[R6RS] my notes on today's conference call (4 July 2006)

William D Clinger will at ccs.neu.edu
Tue Jul 4 13:56:13 EDT 2006

Conference call July 4 2006 12:45pm
All present by about 12:46pm:
Kent, Matthew, Will, Anton, Mike

0. finalize agenda (1 minute)

1. action items from 6/27/2006 (5 minutes)
   - draft and commit eval/environment proposal (Kent)
   - review hash-table draft before 7/04 meeting (All)
        to discuss today
   - review safe/unsafe mode draft before 7/04 meeting (All)
        assumed done
   - update hash table draft---see 6/27 minutes (Anton)
   - carried over:
     - update library srfi (Matthew and Kent)
        note sent out concerning syntax; more coming on phases
     - update reference implementation for arithmetic SRFI (Will)
        complete except for naming issue, to be discussed today
     - complete unicode reference implementation (Will)
        in progress
     - comment on core/library split (All)

2. naming conventions for fx=/fx=? etc. (10 minutes)
   - option 0: = fixnum= fx= fl= exact=? inexact=?
   - option 1: = fixnum= fx= fl= exact= inexact=
     (no question mark for predicates corresponding to =, <, etc.)
   - option 2: = fixnum=? fx=? fl=? exact=? inexact=?
     (include question mark if predicate name constains any letters)
   - other options?
        Kent suggested option 3:
        pronounceable names have question marks at the end;
        unpronounceable don't (e.g. fixnum=?, fx=)
    option 0: 
    option 1: Will, Anton
    option 2: Mike
    option 3: Kent, Matthew
    option 1: Will, Anton
    option 3: Kent, Matthew, Mike
Option 3 has the majority, but option 2 might be a better compromise:
Runoff between 2 and 3:
    option 2: Will, Matthew, Anton, Mike
    option 3: Kent
We'll go with option 2
   - should we rename char/string equivalence predicates as well
     if we go with Option 1?
        We didn't go with option 1.

3. multiple return values (10 minutes)
   - revisit exceptions vote #42
   - see https://r6rs.scheming.org/node/286
   - see 6/27 minutes
        Non-binding straw poll (because some aren't ready to vote):
        Kent, Matthew, and Anton prefer Kent's preferred semantics
        Will, Mike prefer Will's preferred semantics
        Non-binding straw poll on leaving it unspecified:
        Will and Kent okay with that;
        Matthew, Anton, and Mike not okay with that
    tabled until next week; must vote next week

4. hash tables (10 minutes)
    changes detailed in recent email; none have had time to read
    Mike worried about the number of -ref and -update! procedures
    Mike doesn't like the exception semantics when a key isn't in
        the table
    Mike moved to ditch hashtable-ref/call, hashtable-ref/thunk,
        and hashtable-ref/get
    Will seconded
    motion passed unanimously
    We are left with hashtable-ref, which takes an optional
    Will moved that the optional argument become mandatory
    Mike seconded
    motion passed without objection
    Mike moved to make the corresponding changes to the update!
    Kent didn't like that because of the potential inefficiency
    Spec needs to clarify whether the hashtable is allowed to
        cache lookups in a way that reduces the number of calls
        to the hash function
    Spec needs to *insist* that the hash function be a pure
        function; Will volunteers to take a whack at cleaning
        this up

Not enough time left for the remaining items:

5. eval/environment (10 minutes)

6. simplified condition type hierarchy (10 minutes)

7. I/O (10 minutes)

Mike asked us to think about setting agendas for the rest
of the summer.

8. adjourned about 1:48pm

More information about the R6RS mailing list