[R6RS] my notes on today's conference call (8 March 2006)
William D Clinger
will at ccs.neu.edu
Thu Mar 9 07:25:41 EST 2006
conference call March 8 2006 12:46pm EST
Kent, Anton, Matthew, Mike, Will
0. finalize agenda (1 minute)
1. action items from 3/1/2006 (5 minutes)
- no new action items
- carried over:
- post note re: van Tonder syntax-case differences (Matthew)
- read exception, byte-vector, and I/O SRFIs (Anton, Kent, Matthew, Will)
- draft hash-table proposal (Anton, Will)
- draft syntax-case SRFI (Kent)
2. multiple-value binding construct (5-10 minutes)
- should we consider mvlet / mvlet*?
okay
- if not, are we ready to vote between options #1 and #4?
option 1: mvlet and mvlet* names; no "values" in patterns
option 4: let and let* names; "values" in patterns
Mike and Will like semantics of #4 with mvlet and mvlet* names;
Kent doesn't like that combination, but should let us vote on it.
3. equal? (5-10 minutes)
- should we adopt Will's equiv? semantics for equal?
mike: no
matthew: no
anton: abstain
will: a good thing to leave for R7RS
- should we define equal? on records?
behavior of equal on records was not discussed
4. status reports (5 minutes each, 15 minutes total)
- arithmetic
general agreement on large issues, controversy on details
write R6RS as though all systems support IEEE-754, but
allow systems to raise &implementation-restriction
exceptions when they can't represent an infinity or NaN
fixnums and flonums to be in libraries
- safe/unsafe mode (declarations)
Kent wants some portable, reasonably detailed semantics
Will says a portable syntax is already valuable so you
can express your general priorities without having
to rewrite all declarations when you port to another
system
- exceptions
people who haven't read SRFI 34 and 35 need to do so
Mike will send out summary of proposed changes to those SRFIs
vote next week
5. unicode requirements (10 minutes)
Should we require all of the Unicode stuff from SRFI 76?
Follow example of full numeric tower, but don't be
so apologetic about it. Allow implementations to
raise &implementation-restriction exceptions if
they can't represent a character or string they
need to represent. Mention the ASCII subset as
reasonable, or sanction an ASCII level of R6RS
conformance. If some implementors wants to build
some subset that's in between ASCII and the full
SRFI 76 spec, then let them worry how to do it.
ISO Latin-1 isn't closed under normalization
Items to discuss next week:
split between core and libraries
Anton will post something
io
see action items
6. adjourned 1:28pm EST
Will
More information about the R6RS
mailing list