[R6RS] draft minutes of 21 September 2004 meeting

[William D Clinger]

Third meeting at Snowbird.
Tuesday, 21 September 2004, 8:15am
Present: Marc, Kent, Will, Manuel, Mike
(Matthew arrived later.)

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

New issues.

Do we really want to keep:
    proper tail recursion
    call-with-current-continuation

Add symbols of the form ->whatever.

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

More module discussion.

Should programmers be required to list implicit imports?

Are exported variables mutable?
Indeed, are variables exported, or are values exported?
    (General opinion: variables are exported.)

Kent wants exported variables to be mutable.  If explicitly
exported variables are not mutable, and we allow implicit
exports without requiring them to be explicitly declared,
then implicitly exported variables should not be mutable
either.

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

New issue.

Should there be a specified unspecified value?

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

List of issues.

We then proceeded to march through the list of issues, trying
to take easier issues first, and voting.

Multiline comments.  Voted 4-0-1 to add #|...|# comments (but
to fix the bug in their specification).  Voted 4-1-0 to add
#;<datum> comments.

(Matthew arrived.)

Square brackets.  Voted 4-1-1 to allow [ and ] to be used in
place of ( and ), but require square brackets to match square
brackets and round brackets to match round brackets.

External representation for circular structures.  Voted 2-2-2
on "Should <DATUM> include #n=<DATUM> (need not precede the
first reference) and #n# syntax?".

#!eof.  It was pointed out that adding #!eof is actually an
incompatible change to the language, since the R5RS says that
an end-of-file object is not the representation of any object
that can be read by READ etc.  We took several votes on this.

Voted 4-1-1 to require only one end-of-file object.
Voted 2-3-1 on "Is there a readable syntax for that object?"
It was moved that the readable syntax be #!eof, but this
motion died for lack of a second.
Voted 6-0-0 to require the end-of-file object to be distinct
from other types.
Voted 3-0-3 to have a procedure that returns the end-of-file
object.

More escape characters.  Voted 6-0-0 to allow \n in string
constants.  Matthew, who is looking into Unicode, will write
a proposal for other escape sequences.

Require #f, #t, and characters to be followed by a delimiter.
Voted 3-0-3 for this requirement.

Add new number syntaxes.  It was pointed out that +0 and -0
are both exact integers that are = and eqv? to 0.  The real
question is whether we should have a standard syntax for
inexact infinities and NaNs.  Considerable discussion of the
syntax ensued.
Voted 3-3-0 with regard to +inf.0, -inf.0, and +nan.0.
Voted 3-3-0 with regard to +inf.0*, -inf.0*, and +nan.0*;
then Kent changed his vote from positive to negative, making
the vote 2-4-0.

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

Arithmetic.

Will then presented the report of the subcommittee on arithmetic.
There was not enough time available to discuss this issue.

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