[ietf-dkim] New Issue: ssp-04 Domain Existence Requirement

[Frank Ellermann]

Douglas Otis wrote:
 
> NNTP might use a TLD of .invalid to ensure an address is not
> mistaken for a valid email-address.

It doesn't.  Individual NNTP servers might allow this, as far
as they check From: email-addresses at all.  Likewise individual
receivers behind a news2mail gateway might allow this, as far
as they check From: email-addresses at all, not limited to ADSP.

TLD ".invalid" in conjunction with the text "valid mail domain"
is clear.  If not the poor reader needs medical help, not more
text in the draft.

> the suggestion to use wildcard domains to publish ADSP TXT 
> records necessitates exclusion of domains without valid host
> name syntax.

AFAIK it doesn't.  A, AAAA, MX, or TXT records are not limited
to valid host name syntax.  That the draft only discusses the
case of MX wildcards might be an issue.  But I think readers
should be able to figure out that it's also applicable to AAAA
and A wildcards.

The statement "multiple wildcard records produce an undefined
ADSP result" is odd.  Defining a proper "subtype" to find the
wanted ADSP needle in a TXT RR haystack is no rocket science.
We can get an IAB review if necessary.

> These exploits can be prevented by insuring valid host name
> syntax.

What has host name syntax as required for 2821bis and 2822upd
to do with wildcards, another DNS oddity I'm not aware of ?
Anything that is no host name can't be a "valid mail domain".

 Frank