I hate the new mail system

Pete Turnbull pete at dunnington.plus.com
Fri Mar 10 14:42:28 CST 2017

On 09/03/2017 09:50, Christian Corti via cctalk wrote:
> On Thu, 9 Mar 2017, Tor Arntsen wrote:
>> I did an strace and I can confirm that the Linux 'whois' client that I
>> used from those various sites sends '-T dn' (or actually -T dn,ace)

> I did a little research on that:
> The '-T' option is passed to the whois server, it's not a client option.

I don't think Tor has suggested it's a client option, and I certainly 
did not.  It's simply part of the string that one particular client 
knows to send to DENIC, and which at least one other client can be 
forced (manually) to send.

> Intelligent or modern clients know what options to pass to the
> appropriate server, in this case '-T dn' to the DENIC whois server.

Well, for one value of modern :-)  The RIPE client which DENIC 
themselves recommend people to use, does not do that.  It has to be 
cajoled into sending the "-T dn" along with the query string.

> This option is completely legal and was introduced at DENIC in an
> attempt to better protect the domain holder's privacy (you know,
> different country, different rules). This was many years ago, but it's
> still there.

Legal in what sense?  It's not part of any RFC, and appears to be unique 
to DENIC.  I regularly use dozens of whois servers and not one other 
honours, let alone requires, that option. Also not that RFC3912 gives a 
single protocol example which does not require any such option, and 
refers to RFC954 for normative references - ie, they way it should be 
done.  No -T there either.

> RFC 3912 doesn't specify what output the whois server is supposed to
> send. Everybody "assumes" that it should be the complete domain
> information, but that's simply not the case.

That's certainly true.  Very few whois servers return all the 
information; many won't disclose phone numbers, for example.  In fact I 
was surprised how much text /is/ returned by DENIC.

> Imposing this assumption is what Mouse does, and that is wrong.

I disagree; Mouse (and I) were simply using standard common whois 
clients and getting either nothing or an error in return to queries.

Pete Turnbull

More information about the cctalk mailing list