Re: the well-known address approach

["D. J. Bernstein" <djb@cr.yp.to>]

Paul Vixie writes:
> we can just designate the all-zeroes address in any given subnet for
> "default route service".  And so on.  This way lies madness

Taking away layers of indirection isn't always a good idea, but it's
inherently self-limiting. Madness lies in the opposite direction, where
people keep adding unnecessary layers of indirection. Examples:

   * ``Instead of specifying port 25 for SMTP, let's use SRV to declare
     site-specific SMTP ports!''
   * ``Instead of specifying IP addresses in NS records, let's specify
     host names, which will be resolved recursively!''
   * ``Instead of specifying a universal character set, let's have every
     piece of text tagged with its character set!''

There's no limit to the amount of ``configurability'' bloat that people
can---and do---add to their designs.

If there's a real advantage to indirection in the router address, DNS
cache address, etc., then someone should be able to state exactly what
that advantage is. So far, all I've seen are religious comments about
how this is ``madness,'' ``not the way the Internet works,'' etc.

---D. J. Bernstein, Associate Professor, Department of Mathematics,
Statistics, and Computer Science, University of Illinois at Chicago
#----------------------------------------------------------------------
# To unsubscribe, send a message to <dnsop-request@cafax.se>.