[R6RS] my notes on today's conference call (12 April 2006)

[William D Clinger]

Conference call April 11 2006 12:30pm
All present by about 12:30pm:
Kent, Anton, Mike, Will, Matthew

0. finalize agenda (1 minute)

1. action items from 4/05/2006 (5 minutes)
   - carried over:
     - look over proposed condition hierarchy and details; comment
       via mailing list (All)
        still pending
     - complete proposed list of libraries (Anton)
        done
     - propose library modifications to allow subsetting, supersetting,
       defining new libraries, and to allow for lexical syntax extensions
       or future changes (Anton, Matthew, Kent)
        still pending

2. plan for completing r6rs document draft (15 minutes)
    Mike sent out a note yesterday
    Would like to pick 2 or 3 new things, already in SRFI form,
        finish them, and get started on integrating them into the
        document.  Main candidates are:
            Unicode
            arithmetic
            records
        Try to get this done by end of month.
    Unicode going out in final form.
    Arithmetic:
        requiring infinities, NaNs
        changing fx+ to throw exception on overflow,
            adding fixnum+ that wraps
        choosing names for div/mod etc
    Exceptions and conditions
    Records
        post message to mailing list asking people to say whatever
            they're still going to say
    Syntax-case: is there still time to put it out as a SRFI?
        could have a syntax-case section in R6RS by mid-June
    Mike wants to postpone semantics until after mid-June

3. core/library split (15 minutes)
    general happiness with Anton's proposal and Mike's comments

4. eval and load (10 minutes)
    environment could become library
    What about load?
    Let's keep eval and drop load.
        Implementations can still support whatever version of
        load they like.
    Are we going to have a top level?
    Anton uncomfortable with dropping top level and load.
    "Let's let Mike chew on that."

5. local define-syntax (10 minutes)
    Mike questions whether the added complexity (of both
        implementation and specification) is worth the
        added convenience.  Suggests we hold off on this.
    Interleaving local define-syntax with define is problem.
    Kent points out that expansion order already matters,
        both locally and at library level.  If we don't
        allow local define-syntax, then macros that expand
        into macros will have to come in two versions: one
        for top level (of a library) and another for bodies
        of expressions.  Restricting to top level (e.g. import)
        is a bad idea; not very Schemely.
    Not allowing local define-syntax in lambda bodies creates
        artificial difference between library bodies and
        lambda bodies.

Next week:
Redo SRFI on bytes?  No.
Withdraw i/o SRFIs?  None of our business; they aren't R6RS SRFIs.
  
6. adjourned about 1:38pm