On Thu, 22 Dec 2005, Arvel Hathcock wrote:
That's fine, and as we know about 1,000,000 domains
and 10 independent interoperable implementations are
good enough for "experimental".
Well, I don't have anywhere near that many. And I'm not pretending to know
what constitutes "wide deployment" or "how many users is enough", etc. This
isn't for me to decide anyway. All I want to do is counter the claims
people make that there are a couple dozen test machines or < 100 domains
using DKIM. I'm a better "marketer" than that <g>. I now have a list as a
result of my 200 customer sampling last night that is 300+ domains.
It is good to have real numbers. As I said it is very difficult to find
number based on looking at dns (which is bad as I think it would have
been better if _dkim prefix was used instead of reusing _domainkey - it
is really not that difficult to setup CNAME if one wants to share). The
only large email-use domain using DKIM I could actually see was cisco.
I do have a preference to seeing the number of separates sites (i.e.
installations as opposed to #of domains), you did mention that just
two sites accounted for 35 domains before...
Also based on what I could see I think it may well be that there is only
one or two commercial products that released dkim in their production
products - you can easily see it based on who talks about interoperability
testing on development list (and I don't think sendmail released their
version to production; neither did cisco beyond their own use, which
leaves only ....)
In any case the amount of use for DKIM (and for that matter even DK) is
still extremely small compared to real-word installation of SMTP and this
all should be enough that group work on standards without preconditions.
--
William Leibzon
Elan Networks
william(_at_)elan(_dot_)net
_______________________________________________
ietf-dkim mailing list
http://dkim.org