--On Friday, 03 March, 2006 10:39 -0600 "Stephen Hayes (TX/EUS)" <[EMAIL PROTECTED]> wrote:
> > >> -----Original Message----- >> From: Brian E Carpenter [mailto:[EMAIL PROTECTED] >> I'm not sure we're in agreement. Let's assume that we switch >> contractors every two years. We can't have even the serial >> numbers (RFCxxxx for sake of argument) changing or resetting >> whenever we change contractor. So I think the IETF has to >> govern all types of identifier, including the basic document >> serial number. >> >> > So a possible requirement is that the publisher only uses >> > identifiers as instructed by the IETF and not also allocate >> > them to a publisher defined series. >> >> Yes. But I didn't get that from your previous paragraph. > > I listed this as a possible requirement due to the > implications. It means that there is no longer an overarching > serial identifier like the current RFC series. A serial > identifier allocated by the IETF is insufficient since > independent submissions do not go through the the IETF. A > serial identifier allocated by the publisher is insufficient > since the IETF can request it not be allocated one of those > identifiers. And the solution, of course, is two identifiers, which is what we have been doing for years, albeit badly for standards-track documents (for reasons that are clearly out of TechSpec scope). If there is a techspec requirement, I think it is that, if there is a serial document identifier (and, IMO, dropping it would be a terrible idea), the numbering system "belong" sufficiently to the IETF that, if there is a change in publisher, the new one can continue the numbers where the old one left off. This does _not_ require IETF management of the number space, consolidation of document numbers with standards-identifiers (numbers or otherwise), or any number of other too-complex possibilities. john > > Stephen _______________________________________________ Techspec mailing list [email protected] https://www1.ietf.org/mailman/listinfo/techspec
