning away from TXT records for other reasons,
> this is not one of them.  ADSP is a bit more safe because like DKIM
> it uses a _adsp label.  If you look up a TXT record with that label
> it makes less sense to expect anything other than ADSP.

As explained in Dave's message:  The "label" approach does not work 
with *wildcards*, for that you'd need a separate record type and/or
a "tag" for the identification in the case of a shared record type
(TXT, SPF, whatever).  But his reasoning was backwards, with a "tag"
for the identification of a desired record in a RR set it would also
work for wildcards until the size issue for shared RR types kills it.

See <http://tools.ietf.org/html/draft-iab-dns-choices#section-3.1>
for an "authoritative" (from my POV) statement about these issues.

 Frank