Date:        Tue, 1 May 2001 19:46:05 +0300 (EEST)
    From:        Pekka Savola <[EMAIL PROTECTED]>
    Message-ID:  <[EMAIL PROTECTED]>

I know I should stay out of this discussion, but ...

  | In mathematics, if you want to prove something [esp. complex], you just
  | can't toss a 1000-page book in front of someone and say "Believe me, it's
  | there! It really _is_ there!

Nonsense, of course you can.  In fact, in Mathematics, it is more common
to just name the theorems after those who proved them first, and then
just say "by fred's theorem" without even providing a reference to where
the thing can be found.

Then there's a difference between showing a proof, and teaching that to
someone.  To teach adequately, you need to do more than point someone at
a reference (usually).

I have no idea how this side of the discussion moved to this list, this
is really namedroppers material - it concerns exactly the definition of
the DNS, and isn't in the slightest bit concerned with ipngwg.   But
just this once...

  | I have the impression that most people consider it either "grey" or
  | wrong; that was the way it was understood by the implementors at least.

Ohta-san is correct.   The problem is that he isn't bothering to even
explain what he is claiming, not that he isn't bothering to prove it.

The DNS requires glue, it requires glue wherever it is necessary.

It is obviously necessary (necessary by inspection) when a domain is
delegated to servers within the domain.  Any other time discovering
whether it is necessary or not requires more work.

Some implementors (recent BIND I believe, though I'm not sure about bind9,
and I know nothing at all about the other implementation mentioned here)
have restricted glue to where it is obviously necessary.   That's not
enough, glue needs to be possible anywhere it is necessary, not only
where it is obviously necessary.

The aol.net delegated to aol.com, and aol.com delegated to aol.net
example (as stupid as such a setup might be) is perfectly legal, and
is a situation where glue is necessary.  It isn't obviously necessary
however.   Some servers refuse to allow the glue to be loaded in a zone.
Those servers are broken.   Other servers (or back end resolvers) reject
the glue if they receive it in an answer to a query.  They're broken as well.

If you want proof that the DNS requires glue - go look in 1034/5.
If you can't be bothered, that isn't my, or anyone else's problem.

Now can we end this absurd name calling exercise please - or if it wants
to continue, take it back to [EMAIL PROTECTED] where the DNS experts
hang out.

kre

--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to [EMAIL PROTECTED]
--------------------------------------------------------------------

Reply via email to