[ppml] CIDR support for whois.arin.net / merging whois with rr data?

Asking for CIDR is good - I tried doing it once before even when it was
"old" ARIN whois server, obviously it has not been implimented...

Regarding copy whois into RR - ARIN will probably not do it since RR
is running on different type of software (at least from what I know - 
somebody from ARIN can clarify), but possibly they could add an option
to sub-lookup in RR database together with ARIN's regular whois. My 
understanding however is that in US RRs are not quite as centric to RIR
as it is with RIPE/APNIC so data can well be in some other RR and ARIN 
does not want to provide such data with regular whois if it can not
trust it - again somebody from ARIN would have to clarify.

On Thu, 14 Sep 2006, Jeroen Massar wrote:

> Hi,
>
> I am not sure if this is the correct place to ask this question, but it seems 
> to be the most appropriate place to ask this. I could not find a policy or 
> any guidelines for this kind of request, if there is any please point me to 
> it and otherwise it might be good to document it somewhere.
>
> Currently when one queries whois.arin.net for an address with a prefixlength 
> it will return that it doesn't support CIDR:
>
> 8<----------------------------------------
> $ whois -h whois.arin.net 2620::/48
>
> CIDR queries are not accepted
>
> No match found for 2620::/48.
> ---------------------------------------->8
> It clearly detects I am asking for a CIDR block.
>
> Simply stripping the /48 helps in this case:
> 8<----------------------------------------
> $ whois -h whois.arin.net 2620::
>
> OrgName:    U.S. Securities & Exchange Commission
> OrgID:      USEC
> ...
> ---------------------------------------->8
>
> The client can of course do this automatically for you, but that is not very 
> useful, especially with IPv4 in mind where one can ask for 192.0.2.0/26 while 
> the that is actually inside a /25, you will get back that there are 2 
> possibilities. For tool writers this is not an optimal solution either, also 
> one will have to do a second query, copying in the name of the network as a 
> query. With IPv6 it will be even more fun, /32 is doable, /48 is also doable, 
> but can you strip it to a /33, /42 etc?
> Next to that CIDR was introduced in 1993 (before I even had internet at home 
> ;) which is already 13 years ago, thus it is definitely way over time that 
> the ARIN whois starts supporting this.
>
>
> Therefor, I would like to request if the whois query match can be made on the 
> prefix including CIDR as that saves a lot of head troubles and is generally 
> more convenient.
>
>
> On a similar subject, there is a rr.arin.net which provides routing 
> information. In RIPE/APNIC/AFRINIC/LACNIC land we are very used to simply 
> querying whois.<registry> and getting back the information that is stored in 
> rr.arin.net. The problem with rr.arin.net is though that it seems that not 
> too many organisations are using it. Next to that, if one queries the RR for 
> a block for which the inetnum/inet6num is known, but the route object isn't 
> one gets back a 0::/0 answer, while it could easily return the information in 
> whois.arin.net.
>
> Thus, to possibly solve both problems in one go, could the ARIN staff 
> populate rr.arin.net with the information that is currently contained in
> whois.arin.net as a first step, and as a second step move whois.arin.net to 
> whois-old.arin.net and propagate rr.arin.net as whois.arin.net ?
>
>
> This will have one drawback, but from certain perspectives not a big one: the 
> format of RPSL is quite different from the current format of the 
> whois.arin.net output. Then again, the other RIR's are using RPSL format 
> already and thus tools are already built to support both formats and should 
> not have a problem with RPSL when that would appear as a response from their 
> queries.
>
> Greets,
> Jeroen