COM zone and serials

Mailman List Mirror (Read Only)
1

FYI the host(1) command offers a MUCH simpler way of checking these
sorts of things. Just a single command, and a single option "host -C
zone" (or add '-u' to do the query/queries with TCP). IMNSHO host(1) is
far easier to use, and far more versatile, than any other DNS query tool.

The latest version of host(1) can always be found here:

  ftp://ftp.nikhef.nl/pub/network/host.tar.Z

Namespace collision!

ftp.nikhef.nl has versions of programs that are very similar to
"canonical" distributions but are different in option syntax and
features. Usually somewhat more featureful, but not always.

For instance, traceroute that differs from the canonical version
(ftp.ee.lbl.gov).
"host" that differs from any of the canonical versions (like that
distributed with bind), etc., etc.

I think promulgating their usage as long as they conflict in name
with more popular versions of the same programs is A Poor Idea.

--jhawk

2

[ On Wed, September 16, 1998 at 20:11:28 (-0400), John Hawkinson wrote: ]

Subject: Re: COM zone and serials

For instance, traceroute that differs from the canonical version
(ftp.ee.lbl.gov).

Although the verions of traceroute at ftp.ee.lbl.gov may be the root of
all traceroutes, I'd never consider it to be the most portable one.

That said though I must admit I've stopped using Eric's version of
traceroute in favour of the version that comes with NetBSD. However in
the days I was still using SunOS-4 on my primary server(s), Eric's
version was enormously better than the LBL version, and much easier to
get working.

"host" that differs from any of the canonical versions (like that
distributed with bind), etc., etc.

I think you'll find the 'host' in the "tools" sub-directory of the BIND
distribution is an extremely ancient predecessor to the version Eric
maintains and that the one in the "contrib/host" sub-directory is in
fact a slightly out-of-date version of Eric's host(1).

I.e. Eric's host(1) *is* the "canonical" version.

I think promulgating their usage as long as they conflict in name
with more popular versions of the same programs is A Poor Idea.

What you call a "canonical" version simply may be the "original
predecessor". Even I maintain and distribute enhanced versions of
programs that originally came from elsewhere (eg. fingerd, newsyslog,
etc.). I rarely, if ever, change their names (though usually I give the
distribution tar archive a different version name :-).

"Use what works. Use what works best."

3

jhawk@bbnplanet.com (John Hawkinson) writes:

... "host" that differs from any of the canonical versions (like that
distributed with bind), etc., etc.

I include the NIKHEF version of "host" in the contrib/ directory of BIND.
I have not put it into bin/ and replaced Charles Hedrick's version (from
which the NIKHEF version was derived) mostly because the NIKHEF version
does not subscribe to BIND's various portability-level #include's but
also because ISC is not the definitive source of the NIKHEF version of
"host" -- and the "bin/" subtree of the BIND source has only things we
are the definitive source of.

But I personally use the NIKHEF version on my own systems since it is a
lot more featureful and better implemented than the BIND version. (That
it, I use NIKHEF "host" whenever BIND "DiG" won't do what I need.)