ietf-clear
[Top] [All Lists]

[clear] Sender's Declaration of Identity

2005-06-10 15:07:53
On 6/9/05 11:21 AM, David MacQuigg sent forth electrons to convey:

Matthew,  I appreciate your response.  I didn't think anyone on this 
list was interested.

At 09:21 AM 6/9/2005 -0700, Matthew Elvey wrote:

The barrier to implementing CSV's CSA is trivial, or at least as 
trivial as implementing this syntax.


Implementation barriers are not what is holding back authentication.  
The problem is lack of a simple standard that will allow all methods 
to be widely deployed.

I would like to see CSV more widely deployed.  Why isn't it happening?

It's new.  There are few working implementations, and little evangelism yet.


Given 3 options:

  1. Do CSV and this neutral syntax.
  2. Do this neutral syntax, but not CSV
  3. Do CSV, but not this neutral syntax.  (If CSV doesn't yield any
     actionable information, proceed with other Identity establishing
     shemes: Check for valid Domain Keys, then useful SPF records, etc.)


Given N methods, there are 2**N options for which methods a sender may 
install.

So?


I see no reason not to go with 3.  (Other than politics - i.e. I see 
no technical reason...)


DNS hunting.  If you don't know what I mean by that term, read the 
introduction to the draft.

I told you, I already read the draft, at least the one that was posted 
at the time.  I found the following quote on your site:
"The large number of authentication records [of CSV] might make this 
method vulnerable to a DoS attack."
This statement is unsubstantiated.  Because far fewer domains need to 
have ANY records in the CSV scheme, this may mean that CSV requires 
fewer DNS records than, say SPF.  Said differently: Ok, with SPF or QR, 
one DNS record can authorize all the servers for an entire domain.  With 
CSV, one DNS record can a server for many domains.  E.g. a webhost that 
hosts 1 million domains, and runs 10 SMTP servers to service those 
million domains.  That's a million or so SPF records, vs. 10 or so CSV 
records.

There is often no 'hunting', because CSV will often provide actionable 
information.  A receiver will therefore often NOT try all possible methods.

I think we just need to agree to disagree.  Your scheme doesn't require 
network traffic, but it does require SMTP server software be upgraded.
Unlike SPF or SID it doesen't have severe negative consequences that I 
can see.  Because your scheme doesn't require ANY network traffic, I can 
see why option 1 above is reasonable.


<line breaks moved>
The proposal is at
http://purl.net/macquigg/email/draft-macquigg-authent-declare
in various formats.  The htm and rtf formats show recent changes 
highlighted in yellow.


This URL still does not work.  Google helped.  A discussion forum 
should be specified in the I-D.  Current draft and discussions are 
where?


This is the first I have heard of any problems with the URL.  Can you 
reach any of the parent domains?  Send me the details off list. 

The error I get is: The requested URL 
/~edatools/home/email/draft-macquigg-authent-declare was not found on 
this server after some redirect.
The problem is on your end (no file type suffix).