On Tue, 7 Dec 2004, Mark Andrews wrote:
> As you would have already seen underscores are not permitted
> by RFC 952. The are also not permitted by RFC 1034 which say:
> The idea is that the name of any
> existing object can be expressed as a domain name with minimal changes.
> However, when assigning a domain name for an object, the prudent user
> will select a name which satisfies both the rules of the domain system
> and any existing rules for the object, whether these rules are published
> or implied by existing programs.
> In this case the rules were published.
> AD itself uses underscores to create heirachies in the namesepace
> that do not clash with hostnames. It does this by using underscores.
> SRV uses underscored so as to not clash with hostnames.
> Other services also use underscore to avoid clashing with hostnames.
> Be aware thay _gc contains a A record so if you are running a version
> of bind which supports hostname checks (BIND 8, 9.3.0) you will need
> to disable the checks atleast for this name.

As always, succinct and to the point. My thanks Mark this is exactly the
info I needed (eg: the _ is a convention used by AD, and allowing that to
creep into the rest of the namespace is not technically allowed,
regardless of one's ability to actually do it.)

As to the people who were for or against user shooting themselves, it's
kinda hard across a large campus to manage user education DNS be really
nice about it. We have a very nice web front end that parses requests,
does error checking and makes them easily digestible for admins to insert.
The question is if this interface should reject things with an _ in them
(and present the user an error) before sending them to us (which then
causes work for us to contact, explain, etc..).

Thanks to everyone for their input, it's been very helpful!


nathan hruby
uga enterprise information technology services
production systems support
metaphysically wrinkle-free