Stewart Bryant <stewart(_dot_)bryant(_at_)gmail(_dot_)com> wrote:
Having thought a little more about this, I am wondering about
unintended consequences in the 5K documents that we have
written since RFC2119 was published.
Actually, there is an obvious meaning of citing RFC xxxx or BCX xx:
which implies, "at the time of this publication".
Nonetheless, I urge including text if leiba-rfc2119-update becomes an
RFC: stating that going forward, a reference to RFC 2119 will continue
to include-by-reference RFC 2119 as published, whereas a reference to
BCP 14 will include-by-reference whatever BCP 14 may be (or have been)
at the time any RFC referencing it is published.
(This really doesn't change much of anything; but the opportunity for
confusion is _so_ great that I'm sure it will arise.)
We must leave it to the RFC Editor to do whatever _can_ be done to
ensure that nobody "unintentionally" cites RFC 2119 or BCP 14 in such
a way as to cause further confusion.
If we effectively change RFC2119 as we propose, is there a danger that
readers will incorrectly interpret old text with new semantics.
RFC 2119 _cannot_ change.
Nor, alas, _can_ we change the confusion in existing documents which
cite RFC 2119.
We can only reduce that confusion going forward -- and that only if
the RFC Editor agrees to discourage citing RFC 2119 in future documents.
I have no idea whether anything of significance will occur but
considering the thought put into terms like SHOULD there exists a risk
that would be mitigated if we picked a new RFC number whereupon the
reader would know which definition the writers and reviewers were using.
I would have recommended this, too, if I were writing the document.
But I stand by my previous statement: we can only reduce the confusion
if the RFC Editor discourages citing RFC 2119 after draft-leiba becomes
an RFC.
--
John Leslie <john(_at_)jlc(_dot_)net>