Re: [dnsext] What is the 'canonical' spelling: RRset of RRSet ?

[Paul Vixie <vixie@isc.org>]

> Date: Thu, 15 Oct 2009 13:15:59 -0400
> From: Donald Eastlake <d3e3e3@gmail.com>
> 
> > I admit to a strong personal bias in favour of "RRset".
> 
> +1

also +1.  as a bonus, here's some historical background on the term itself.

RFC 2136 was the first time the atomicity of resource record sets denoted
by <name,class,type> was made explicit, because it was very important to
the operation of the DNS UPDATE protocol.  i coined the term "RRset" for
RFC 2136 but the atomicity itself was always implicit in RFC 1034 and 1035,
so it was not new information.  BIND itself wasn't as strict about resource
record set atomicity until version 4.9 -- which was the first release i had
a hand in -- and so it was a hot topic for me personally during the
production of RFC 2136.  an early draft of DNS UPDATE attempted to make
updates on a per-RR basis, and DNS IXFR as far as i know still does so.