Re: [dnsext] draft-mohan-dns-query-xml-00.txt

[Paul Vixie <[e-mail]>]

On Monday, October 03, 2011 18:31:52 Alex Bligh wrote:
> I presume we could base64 encode POST data if we couldn't send it
> binary, whereas IIRC we couldn't do much more than hex encode GET URLs.

the ability to base64-encode POST data is part of the HTTP spec, so yes.  the 
response could also come back encoded that way.  this is supposed to be "just 
web traffic."

> So +1 for POST here, with body content as close to UDP message format
> as possible, save that we should surely be able to junk a pile of UDP
> restrictions (e.g. we can surely mandate EDNS, assume 'packet' size
> can 2^32 bytes, etc.). I presume we also don't need UDP headers etc.

i don't think we should overly constrain the message (like mandating EDNS) 
since this adds test cases to compliance suites and precludes future DNS 
protocol extensions that could be better than EDNS.

we would only include the DNS message, which UDP thinks of as a "payload";
not the transport or network headers.  also note, this isn't necessarily a 
tunneling protocol -- if it's used by a stub then there is no underlying UDP 
packet to pull anything from.  and where it is used as a tunneling protocol, 
the underlying DNS message could have arrived via TCP/53 or even HTTP.

paul
_______________________________________________
dnsext mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/dnsext