Sanjaya wrote:
> Just a thought, maybe we could redefine the RIR extended delegated
> file format to incorporate both cc values of inet(6)num|autnum and
> organisation Whois objects like this:
>
> registry|xx|type|start|value|date|status|yy-custodianid
> Where:
> xx = country code of inet(6)num|autnum
> yy = country code of organisation
>
> This will require agreement among the RIRs and some adjustments
> by the relying parties, but provides a clearer mapping between
> Whois objects and the stats file records. My 2 cents.
The stats file could have its structure adjusted but remember that "country:" is a multiple attribute, so there could technically be as many countries listed as there are valid ISO 3166 codes.
Also, aligning the stats file and inet(6)numns doesn't solve the fundamental issue, which is that the "country:" attribute in inet(6)nums has never had a clearly defined meaning. Readers and writers of the data almost certainly mean and understand different things. As there is already a lack of shared understanding and a very large deployed base, which includes lots of legacy software, managing a change that leaves the structure intact while potentially changing how that data should be used sounds dangerous to me. That's the sort of change that leads to subtle, quiet errors that are hard to debug. I think it is safer to break things in a way that is very obvious, so that they get noticed and fixed.
Kind regards,
Leo Vegoda
