Scheme RFCs

[Alan Bawden <Alan@lcs.mit.edu>]

I agree with Olin that the proposed SRFI mechanism seems too restrictive.

When the idea of a Scheme RFC mechanism was first proposed, I naturally
assumed that we were talking about emulating the highly successful Internet
RFC process.  I imagined that the standards for accepting a Scheme RFC
would be kept relatively open.  If it was in the right format, it was
intelligibly written, and it was relevant to the project of improving the
art of Scheme, it would be accepted.  Proposed Scheme RFCs would be
reviewed by the Scheme RFC editor and possibly by other reviewers.  The
editor or reviewers might suggest clarifications and improvements, but in
the end the editor would have the final say, and she could only reject a
Scheme RFC if it was badly formatted, incoherent, or irrelevant.

Publication as a Scheme RFC would not be a stamp of approval about the
implementability of the proposal.  If something wasn't implementable,
implementors would discover that fact soon enough.

Documents could be considered for publication as Scheme RFCs other than
descriptions of specific extension packages.  The Internet RFC series is
full of documents that do not describe new proposed protocols or standards.
Let me pull two out of the catalog to illustrate the other kinds of
documents published as Internet RFCs:

  o RFC 2416: "When TCP Starts Up With Four Packets Into Only Three
    Buffers".  This RFC documents the results of an experiment performed by
    the authors.  It's not a result that's really suitable for a conference
    paper or journal article -- I gather that they are just settling an
    issue raised during a mailing list discussion -- but its an interesting
    little factoid worth preserving as part of the collective networking
    wisdom, so publication as an RFC is perfect.

  o RFC 2291: "Requirements for a Distributed Authoring and Versioning
    Protocol for the World Wide Web".  This RFC is the product of an IETF
    working group that is trying to develop a new standard.  The group
    thought it was a good idea to start by publishing exactly what their
    -goals- were.  Publication as an RFC is a good way to solicit feedback
    about whether they are heading in the right direction.

Requirements documents, such as RFC 2291, seem to me to be something that
the Scheme community could use more of.

I also note that under the current guidelines the document that describes
the Scheme RFC mechanism does not itself qualify as a Scheme RFC.  I don't
see why it shouldn't.  And how about the minutes of the Scheme workshop?
Surely they should be part of the permanent record.

Seems to me that such an open RFC process could work very well for Scheme.