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

[William D Clinger]

Conference call April 5 2006 12:45pm
All present by about 12:47pm:
Kent, Anton, Matthew, Will, Mike

Matthew will not be available next Wednesday
Tuesday at 12:30pm might work

0. finalize agenda (1 minute)
    no changes

1. action items from 3/29/2006 (5 minutes)
   - read syntax-case SRFI for 4/5 meeting (All)
   - propose library modifications to allow subsetting, supersetting,
     defining new libraries, and to allow for lexical syntax extensions
     or future changes (Anton, Matthew, Kent)
        not done yet
   - propose enumeration mechanism to replace ad-hoc enumerations in
     the I/O, binary SRFIs (Mike)
        proposed
   - propose a better name than ``blob'' for byte-addressible octet vectors
     (Anton)
        bytes
   - complete proposed list of libraries (Anton)
        not done yet
   - carried over:
     - look over proposed condition hierarchy and details (All)
        to be discussed electronically

2. syntax-case macros (20 minutes)
    no reference implementation yet
    opacity of syntax objects
        three reasonable positions:
            opaque, no exceptions, as in PLT Scheme
            list structure except for identifiers, as in SRFI 72
            allow both of the above extremes and everything between
                would probably create portability problems
                but we might have those problems anyway if something
                    like SRFI 72 becomes a popular alternative
    specification needs to be more precise from user's perspective
        matching described using list structure, not syntax objects
        marks mentioned only at end of section 3.1:
            "Hygiene is enforced by attaching a fresh mark to the
                output of the introduced portions of each transformer
                result...."
            one mark per introduced portion?  or same mark to all?
            when syntax is used, or when transformer returns?
            what about recursive transformers, with multiple returns?
            marks never used to describe semantics of anything:
                matching of identifiers (but see literal-identifier=?)
                literal-identifier=?
        bound-identifier=? and free-identifier=? described in terms
            of bound and free identifiers, but bound and free with
            respect to what?
    probably need both:
        a formal semantics for language lawyers and implementors
        an informal description for programmers, who can refer to
            the formal semantics in hard cases or ask a lawyer
            or ask an implementation
    potential interactions with help procedures
        make precise semantics even more important

3. eval: flush from or adapt for r6rs?

for next week:
process and schedule
library split
eval and load

4. adjourned about 1:43pm