Re: Documentation for car and cdr

[Kevin Rodgers]

Peter Whaite wrote:
> Luc Teirlinck <address@hidden> wrote:
> > The only list that is not a cons cell is nil.
>
> Actually I seem to remember learning that nil _was_ a cons cell where
> both the car and cdr pointed to itself.  That makes it unnecessary to
> treat nil as a special case.
>
> But that was a long time ago.

I learned Lisp 20+ years ago, and nil has always been an atom (in fact,
a symbol whose value is itself, just like t and now keywords).  A cons
cell like you describe would be a circular structure:

        +----+   +----+
        |    |   |    |
        |    v   v    |
        |  +---+---+  |
        |  |   |   +--+
        |  +-+-+---+
        |    |
        +----+

(let ((print-circle t)
      (cons-cell (cons nil nil)))
  (setcar cons-cell cons-cell)
  (setcdr cons-cell cons-cell)
  cons-cell) ==>
(#0 . #0)

If the symbol nil had such a value, (eq nil 'nil) would not be true.

--
Kevin Rodgers