[R6RS] my notes on today's conference call (11 April 2007)

William D Clinger will at ccs.neu.edu
Wed Apr 11 10:17:22 EDT 2007

telephone conference April 11 2007 8:00am-10:00am EDT

Mike, Anton, Will, Kent present by about 8:02am;
Matthew (in China) joined temporarily at 8:27am,
on a very poor connection, and rejoined on a better
connection a few minutes later.  We skipped around
the agenda some to delay our consideration of agenda
items that needed Matthew's expertise or vote, but
my notes preserve the original order of the agenda,
not the order in which we considered the items.

0. finalize agenda (1 minute)

1. action items (1 minute)
   - record response votes (All)
   - draft concrete proposal for making conditions more like
     records---or not (ticket 210) (Mike)
   - draft concrete proposal for change in exception handling
     protocol---or not (ticket 221) (Mike)
   - redraft response for ticket 201 (Will) [done]
   - revise (shorten) draft response for ticket 200 (Kent) [done]
   - draft responses for ticket numbers
     208: eliminate library export immutability loophole
        to be done after today's meeting
     209: Syntactic datums and datum values
     210: Why are simple conditions so different from records
     221: Recursive exception handling considered harmful
        done last night
   - update ticket-nnn responses to reflect decisions:
     151: reject formal comment
     204: go with option B
     211: move when and unless into base library
     214: reject suggestion to replace "lookahead" with "look-ahead"
     216: reject comment; possibly add footnote acknowledging Boolean
     220: map should not modify earlier return values
     223: will add string <=> unicode conversion procedures with
          "replace" semantics
     229: mention that bytevectors can be used with record wrapper
          and srfi can replicate string functionality; string-ref
          and string-set *should* be O(1)
   - update r6rs to reflect additional decisions:
     - probably use "should" for programmer requirements that implementations
       are not required to enforce
     - change the specifications of char-alphabetic?, char-whitespace?, and
       string-titlecase to conform with Unicode 5?
     - adopt Posix semantics for set-port-position! when underlying
       object is a file or bytevector output port, with unwritten bytes
       having unspecified values
     - move formal semantics to a nonnormative appendix
   - deferred until after April 15
     - add port-length and port-has-port-length? procedures
       and corresponding arguments to custom port makers?

NOTE: tickets whose responses are not yet accepted and by whom:

 *135: Will, Matthew, Mike, Anton      150: Will, Matthew, Mike, Anton
 +151: Kent, Mike, Anton               152: Mike, Anton
 -155: Will, Anton                     156: Mike
  160: Mike, Anton                     161: Anton
 -164: All                            -165: All
  166: Mike, Anton                    -170: Will, Anton
 -176: Will, Kent, Matthew, Anton      183: Mike
  184: Mike                            186: Mike
 *189: Will, Mike, Anton               198: Mike
 *200: Will, Mike, Anton               201: Mike, Anton
  202: Will, Matthew, Mike, Anton     +204: All
  205: Mike                            207: Mike
 -208: All                            -209: All
 -210: All                            +211: All
  212: Mike, Anton                     213: Mike, Anton
  214: All                             215: Mike, Anton
 +216: Will, Kent, Matthew, Anton      217: Will, Mike, Anton
 +220: Will, Kent, Matthew, Anton     -221: Will, Kent, Matthew, Anton
 -222: Will, Matthew, Mike, Anton      223: Mike, Anton
 -224: Will, Kent, Matthew, Anton      225: Mike, Anton
 -226: Will, Kent, Matthew, Anton     -227: Will, Kent, Matthew, Anton
 +229: All

   - awaiting technical decision

   + response does not yet reflect technical decision

   * questions/comments:
     135: Kent needs to fix first bullet point [done]
     189: Anton would like help with this response
     200: Kent needs to shorten this response [done]

2. end game
   - we have four days left to publish responses
   - same process as last time?
   - volunteer to clean up tickets
        Anton volunteered
   - meetings before and after April 15
        deferred to end of conference

