ISSUE: Document subseries name for IETF Administration
moore at cs.utk.edu
Mon Jun 16 22:22:22 CEST 2003
> Sure. In 1977, the RFC series, which was already 8 years old, dealt
> mostly with the ARPAnet. In 1977 [D]ARPA began funding research to
> develop a catenet (internet) of networks, based upon the TCP[/IP]
> defined in the Kahn/Cerf paper. Jon (and others?) decided that the new
> project should have a new document series, so they began the Internet
> Experiment Notes (IEN) series. This continued until 1982; Jon Postel
> was the editor of both RFCs and IENs. During the 5 year period, he
> published approximately 200 IENs and 100 RFCs.
> But Jon found it increasingly difficult to decide which series to use
> to publish a given document, and the two series caused a lot of
> confusion in the community. In 1982 he gave up on the IENs and did all
> further publication related to networking only in the RFC series.
> Jon came to the conclusion that having a primary numbering space that
> is mult-dimensional is bound to lead to confusion. There ought to be a
> single master index sequence number -- the RFC #s -- for all documents
> in the archive. For secondary categorization, to establish
> sub-collections, he invented the subseries (STD, FYI, BCP).
> (I see little reason to restrict the growth of such subseries,
> BTW, as long as you have the single unique index of RFC #).
A conclusion that was valid in 1982 with ~900 documents in the library
might not be valid 21 years later with ~3500 documents in the library.
We have some single protocols now with more pages of specification, and
more documents, than the entire Internet protocol suite had in 1982.
To be sure, there's nothing wrong with assigning a serial number to
each document, any more than assigning an ISBN to each book published.
But we don't tend to refer to such works by their ISBNs except in very
specific contexts. And we do find other kinds of document classification
More information about the Problem-statement