Things used in S&ICP

[Jonathan A Rees <JAR@MC.LCS.MIT.EDU>]

    Date: Mon 24 Feb 86 22:22:37-EST
    From: Gerald Jay Sussman <GJS%OZ.AI.MIT.EDU at XX.LCS.MIT.EDU>

    I feel strongly that we should have >, <, =, because I used them in
    S&I.  I see no reasons why we should not also have other people's
    favorites.

Speaking of what's in S&I, there certainly are many things used in the
book that AREN'T described in the RRRS.  Namely, these special forms:

	COLLECT
	CONS-STREAM
	DELAY
	DEFINE-MACHINE
	MAKE-ENVIRONMENT

and these bindings:

	ATOM?
	EMPTY-STREAM?
	ERROR
	EVAL
	EXPLODE
	FORCE
	GET
	HEAD
	MAPC
	IMPLODE
	MAPCAR
	PRINC
	PRINT
	PUT
	TAIL
	THE-EMPTY-STREAM
	USER-INITIAL-ENVIRONMENT

I would like to hear a principled argument for including or not
including these various things in the RRRS.  (Not a separate argument
for each one, please!)  I don't much care what position is taken so long
as it's pretty much consistent.  The current RRRS is not consistent,
since it includes SEQUENCE (non-essentially) only because it's in S&I,
but doesn't even include the trivial but important special forms DELAY
and CONS-STREAM.

Suggestion: document them all, each one very briefly, in an appendix, as
non-essential features.