Message-Id: <9310200403.AA12662@thud.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: Jon Knight <J.P.Knight@lut.ac.uk>
Subject: Re: Final edits - summary of outstanding points.
In-Reply-To: Your message of "Wed, 20 Oct 1993 00:50:39 -0000."
<9310192350.AA05355@sun-co15.lut.ac.uk>
Date: Wed, 20 Oct 1993 00:03:38 -0400
> Recently Mitra said:
> >
> > 17) Do we allow trailing "." in FQDNs - I forget who took which side here
> ?
> >
>
> I vote for allowing the trailing ``.'' in FQDNs in URLs. Although many
> people drop the trailing dot in day-to-day usage, the _proper_ format of
> a FQDN does have a trailing dot.
That's odd. I've always thought a proper FQDN omits a trailing dot. The
grammars listed in RFCs 1035, 821 and 822 forbid a trailing dot.
> This can also make DNS lookups
> slightly faster and safer as the software knows not to attach any local
> domains to it (there was a recent thread in one of the DNS or BIND
> related mailing lists about an obscure security flaw with some BIND
> versions when you present a domain name without a trailing dot).
Ah, now I see what you are talking about. The syntax used to configure
domain name servers does use a trailing dot to signify absolute domain names
as opposed to those relative to some other RR. (RFC 1034, section 3.1)
I suggest that the use of trailing dot to mark a domain name as "absolute"
should be confined to DNS config files. DNS config files need such a
mechanism, since they will be defining several RRs within a single domain,
and it would be tedious and error-prone to list the FQDN each time.
URNs need to be unambiguous by themselves, so they should always use
absolute domain names. Therefore no trailing dot is needed.
Implementations of gethostbyname() that append default domains to domain
addresses (i.e. those already containing a '.') are BROKEN. Software which
makes use of such broken routines can append the '.' themselves as a
workaround, but we should not require it in URLs.
Keith Moore