Re: [dnsext] Re: I-D ACTION:draft-vandergaast-edns-client-ip-00.txt

[Carlo Contavalli <[e-mail]>]

On Fri, Jan 29, 2010 at 11:32 AM, Stephane Bortzmeyer <[e-mail]> wrote:
> 1) A small an reasonable one, deleting every mention of the
> load-balancing algorithm and just describing an EDNS option for
> carrying the IP address of the desktop client to the authoritative
> name server (which, being authoritative, could do what it wants with
> it). This would suppress one opportunity to troll and would be in line
> with the general IETF rule "protocol, not policy".
That was the intention :-), we'll try to review the document, and
clean up references to any such algorithm if there's still any.

> 2) A more ambitious one (may be too ambitious), to have an EDNS option
> code "Client info", with sub-codes and various fields (and a registry
> at IANA to register these fields) to carry absolutely everything from
> the desktop client to the authoritative name server. IP address
> information would then be just a special case.

Personally, I'd rather have different EDNS options. Eg, we're now
/trying/ to standardize an option to carry the user ip, there's 16 bit
of space to register other options, it's already possible to register
an option to carry my location, or whatever else I want to carry
there?

So we wouldn't have to create a new register, and we wouldn't have to
nest another header that would look very similar to the enclosing
edns0 header, and we wouldn't have to write more code to parse or
handle those options.

But I don't feel strongly, it's just my 2 cents,
Carlo