[bind10-dev] sorting RDATA in-memory

[JINMEI Tatuya / 神明達哉]

At Tue, 12 Feb 2013 00:47:55 -0500,
Robert Edmonds <edmonds at isc.org> wrote:

> > BIND 9 sorts RRs of the same RRset in the DNSSEC order when it loads
> > the records into memory (unless the special compile time options is
> > specified).  Is there any reason for this behavior?  Obviously we need
> > to sort them when we need to compute the signature of the RRset, but I
> > don't see the need for it in answering queries.  I'm curious about
> > whether there's any other reason for the convenience of calculating
> > the signature.
> 
> sorting the RRset is probably the easiest way to do duplicate
> suppression:

Yeah, I know BIND 9 exploits the internal sorted order to eliminate
duplicates.  But that's a matter of internal implementation
convenience.  I wondered whether there's a reason that's related to
externally visible behavior.

---
JINMEI, Tatuya