Issue: SETF-METHOD-FOR-SYMBOLS (version 2)

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

    Date: 23 Oct 87 15:22 PDT
    From: Masinter.pa@Xerox.COM

    I think I tend to agree with Scott that I'd
    rather see a little more shuffling in GETF and LDB and less in SETF of
    symbols.

It seems to me that the issue is whether we believe what CLtL says on
p.105 or what it says on p.106.  They contradict each other, because if
you combine that setf method for LDB with that setf method for symbols,
you get a Common Lisp that doesn't work.

I claim that user programs are much more likely to have copied the setf
method for LDB given on p.106 than the setf method for symbols
schematized on p.105, and therefore if we disbelieve p.106 the change is
more incompatible.  I withdraw the mistaken claim I made originally
that it is impossible to make a valid Common Lisp while believing
p.105.

    Hopefully, SETF-FUNCTION-VS-MACRO would reduce the amount of
    hair that was involved for users of setf, anyway. 

In general yes, but not in the cases at issue here; SETF of GETF
and of LDB cannot be implemented as setf functions, for reasons that
are obvious as soon as you think about it.

I might have further comments later, but I thought I'd fire this one
off now since I'm on my way home.