From ...
From: Erik Naggum <erik@naggum.no>
Subject: Re: setf implementation
Date: 2000/01/24
Message-ID: <3157726372992359@naggum.no>#1/1
X-Deja-AN: 577002782
References: <86i2o6$o2t$1@monet.op.net>
mail-copies-to: never
Content-Type: text/plain; charset=us-ascii
X-Complaints-To: newsmaster@eunet.no
X-Trace: oslo-nntp.eunet.no 948737811 20273 195.0.192.66 (24 Jan 2000 18:16:51 GMT)
Organization: Naggum Software; +47 8800 8879 or +1 510 435 8604; fax: +47 2210 9077; http://www.naggum.no
User-Agent: Gnus/5.0802 (Gnus v5.8.2) Emacs/20.5
Mime-Version: 1.0
NNTP-Posting-Date: 24 Jan 2000 18:16:51 GMT
Newsgroups: comp.lang.lisp

* mjd@op.net (Mark-Jason Dominus)
| I was surprised by this, because I had picked up a different idea from
| somewhere.

  it's a notion from C's pointer concept.  as such, it's pretty bogus.

| When you evaluate (cdr ...), you get back a pointer to the cdr, and
| then the enclosing function receives this pointer as an argument.

  that's precisely what you don't do in Common Lisp.  you don't get a
  pointer to the CDR, you get the value of the CDR, which may or may not be
  a pointer, depending on its type.  in any case, you're not getting a
  pointer to something settable.

  consider this schematic (yikes, I'm having to resort to graphics!):

        +-------+-------+
    X:  |  CAR  |  CDR  |--> ZOT
        +-------+-------+
            |
            V
           FOO

  when you evaluate (CDR X), you get ZOT back.  if you wish the CDR of X to
  point to something else, you would do (setf (cdr x) 'bar), but if you
  have already obtained ZOT, you don't know where to store BAR.

  therefore, in C terminology, what SETF gives you is the address of the
  slot you wish to change, but that is not at all similar to a pointer to
  the value _in_ that slot.

  therefore, your approach won't work, since it misses out on the crucial
  level of indirection.

#:Erik