Re: [dnsext] Draft of an RRTYPE extension language

[George Michaelson <[e-mail]>]

On 19/08/2011, at 3:26 PM, John R. Levine wrote:

>> I wonder if its going to be sufficiently extensible for forseeable requirements. SRV for instance encodes structures like regex() which is 'powerful' (ie complicated)
> 
> SRV is three integers and a domain name.  Do you mean NAPTR?  It includes a regex, but the regex isn't interpreted within the DNS, only by clients, so as far as the DNS is concerned, it's just a string.
> 
> NAPTR:35 I2 I2 S S S N[A]
> 

Yes I meant NAPTR.

And, as you say...

> I should probably emphasize that this isn't supposed to be the FUSRP*. Some RRs require that the DNS server do something new, so if for example we added a BNAME record, it'd be easy enough to describe the syntax, but that wouldn't tell the DNS server about all the new synthesis rules.

its descriptive only, in a structural Octet-level-encoding sense. 

> 
> Regards,
> John Levine, [e-mail], Primary Perpetrator of "The Internet for Dummies",
> Please consider the environment before reading this e-mail. http://jl.ly
> 
> * - Final Ultimate Solution to the RRTYPE Problem

I really hate use of the words Final and Solution in one sentence. we've both just aquired a huge amount of Baysian for Godwin filters.

-G