Re: static versus dynamic nature of DNS: rate limiting

On Feb 22, 2004, at 10:49 PM, Patrik Fältström wrote:

> On 2004-02-22, at 22.28, Ian Peter wrote:
>
>> I wonder if its possible before the IETF meeting to have at least a
>> brief exchange as regards SPF. SPF has gained quite a degree of
>> momentum, and in my mind at least it's best to back winners in
>> situations like this and put some of the other issues aside.
>
> Problems with SPF (as I see it):
>
> - It uses TXT records and not a new RR (it should use a new RR type)

This seems to me to simply be a crutch to allow immediate
implementation (which accounts for it's momentum, IMHO).  I don't think
anyone in the SPF community would be adverse to having a new record
type.


> - It has a (too complicated) macro language which I am nervous can
> create a target for an attack. Allowing "everything which is possible
> in perl" is not a generically good idea. I definitely want to see a
> security analysis of the language and security threats (mis
> configuration, ability to write macros which doesn't converge in
> memory and/or time) etc.

It certainly doesn't allow 'everything which is possible in perl',
though I agree it's rather complex.  Much of this is rooted in getting
around the length limit for TXT records (and the UDP packet size
limit.)

George