REMF-DESTRUCTION-UNSPECIFIED (Version 2)

["Scott E. Fahlman" <Fahlman@C.CS.CMU.EDU>]

I've wavered back and forth a few times on this in last hour, but I seem
to be converging toward MAKE-EXPLICITLY-VAGUE.

I argued before that it was treacherous, if not clearly illegal, for
implementations to play around with list cells in unexpected ways.  Most
users will have a naive model of what is going on, more or less
equivalent to what is described in MAKE-EXPLICITLY-DEFINED, and it is
likely to screw these users in very subtle ways to go in and manipulate
the cells in some toher way that happens to make CDR-coding happy.  But
if we give the users a sufficiently clear warning that they can't count
on the naive model, then I think it's acceptable to allow implementors
this freedom.

However, we should only go with MAKE-EXPLICITLY-VAGUE if the CDR-coding
people really think it is going to make a significant difference in the
efficiency of some real programs.  I tend to doubt this, but I haven't
played much with CDR-coding and I'll yield to those with more
experience.  If the added freedom only is going to make a tiny
difference, we should go with MAKE-EXPLICITLY-DEFINED, and give the
poor users one less confusing trap to fall into.

-- Scott