Issue: REMF-DESTRUCTION-UNSPECIFIED (Version 5)

[David A. Moon <Moon@STONY-BROOK.SCRC.Symbolics.COM>]

    Date: Thu, 16 Mar 89 17:22 EST
    From: Barry Margolin <barmar@Think.COM>

	Date: Thu, 16 Mar 89 16:11 EST
	From: David A. Moon <Moon@stony-brook.scrc.symbolics.com>

	I don't have the mail message you referred to ("I suggest that the text
	Amemdment I from my 14 February mail be used verbatim as the proposal.").
	However, I like the way you dealt with NCONC in version 5, I don't
	see any need for a separate proposal.

	Generally this looks okay.  I thought you were going to remove this:

	 (NSUBSTITUTE new-object old-object sequence ...)
	 (NSUBSTITUTE-IF new-object test sequence ...)
	  when sequence is a list, is permitted to SETF any part, CAR or
	   CDR, of the top-level list structure in that sequence.
	  when sequence is an array is permitted to SETF the contents of
	   any cell in that array which must be replaced by NEW-OBJECT.

	since in fact there is no need to give NSUBSTITUTE the freedom
	to modify anything more than what it is required to modify.  I
	still think it should be removed, just as NSUBST was removed.

    I think I remember my reason (it's been a month since I wrote the
    amendment).  I removed NSUBST because CLtL was already very specific
    about what it does.  CLtL's description of NSUBSTITUTE isn't very
    specific; removing it from the proposal would just leave it implicitly
    vague instead of making it explicitly vague, and I'd prefer to be
    explicit.

I see.

    I would support a version that makes NSUBSTITUTE* be explicitly
    specific.  I see no reason not to require it to use SETF of CAR or AREF,
    as appropriate.  Unless someone is working on CAR-coding, I don't think
    it precludes any known optimizations.

I'd support that too.  In other words, describe NSUBSTITUTE with language
similar to the description of NSUBST.