Re: [dnsext] WGLC ENDS0-bis

[Edward Lewis <[e-mail]>]

At 15:19 +0000 5/12/11, Ray Bellis wrote:

>I believe the intent of 6.8 is to reflect 5625, where I put in text along
>the lines of "if you're trying to be transparent, make sure you don't strip
>the EDNS0 options".

I think that is a fine principle and well stated - what the section 
is trying to impart.  But I'm talking more basically about a lack of 
a definition of middle box in this document.  What I'm hinting at is 
including a reference where the term is already defined or a 
paragraph defining the term.

>In both cases I wanted to cite the hop-by-hop principle but couldn't find
>relevant chapter and verse to back it up.

More text is needed then.  Given some time I'll try to come up with something.

In 
https://www.ietf.org/ibin/c5i?mid=6&rid=49&gid=0&k1=933&k2=56283&tid=1305232265 
I wrote this "the fact that the DNS is not a client-server  protocol, 
as it is usually treated in text, but a client-cache-server 
 protocol".  Beginning with that, there is a difference between 
hop-by-hop and end-to-end in most DNS messages supporting a 
transaction.  The data model, backed up by DNSSEC, is pretty much 
built to be end-to-end.  EDNS0 on the other hand is hop-by-hop, 
taking parameters governing the message exchange from one entity to 
the next, not along the entire end-to-end path that the data travels.

...That's a first scratch of what is needed.  Admittedly it is not 
fleshed out enough to be clear and come to the point.  But I'm just 
squeezed for time now.

-- 
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
Edward Lewis             
NeuStar                    You can leave a voice message at [phone]

Now, don't say I'm always complaining.
Wait, that's a complaint, isn't it?