Re: [Inip-discuss] Domain Names

[Andrew Sullivan <[e-mail]>]

On Sun, Jan 10, 2016 at 03:43:19PM -0500, Avri Doria wrote:
> 
> Is this a definition specific to the IN class domain name space?

That's an excellent question.  Sitting parked on my wet-drive (the thing
in between my ears before it comes out my fingers) is an I-D entitled,
"DNS CLASS Considered Useless"

If you read carefully the way CLASS is described, it turns out to
assume that the name tree is parallel.  If that is true, then some
things follow:

    1.  All names must subsist in all classes.
    2.  No name with a cross-class type can result in different
    resolution across classes.

If those are both true, then classes simply don't work, because of
[CD]NAME.  And if you think about the way the DNS is defined, this is
not surprising.

DNS is defined as a name space subordinate to which is the type and
class.  If classes worked, then the namespace would need to be
subordinate to class: you'd look up a name under a class with a
specific type.  But that's not how it is defined.

If you look at the very earliest namedroppers archives, actually, it
seems that this was very much the concern some commenters on DNS were
asking, though they didn't put it quite this starkly.  I think that
may be because they didn't yet have the experience with DNS that we
now do.  But I think those earliest questions were right on, and I
think that the complete failure of classes to take off is a pretty
good indication they were right.

I swore I'd submit that I-D before the end of 2015, but it's not even
been typed out yet.  I am still keen to make it happen.  I think we
should close the CLASS registry and admit that it's the appendix of
the DNS.

A

-- 
Andrew Sullivan
[e-mail]