differences between R4RS and draft IEEE std

[William Clinger <will@fog.cs.uoregon.edu>]

   Date: Fri, 27 Jan 89 14:46:12 PST
   From: Pavel.pa@Xerox.COM

   I note with some misgivings several differences between R4RS and the draft
   IEEE standard such that the standard is not a subset of R4RS....

I predict that the R4RS will adopt almost all of the changes urged upon it by
the people writing the IEEE standard, making the draft IEEE standard very
close to being a subset of the R4RS.  There are only two things in the draft
IEEE standard that might not belong in the R4RS:

  WITH-INPUT-FROM-PORT
  WITH-OUTPUT-TO-PORT

Many people have wanted these procedures, but I do not feel that the authors
have yet approved their inclusion in R4RS.  (These also happen to be the
features of the draft IEEE standard that I am least comfortable with, as
discussed in a separate message.)

I consider that addition of the STRING procedure to R4RS is warranted as a
"regularization of procedures", which the authors indicated they were willing
to consider by electronic mail.

The subcommittee delegated to clean up numbers was specifically charged with
the task of simplifying NUMBER->STRING and STRING->NUMBER.  If that subcommittee
recommends that the R4RS description of these procedures be changed to match
the draft IEEE standard, as I am sure they will, then the change will happen.

The changes to CHAR-UPPER-CASE? and CHAR-LOWER-CASE? are non-controversial and
should fall within the discretion of the editors.

Requiring that CHAR->INTEGER be given an exact non-negative integer is
needed for consistency with R4RS's requirement that INTEGER->CHAR return an
exact non-negative integer.  (Actually the non-negative part was not explicit
when this was approved at Snowbird, but I think that was just an oversight on
my part.  Does anyone disagree?)  I do not feel that the R4RS should drop the
requirement of order isomorphism, but the draft IEEE standard can drop it
without endangering its status as a subset.

Peace, Will
acting editor, R4RS