Just 2p worth on this subject...
Bruno Vuillemin <Bruno.Vuillemin@???> writes
> Hm rather new ....March 1998.
Not really: the document had a long life (too long for comfort, in fact) as
an Internet Draft before it became an RFC, and existed as an experimental
method known to cognoscenti even before that. It's already widely used by
ISPs delegating micro-subnets, and it's almost certain to spread further.
Peter Radcliffe <pir@???> writes
> The problem with RFC 2317 is that its a hack.
Well, it depends on your definition of a hack :-) There's absolutely no change
of specification involved: nothing all the way back to RFCs 1033-5 says that
you can't have a CNAME pointing to a PTR. It just wasn't in practice done
until the idea of using it for delegation arrived.
And so, sadly...
> Some older resolvers just
> plain break. Most Sun resolvers (Solaris and SunOS with bind named hacked
> into libc are known good) are ok, but older versions of Linux, *BSD and
> others are not, to my knowledge.
> I can't get onto any of my more modern FreeBSD boxes to check those right
> now, but NetBSD 1.3.2 is ok. FreeBSD 2.1.6 or so was broken (they are now
> on 2.2.7).
The canonical broken resolvers were beta versions of 4.9.3 (up to beta25,
I think), and AFAIK those are what got into the older Linux and FreeBSD
versions.
Unfortunately, there are reliable reports that the Solaris 2.6 resolver
library [although not earlier Solaris 2.x ones] has a similar bug. I'm
still trying to determine the precise details.
Chris Thompson Cambridge University Computing Service,
Email: cet1@??? New Museums Site, Cambridge CB2 3QG,
Phone: +44 1223 334715 United Kingdom.
--
*** Exim information can be found at
http://www.exim.org/ ***
