Re: Issue: ALIST-NIL (Version 4)

[masinter.pa@Xerox.COM]

Kent and I spoke on the phone. The motivation for this proposal is that the
idea of NIL in an alist is confusing.

One possibility we discussed was to document that an "alist" is in fact a
list of pairs, but to document that the functions ASSOC, RASSOC, ASSOC-IF,
ASSOC-IF-NOT RASSOC-IF and RASSOC-IF-NOT will accept not only an alist but
any list whose elements are either pairs or NIL, and that NIL elements are
ignored. This means that the concept of "alist" is kept to the traditional
interpretation, but that the behavior of some of the functions are
extended.

In general we should try very hard to find ways to "simplify the language"
by changing its description, before we resort to changing its definition.