> On Apr 12, 2018, at 8:27 AM, Ken O'Driscoll via mailop <email@example.com>
> Hi all,
> While the "v=DKIM1" is RECOMMENDED as opposed to REQUIRED I have always
> included it in the DNS record and this appears to be the norm.
> However, I have recently been dealing with a provider doesn't include it in
> either their shared public key or when providing the generated public key
> to the client (for whitelabeling).
> Personally I don't like this practice because I believe that the "v=DKIM1"
> has become so ubiquitous that at least someone has coded a validator that
> treats it as required.
> Interested in what others think of this.
I believe the history behind it not being REQUIRED is so that during transition
you could publish a single record that was both a valid DKIM key and a
valid DomainKeys key.
If I were writing a validator I wouldn't require it. If I were publishing a
key record I would, even though leaving it out is entirely within spec and
even though the only semantic value it really adds is "this is not a
If nothing else, it makes it clearer what someone's intent was when
that RHS is in a TXT record at example.com or _spf.example.com or
_dkim.example.com or _dmarc.example.com ... If you're providing it
to a customer for them to put into their DNS that's probably relevant.
mailop mailing list