[R6RS] my notes on today's conference call (5 September 2006)

[William D Clinger]

Conference call September 5 2006 8:00am-10:00am EDT
All present by about 8:02am EDT:
Kent, Anton, Matthew, Mike, Will

Thanks to Kent and Mike for all the work they've done...
so far...

0. finalize agenda (1 minute)

1. action items from 8/29/2006 (5 minutes)
   - publish the not-for-public-release r6rs
        done

2. termination language for identifiers, characters, etc. (5 minutes)
   - are termination requirements the same for all of these?
   - who will fix the language?
        Second paragraph of 3.2.1 should become something like:
        Lexemes that require implicit termination (identifiers,
        numbers, characters, booleans, and dot) are terminated
        by any <delimiter> or by end of file, but not necessarily
        by anything else.

3. export "for" syntax" (10 minutes)
   - there are different interpretations about the semantics and
     disagreement about the proper interpretation mong current
     designers/implementors (Flatt, Dybvig, van Tonder, Ghuloum)
   - the export "for" syntax turns out also to be essentially redundant
   - recommendation: flush export "for" syntax and state that
     exported bindings within a library are exported for run, and
     reexported identifiers are exported into their import phases
        Matthew so moved; Kent seconded.
        Vote:
            yes: Kent, Matthew
            no:
            abstain: Will, Anton, Mike
        Motion passed.
    At what phases should the standard libraries be exported?
        Matthew suggested that (r6rs base) export everything
            for run only, and that (r6rs) export everything
            for both run and for export.
    Why is eval in the (r6rs) library?
        Will moved that eval be removed from (r6rs) library;
        Kent seconded.  Motion carried by unanimous vote.

4. script syntax
   - ignore first line iff it starts with "#!<space>" or "#!/"?
   - ignore first line regardless?
        An R6RS script has <blah> as its first line.
        Implementations are required to ignore the first line
            of a script, even if it is not <blah>.
        No objections.

5. perceived r6rs size (10 minutes)
   - should we make use of headers, page breaks, Part numbers, etc. to
     make clear distinction between base and libraries?
   - should we add a rationale for the size and explain that the base
     is approximately the size of r5rs?
   - should we mention that r6rs is the first Scheme standard that a
     reasonable Scheme implementor might choose *not* to extend?
        No.

6. six-month public comment process (15 minutes)
   - expectations for editor participation
        editors are expected to participate
        we don't have to present a unified front
        editors may make public comments
        editors will also deliberate in private
   - tracking public comments
   - tracking our own deliberations
   - trac wiki?

Mitch suggested we set up two mailing lists.
    Anyone could submit to one, but only Kent could read it.
    Kent would verify that it is sent in the proper form.
    Then Kent would resubmit the comment to the second list,
    where everyone could participate.
    Kent didn't want to do that task manually.
Kent suggested they delay public comments for a week
    so we can put together a web form that would enforce
    the correct form.  Those comments would be sent to
    Kent.  The editors would be obliged to respond to
    those comments, but not to comments sent to the
    second list.
Kent volunteered to do the web form.
Will volunteered to set up a Trac system for the editors.
    Only the editors will have access.
Anton volunteered to look into other systems that might
    have a more automated submission process.
Mailing list will probably be at r6rs.org.
    The editors should not be managing that mailing list.

7. publishing discussion archives
   - where and when
        Kent feels the archive should not be published
            until after the R6RS is accepted.
        Let's think about this for a week.

8. publishing reference implementations
   - where and when
        when: some (possibly preliminary) version by next Wednesday
            except as noted below
        libraries and syntax-case
            Kent, maybe another from van Tonder
        arithmetic
            Will
        records
            Matthew
        Unicode library
            Will
        Unicode reader (as part of i/o, probably by next Wednesday)
            Mike
        hash tables (maybe by next Wednesday)
            Anton
        exceptions and conditions
            Mike
        list utilities (don't forget tortoise and hare)
            Anton
        enumerations
            Will
        bytes
            Will
        i/o (in October)
            Mike

Meet again next Tuesday, same time.
The week after that is ICFP.

9. adjourned about 9:45am EDT

Will