MUST for the implementation of for the operator? (Was: version -01 of draft-ietf-dnsext-dns-protocol-profile sent

[Stephane Bortzmeyer <bortzmeyer@nic.fr>]

On Sun, Jan 20, 2008 at 05:14:18PM +1000,
 George Michaelson <ggm@apnic.net> wrote 
 a message of 94 lines which said:

> I have just submitted a 01 revision of
> draft-ietf-dnsext-dns-protocol- profile,

It seems we still have a meta-issue. Is the document mostly for
implementors or mostly for operations people? 

From the fact that it is on this WG, or from notes like in section 1
("In particular, normative directions on features..."), one could say
that it is mostly for implementors.

But it also has a normative reference to RFCs 1033 (a very old purely
"operations" RFC), 1912, 2010...

And the text on TCP, section 4.2.2, says:

> The server MAY accept TCP connections.

which is very different from the section on EDNS0 (which says that
implementation MUST support it for name servers MAY disable it). Why
not a MUST here? Every name server software I know supports TCP, even
djbdns (http://cr.yp.to/djbdns/tcp.html). If we make a document for
implementors, a MUST here seems reasonable.

PS: an editorial comment: it's the first time I see this document
repository in IETF references:

   [NLNet-1]  Wijngaards, W., "NSD Requirements and Specifications",
              July 2006, </dev/null>.

--
to unsubscribe send a message to namedroppers-request@ops.ietf.org with
the word 'unsubscribe' in a single line as the message text body.
archive: <http://ops.ietf.org/lists/namedroppers/>