--On Monday, 30 October, 2006 18:10 -0500 Edward Lewis
<Ed.Lewis@xxxxxxxxxxx> wrote:
> At 17:38 -0500 10/30/06, John C Klensin wrote:
>
>> It seems to me that a reference from the draft to the code
>> description or to Bind 9 more generally, with a note to the
>> effect that Bind 9 is believed to contain an implementation of
>> what is being described in the document, could head off a
>> great deal of confusion... including all of the confusion we
>> have seen in the last week or so.
>
> The DNS WGs[1] have worked hard to stress that BIND is not
> DNS. BIND is not the reference implementation of the DNS
> protocol. With that in mind I think it is wrong to have a
> statement in the document declaring that "the *definition* is
> compliant with BIND 9."
That isn't what I said, and I certainly agree with the
principle. I was suggesting a note that indicated that the Bind
9 implementation was believed, by some, to be compliant with the
definition (or compliant as far as it goes, or whatever). That
is all: no definitional compliance with Bind 9 (or anything
else), not an IETF assertion that the Bind 9 implementation is,
in fact, conformant, etc.
On the other hand, unless the DNS WGs want to standardize this,
it doesn't seem to me that the above matters a lot. Sam could
be describing an interesting approach that is similar to the
implementation, or the implementation could be similar to Sam's
approach, or any number of other situations could apply. I
believe that, for an _informational_ document, it would be
useful for Sam to make some comment about a known
implementation. That comment could be "that implementation is
believed to be consistent with this spec and, if it isn't, it is
wrong and should be fixed" or "the author of this spec has had a
discussion with the authors of Bind 9, they are not
intentionally inconsistent, and, if they turn out to be, the
code will be fixed" or anything in between.
Coming back to Phil's comment, I don't believe that the IESG
should be trying to guarantee anything. However, when the IESG,
or some AD, takes it upon itself to sponsor a non-WG
informational document, rather than sending the author off to
make an independent submission to the RFC Editor, it seems to me
that the IESG takes on the same responsibility to not request
publication of documents that raise unnecessary questions or
cause unnecessary confusion that the RFC Editor does for
documents submitted directly. And, if the existence of an
implementation is known, some sort of brief discussion of known
differences (if any) between the specification and the
implementation, and maybe even some comments about why the
specification is preferable, would seem to me to be very much in
order. I hope that no one will read that comment as suggesting
that the IESG do any special work -- all that is needed is to
say "we Last Called this, it appears that some of the community
has asked for some clarification on this point and no one has
objected, so why don't you supply that before publication".
john
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf