[rfc-i] What standards?

Bob Braden braden at ISI.EDU
Tue May 4 14:24:13 PDT 2004


I just noticed your Feb 2004 Internet Draft
"draft-loughney-what-standards-01.txt", and I like it.  It makes an
important point: the IETF community has (relatively recently, on the
Internet time scale) fallen into the very bad habit of referring to a
protocol by an RFC numbers rather than by name.  RFC numbers are labels
on specification DOCUMENTS, not the names of the protocols these
documents specify.  When a specifications is modified, a new RFC is
issued, but that should not change the protocol name.  The email
protocol is "SMTP" (or "Simple Mail Transfer Protocol", in a more
relaxed world than the one we inhabit), and the protocol is now defined
in RFC 2821. Yet I frequently hear people refer to it as "RFC 821", and
I see confusion arising as a result.

STD numbers were an attempt to provide a distinct name space for protocols.
But they were designed in the good old days when we naively assumed that
all successful protocols would fairly quickly progress from Proposed Std to
Draft Std to Std category, so STDs are applied only to Standard documents.

If you look at the Official Internet Protocol Standards tables, which
Jon started publishing in 1988 (RFC 1083), each protocol has a name,
like Transmission Control Protocol, and a mnemonic, like "TCP".  In
many cases Jon had to invent mnemonics and the results got a little
wierd at times, but it did solve the problem you are concerned with.

We have dropped the definition of protocol mnemonics as Jon had them,
and perhaps they need to be resurrected.  A new take would require some
thought to create structure; the protocol world is much more complex than
it was in 1988!

Bob Braden

More information about the rfc-interest mailing list