3. eliminate library export immutability loophole? (5 minutes)
   - ticket 208
   - motion 1: include implicit only-once wrapper around rhs of each
     library definition
     - options: should or must raise exception?
   - motion 2: include implicit only-once wrapper around rhs of each
     definition, letrec, and letrec* rhs
     - options: should or must raise exception?
        Kent makes motion 2 with "should"; Will seconded
            yes: Will, Kent, Mike, Anton
            no: Matthew

4. keep the other little library: (r6rs case-lambda)? (5 minutes)
   - we moved when-unless into (r6rs base), what about case-lambda?
        no one here opposes putting all three into a separate library,
        but what to name it?
        Anton moved we name it (r6rs control), and put
            when, unless, and case-lambda in it; Mike seconded
            yes: Will, Matthew, Mike, Anton
            abstain: Kent
        Mike moved we move do from (r6rs base) into (r6rs control);
        Matthew seconded
            yes: Will, Kent, Matthew, Mike, Anton
        Will moved we move call/cc and dynamic-wind into (r6rs control),
            but the motion failed for lack of a second

5. Export bindings for various literals? (10 minutes)
   - ticket 164
   - literals that can appear where non-literals are legitimate:
     - syntax-rules underscore (_), ellipses (...)
     - cond else
     - quasiquote unquote, unquote-splicing
     - quasisyntax unsyntax, unsyntax-splicing
     - others?
   - literals that can appear only where literals are legitimate:
     - case else
     - define-record-type fields, mutable, etc.
     - others?
   - literals in the library syntax not (presently) an issue
     - export, import
     - only, rename, etc.
     - and, or, >, etc.
   - tabled last week when Matthew dropped out
        go with the simple solution: export bindings for all identifiers

6. change formal semantics? (5 minutes)
   - tickets 226 (drop library toplevel), 227 (drop define and begin^f)
   - Will says we should do what the formal comment suggests
        we can say we will pass the comment on
            to the authors of the formal semantics

7. should formal semantics contain "complicating optimizations"
   - ticket 222
        we can say we will pass the comment on
            to the authors of the formal semantics

8. eq? and eqv? should apply to all standardized objects (5 minutes)
   - ticket 155
   - Will believes the draft response is incorrect
        but Will has no problem with the recently revised response

9. asymmetry between fold-left and fold-right (5 minutes)
   - ticket 170
   - Will says procedure arguments should come first
        Anton will draft another response and we'll discuss by email
        when we came back to this later, we decided to keep the
            current order of arguments, rejecting the formal comment

10. rename named let? (5 minutes)
   - ticket 165
        Anton will reword with "a majority of the editors" and
        "<bindings>" for "bindings"
            yes: Will, Kent, Matthew, Mike, Anton

11. Syntactic datums and datum values (5 minutes)
   - ticket 209
        The first line of the current draft response is adequate.
        The rest of the current draft response should be deleted.

12. get-bytevector-some may be difficult to implement (5 minutes)
   - ticket 224
        response will be revised to agree with the second fix,
        substituting "bytes" for "characters"

13. Raising of exceptions should not be ambiguous or confusing (5 minutes)
   - ticket 176
   - Will says the response defends early arity checks
        Will wouldn't have a problem with the current response
        if the formal comment had not been truncated

We'll plan to meet again next week (18 April) in case we
haven't approved all of the responses to draft comments.

Discussion of unapproved draft responses to various tickets.
170: keep argument order consistent with Haskell etc;
    point out that SRFI-1 doesn't use the name fold-left
151: should refer to response to 212
152: current draft response no longer explains our position(s)

Will volunteers to draft response regarding text slices etc
Kent will draft responses for the other two recent formal comments

The response to formal comment 223 should mention our recent
approval of bytevector <-> string conversions.

14. adjourned about 9:45am


More information about the R6RS mailing list