Re: Ripe change to whois

[Niall O'Reilly <]

On 30 Nov 98, at 17:14, Hank Nussbacher quoted someone
who had written:

> >
> >As a matter of fact the referral mechanism was seen as a way for ccTLD 
> >administrators to include information in the Database pointing to their own 
> >authoritative sources of information.

I can confirm that this was indeed the intention when this was first proposed in 
the RIPE (25, I think) DNS-WG meeting.
 
> > * > The RIPE NCC is therefore acting according to users directives that were 
> > * > discussed and agreed on an open way.
> > * 
> 

Yes and no, I have to say.  

At RIPE 31, the functionality proposed certainly appeared to me to be what had 
been sought.  I am no longer certain that this is indeed the case.  In particular,
the following surprises have arisen.

First, the key notion of making the referral function available per TLD at the 
option of the TLD registry concerned seems to have been lost.  This is a 
fundamental element of the original concept of over two years ago.

Second, the "feature" of returning the data relating to the parent domain in the 
case that the specified domain cannot be found seems to me to have nothing 
to do with referral.  The referral concept has to do with re-directing the query to 
an authoritative whois server, not with second-guessing the intent behind the 
query and substituting other data which is available locally.  I suggest that this 
"feature" be declared a bug and removed.

I am sure that these technical problems are based on misunderstandings which 
can easily be resolved.

A more significant surprise is that TLD registries, who have had no specific 
prior notification, now find that data about their domains is being returned in 
response to a class of query for which this was not formerly the case, and that 
they have to deal with the unexpected fallout.


Niall O'Reilly

IE Domain Registry at UCD
Chair, RIPE TLD-WG