[rfc-i] Towards Consensus

Andrew Sullivan ajs at anvilwalrusden.com
Fri May 25 10:10:35 PDT 2012

On Fri, May 25, 2012 at 11:46:34AM -0400, Phillip Hallam-Baker wrote:

> I find the arguments about 'Canonical' format to be total nonsense. As far
> as IETF tradition goes, the standard has always been set by bits on the
> wire rather than the documents.

I fully disagree with the claim that it is nonsense, and I fully
disagree with the claims about tradition.  

RFCs form an archival series.  That's why they're not allowed to be
updated.  For that reason, there must be one formally-blessed
publication format that is the one that is to be consulted when there
is a dispute about the meaning of a passage.  Otherwise, it is
logically possible (unlikely, but logically possible) that a bug in a
transformation for presentation could introduce a difference in
meaning.  (The most obvious way this could happen is incorrect
comment-line handling, where the source document includes a comment
line that mistakenly ends up in the document or processes as a comment
something that was in fact normative text.  But it doesn't matter how
it happens, as long as it is a possibility.)

As for tradition: more than once in the past few years in DNSEXT we
have been faced with disputes over what behaviour ought to be, and
differences among different deployed software.  The way we resolve
those things is often to compare them with the text (and then
sometimes, to decide the RFC was wrong, in which case the software in
question is still strictly speaking non-conforming to that RFC). 

Some of your other propositions I'm not sure I agree with (the xml2rfc
stuff I have no strong feeling about), but there must be exactly one
thing to which we can all appeal in the event of a dispute.  If we
want to call that "canonical", "basic", "fundamental", or "BillyBob's
Favourite Format", I don't care.



Andrew Sullivan
ajs at anvilwalrusden.com

More information about the rfc-interest mailing list