I disagree with Valdis' description. I actually worked for a standards
organization that used a codebase as part of the definition of the
standards it produced. The concept of "Running code" is good, but where
are the IETF reference implementations? If "running code" is the actual
standard, then you have to have some code to run. This is not the way the
IETF has ever done things. The IETF references to "Running code" simply
mean that there have been (or must be) demonstrations that the standard
works, and is not a hypothetical document awaiting implementation. But
"running code" is not the actual IETF standard, as Valdis seems to be
claiming. The actual standard is the text of the RFC in the "standard"
state. The status of each RFC is given in the current "Internet Offical
Protocol Standards" which is updated every hundred RFC's, and is
currently RFC 3600. Reviewing the RFC's with this title (or rather the
changes between each) will give you an idea of the process the IETF goes
through. Compare this to RFC 2026 and other documents.
Also, it was my understanding that 2119 and 2026 were "BCP" status because
they didn't refer to computer protocols but rather the rules of the
organization, and can't be properly classifed as "standards". They reflect
the current practice of the organization. However, it is not the case
that they were meant ONLY to be vaguely 'generally useful' ideas that
could be ignored by the IETF. Although under the current leadership, the
rules haven't been followed very closely. This is a failure of the current
leadership, not a failure of the organization to adopt rules. An
organization can't run without rules. Otherwise, it isn't an organization,
its a Cluster-bleep.
Rules that aren't really rules, aren't rules at all. A standards
organization that has no rules can't be open, transparent and fair. If it
is not open, transparent, and fair, then it is closed, opaque, and
proprietary. And if its not open, transparent, and fair, then it isn't
really a standards organization, but a good-old-boy/chums network
operating to the advantage of that group, attempting to create defacto
standards just like other proprietary groups such as Microsoft. While
this might have been OK many years ago, now far too many people have a
stake in the Internet and in Internet Protocols to let the Internet be
specified by a closed, opaque, proprietary good-old-boy operation.
Your goal of having government RFP's refer to STDxxx's is good. It has
been pointed out that relatively few documents ever reach this stage, but
the most important ones do. If the IETF were to attend more to the
administrative procedures it has adopted, then more RFC's would either
reach the "mature" STD or BCP or INFORMATIONAL level or be dropped after
development stops and interest fades. I suspect that some old RFCs' are
kept around as a way for the leadership to promote bad ideas that never
caught on, but that the leadership still favors and hopes someday will
catch on, even though interest in further development was lost long ago.
--Dean
On Thu, 3 Jun 2004 Valdis(_dot_)Kletnieks(_at_)vt(_dot_)edu wrote:
On Thu, 03 Jun 2004 11:04:06 PDT, Joe Touch said:
STD-5 is a nice choice - it actually refers to 6 different RFCs.
So which one redirects to STD005.txt, and what is in it?
(To see this noted in the RFCs themselves, see STD-62, which refers to a
set of 8 different RFCs.)
And what happens when a STD is updated/revised?
This is the IETF, we believe in rough consensus and running code. Note
carefully that RFC2119 and RFC2026 are both BCP rather than Standard,
indicating that even the meta-standard of how we produce standards is "running
code" rather than an actual standard itself... ;)
But anyhow, if we ever update STD005, we'll just do the obvious - create
STD079 or whatever we're up to, stick an "Obsoletes: STD005" on it, and stick
an "Obsoleted By: STD079" on STD005, and the RFC Editor will turn the crank
and
make sure all the appropriate indexes and webpages are updated to match.
Again, this is all "running code" - we know how to issue an RFC that updates
or
obsoletes one (modulo the occasional corner case that crops up), and there's
no
obvious reason that the scheme that works for I-D, Draft, Proposed shouldn't
work for a full Standard...
_______________________________________________
Ietf mailing list
Ietf(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/ietf