[R6RS] draft minutes of 20 September 2004 meeting

[William D Clinger]

Second meeting at Snowbird.
20 September 2004, 1:45pm
Present: Marc, Matthew, Kent, Will, Mike
(Manuel arrived later.)

****************************************************************

Agenda.

Finish going through the list of issues.

Also

  * macros                     Kent, Matthew
  * records                    Will (Kent, Marc will review)
  * arithmetic                 Will, Mike, Kent
  * modules                    Kent, Mike, Manuel
  * pattern matching           Manuel
    Unicode                    Matthew, Mike, Marc
    i/o                                        [look at SML]
    exceptions                 Will
    continuations

****************************************************************

Modules.

Kent presented a slide show.

A long discussion ensued.  Issues included the nature of
separate compilation, et cetera.

Kent and Matthew appear to be within range of agreement, but
asked for endorsement of general concept before trying to work
out the final details.  Manuel and Mike aren't so hot on the
general concept; they will get together and write up their own
counterproposal.

****************************************************************

Hash tables.

Manuel presented his proposal.  One issue is the use of keyword
syntax; another is whether certain features, such as capacity,
are really needed.  The distinction between hashtable-count and
hashtable-count-keys was unclear.  The general feeling is that
hash tables could almost be written in portable Scheme as a
library module, if hash-on-eqv? modules were available as a
primitive.  There is also an interaction with weak pointers.

****************************************************************

New issue:  weak-cons

****************************************************************

Records.

Will presented the record proposal he had written up, allegedly
based upon Marc's proposal, Chez Scheme, and Larceny.  There
were several problems with the proposal.  What people seem to
want is "static generative" types, where each type declaration
in the code generates exactly one new type, no matter how many
times that declaration is executed.

****************************************************************

Public archive of R6RS mailing list.

The general feeling was that the archives of the R6RS mailing
list should become public when R6RS is history.  Indeed, that
seems to be required by our charter.

A motion to open the mailing list to the public was moved and
seconded.  The motion failed on a 1-5 vote.

****************************************************************