Just what *is* a hostname (rfc952/1123, etc)

[Barry Margolin]

In article <a2n7aj$qjg at pub3.rc.vix.com>,
David Carmean  <dlc-bu at halibut.com> wrote:
>
>On Wed, Jan 23, 2002 at 03:44:43PM +0000, Barry Margolin wrote:
>
>> David Carmean  <dlc-bu at halibut.com> wrote:
>
>[snip]
>
>> >So far I haven't been able to figure out from the RFCs whether a 
>> >hostname is just the leftmost label in a domain name, or whether 
>> >the *entire* domain name (fully-qualified or relative) must be 
>> >considered subject to the RFC952/1123 restrictions?
>> 
>> What difference does it make?  IIRC, the rules about hostnames apply to
>> each label, not the fully-qualified name.
>
>It seems to be perfectly legal to create a domain named "foo_b#r&.com.", 
>if I'm reading RFC2181 correctly; the question is whether the name of 
>any node in the entire subtree below that zone cut can be a legal 
>Internet hostname.

I don't think so.  The name of a node in that zone would be something like
hostx.foo_b#r&.com.  The hostname rules apply to each and every label in
the name, and since the middle label breaks the rule, it's not a valid
hostname.

Perhaps a better way to answer this is that non-fully-qualified name are
just an abbreviation mechanism that can be used in certain contexts
(e.g. in zone files they're just a shorthand for the name with the $ORIGIN
appended, and when invoking a resolver it will try appending the suffixes
in the search list).  As far as on-the-wire protocols are concerned, names
are always fully qualified.

-- 
Barry Margolin, barmar at genuity.net
Genuity, Woburn, MA
*** DON'T SEND TECHNICAL QUESTIONS DIRECTLY TO ME, post them to newsgroups.
Please DON'T copy followups to me -- I'll assume it wasn't posted to the group.