Re: regarding the respsize draft: preferring glue of certain types

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

Paul Vixie writes:
> to prevent pathological starvation, at most one glue rrset of each type
> should be present even in a truncated response, unless the rrsets
> are such that the smallest rrset of each glue type would not fit.

I found this unnecessarily confusing. Can we please reserve the word
``truncation'' for the TC bit? See RFC 1035.

There are three separate issues:

   (1) the strategy for adding delegation records;
   (2) the strategy for adding AU+AR to authoritative responses; and
   (3) how to handle truncation.

For example, with tinydns,

   (1) NS and corresponding A glue is always added to delegations;
   (2) AU (or AU+AR) is added to authoritative answers only if the
       entire AU (or AU+AR) ends before the 512-byte boundary; and
   (3) if truncation is required, all records are eliminated.

As for mixing A and AAAA in delegations, I don't understand the benefit
to having both in the first place. If a child server provides only one
protocol, there's no decision for the parent to make. If a child server
provides both protocols, and the query to the parent was made through
IPv4, I don't understand the point of including AAAA; and vice versa.

---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>.