[Top] [All Lists]

2821bis- meaning of "host"

2005-07-14 07:21:52

Robert A. Rosenberg <hal9001(_at_)panix(_dot_)com> wrote:
At 02:33 -0400 on 07/14/2005, Valdis(_dot_)Kletnieks(_at_)vt(_dot_)edu 
On Thu, 14 Jul 2005 00:58:32 EDT, "Robert A. Rosenberg" said:
At 14:51 +0100 on 07/13/2005, willemien(_at_)amidatrust(_dot_)com wrote...

section 2.3.10 has a "fuzzy" reference
"the host specified in the domain part of the address."
in general the host is NOT the domain part so what is meant by the 
host specified in the domain part of the address??
The address is of the form UserID(_at_)Domain (with the "Domain Part"
being everything to the right of the @-sign). The Domain part has a
host if there are more levels than the number defined for the WhoIs
registration. If the domain is but the Domain in the
address is X.SMTP.EXAMPLE.COM then X is the host in the SMTP
subdomain of domain EXAMPLE.COM.

Are you sure that basing it off the whois is the proper referent for this?

Yes. I am defining "Domain" based on ownership and control of that 
level (and those to the left of it) in a FQDN. For TLDs ORG/NET/COM 
and other non-National TLDs this is at the 2nd level. IOW, Panix.Com 
is a Domain Name and all *.Panix.Com FQDNs are owned/controlled by 
the owner of Panix.Com (who can be determined from a WhoIs lookup). 
For national TLDs such as UK or AU, there are addition levels before 
you get to the ownership level (IOW: If Panix were registered as a UK 
Domain it would be Panix.CO.UK not just Panix.UK). What I was 
pointing out (or trying to say) is that a .COM domain name is 2 
levels. If there was a 3-level .COM FQDN then the first (Left Most) 
level is a Hostname in the 2-level domain. Any 4 or longer .COM FQDN 
still has the first level as Hostname while all other levels to the 
left of the 2nd level are subdomains in that 2-level domain.

   I must admit to a little difficulty following this discussion.

   First, I must admit to having not yet completed my first reading
of 2821bis. So far I have two strong reactions:

1. It's _much_ too long; and
2. It _really_ needs a section on definitions.

   (Yes, I'm aware of the contradictory nature of these reactions.)

   In any case, I agree with Robert that his interpretation of "host"
in the context of 2.3.10 is as good as any I can think of.

   My problem is, when might "host" mean that in the rest of the
document? Is there really any hope of dozens of implementers coming
to the same conclusion about each meaning of "host"?

   I seriously recommend that we try to avoid any terms which mean
one thing in one part of the document, and something else in another.

   And I really do believe a section on definitions would help. (I'm
even willing to help put it together.)

John Leslie <john(_at_)jlc(_dot_)net>