[R6RS] my notes on today's conference call (27 June 2006)

[William D Clinger]

Conference call June 20 2006 12:45pm
All present by about 12:46pm:
Kent, Mike, Will, Anton, Matthew

0. finalize agenda (1 minute)

Will's motions re multiple return values

1. action items from 6/20/2006 (5 minutes)
   - submit status report (Kent)
        done
   - submit syntax srfi (Kent)
        done
   - commit latest enumerations draft (Will)
        done
   - review byte-vector draft before 6/27 meeting (All)
        done, presumably
   - review hash-table draft before 6/27 meeting (All)
        done, presumably
   - review enumerations draft before 6/27 meeting (All)
        done, presumably
   - review safe/unsafe mode draft before 6/27 meeting (All)
        done, presumably
   - carried over:
     - update library srfi (Matthew and Kent)
        in progress
     - update reference implementation for arithmetic SRFI (Will)
        no progress since last week
     - comment on eval/environment proposal (All)
        done; proposal will be drafted
     - complete unicode reference implementation (Will)
        no progress since last week
     - comment on core/library split (All)
        in progress

2. wrong number of values (10 minutes)
    Will moved:
        The second (alternative) expansion of begin, from
        R5RS 7.3, shall be revised for consistency with the rest
        of the R6RS, but some such alternative expansion of begin
        shall remain in the corresponding section of R6RS.
    Mike seconded
    Vote:
        yes: Will, Mike
        no: 
        abstain: Kent, Matthew, Anton
        (Motion passed by a plurality)
   - passed to a procedure
     - should we require an exception to be raised?
     - see https://r6rs.scheming.org/node/286#comment-1472
        Kent moved that an exception must be raised
        if a procedure receives a number of arguments
        it does not expect (as determined by the formals
        of the lambda expression from which the procedure
        was created)
        Will seconded
        Vote: yes, unanimously
   - passed to single-value continuation
     - revisit exceptions vote #42
     - see https://r6rs.scheming.org/node/286
        three proposals and a related wrinkle:
            Kent's preferred semantics
                begin continuations accept any number
                cwv accept any number and pass along to receiver
                all other continuation accept exactly one
                    must raise an exception if some other
                    number of values is received
            Will's preferred semantics
                begin continuations accept any number
                cwv accept any number and pass along to receiver
                all other continuation accept any number
                    of values and coerce as follows:
                    zero values: coerce to #!unspecified
                    more than one: ignore all but first
            Will's alternative semantics
                cwv accept any number and pass along to receiver
                all other continuations accept zero or one value
                    zero values: coerce to #!unspecified
                    two or more values might be allowed to raise
                        or required to raise an exception
            related wrinkle that becomes possible with the
                second or third semantics above:
                set!, set-car!, set-cdr! could be required or
                allowed to return 0 values
        tabled for a week so people can think about it

3. byte vectors (5 minutes)
   - approve essentially as is?
        yes, with no objections

4. Bawden quasiquote extension (5 minutes)
   - adopt despite r5rs compatibility?
        really an extension, not an incompatibility
        examples are present in email
        Mike moved to adopt, Kent seconded
        Vote:
            yes: Kent, Matthew, Mike, Anton
            no:
            abstain: Will
        adopted

5. enumerations (10 minutes)
   - approve essentially as is?
        approved without objection

6. hash tables (10 minutes)
   - approve essentially as is?
        not yet; some changes still need to be made, e.g. to
            use enumeration sets
        Kent cited customer who found use for initial capacity
            (because hash function might be expensive)
   - limit to eq? hash tables?
        Kent, Anton, Will want more than just eq?, eqv? tables
   - include equal? hash tables?

Did not get to this agenda item:

7. equal? generalizations (10 minutes)
   - generalize to one or more arguments
   - add pred-equal? (equalp?)

next week:
multiple return values
hash tables
equal? generalizations
eval environment
i/o
Matthew's proposal to reduce condition type hierarchy

8. adjourned at 1:54pm