Re: Walking a list



"Leslie P. Polzer" <leslie.polzer@xxxxxxx> wrote:
it doesn't let me *save a pointer* to some element in between,
which is the central problem I want to solve in a Lisp way.

Giorgos Keramidas wrote:
There are no 'pointers' in Lisp. Once you get past the terminology gap
which separates the two worlds -- the one of C, and the one of Lisp --
you will realize that any Lisp symbol can refer to arbitrary places
within the list

Dan Bensen <randomgeek@xxxxxxxxxxxxxx> wrote:
Getting past the 'terminology gap' is not necessary. All the C pro-
grammer has to remember is that all Lisp variables are references,
i.e. implicit pointers with no C-style pointer arithmetic allowed.

Giorgos Keramidas wrote:
I don't really like the confusion which may result from treating Lisp ``symbols'' are C ``pointers''.

There's no confusion for C programmers, and it's variables that are
pointers, not symbols. As Pascal said,

A symbol can conceptually be considered as a structure of 5 slots:
(defstruct symbol package name function value plist)

This description makes perfect sense to a C programmer. A variable
can point to a symbol, and the slots in the symbol point to data.

Mostly because then the need arises for clarifications
of the sort you just mentioned (``implicit pointers
with no C-style pointer arithmetic allowed'').

Again, that's not a problem for a C programmer.

Referring to Lisp symbols as ``pointers'' is not exactly correct.

It's not even close to correct, but I think we've covered that now.

There is a perfectly fine way to refer to them, and that is by
using the word ``symbols'', so that's why I mentioned terminology.

That's not familiar to a C programmer. C programmers understand
pointers very well, but not symbols.

Sometimes, getting the wording correct facilitates better understanding

That applies to all of us. Here's what I said:

All the C programmer has to remember is that all Lisp ***variables*** are references


but YMMV since we are human after all, and everyone has their own way of understanding something :)

Yes, that's true, and I understand that what you're saying probably
makes sense to a long-time Lisper who may not know C. The OP isn't
an experienced Lisper though, so using their language is probably
more helpful.

To summarize, in Lisp, you don't need `pointers' in the C sense, because
every symbol is essentially a pointer to a particular value, and an
arbitrary number of symbols can be bound to the same value.

The OP explicitly asked for the "Lisp way" of solving the problem,
which does require implicit pointers in the Lisp sense, and every C
programmer knows that multiple references can point to the same memory
location. So all the OP really needed to know was how to do what they
were describing.

Which I demonstrated with an example of two symbols pointing to parts of
the same Lisp object. I don't see what you found wrong about my post

The "terminology gap" comment ("you will realize") suggested that
(1) the OP didn't know that more than one reference can point to the same object, and (2) the only way to cure that ignorance was to submit
to the Lisp Way (TM).

"At this point, stand back ... You will see ..."

Again, this comment suggests that the reader must Behold the
Magnificence of Lisp to understand. Note, btw, that your little arrows
seem to be *pointing* at the data structures, as you yourself even
mentioned:

"To traverse a list, you just need a symbol to 'point'
somewhere inside the list."

Where, to beat an already-dead horse, "symbol" should read "variable".

I understand that symbols are obviously an integral part of Lisp,
but there really are other ways to describe things that people with
other backgrounds will understand better. To more fully address
this comment at the beginning of your post:

"There are no 'pointers' in Lisp."

Lisp is ALL pointers. EVERY variable, EVERY slot in EVERY object
or structure, is a reference to a piece of data. In Lisp, there's
no such thing as directly embedding one object inside another,
as is done in C and C++. There's no such thing as actually
passing an object to a function. The only thing you can contain in
an object or pass to a function is an implicit pointer to the object. If you want to pass a new copy of the object, you have to explicitly
create the copy, and pass a reference to it.

--
Dan
www.prairienet.org/~dsb/
.



Relevant Pages

  • Re: Paul Graham explains Ruby symbols
    ... >> "Symbols are effectively pointers to strings stored in a hash table. ... Actually it's five in Common Lisp: ... A symbol can belong to a package. ...
    (comp.lang.ruby)
  • Re: Confirmed: never, Ever, EVER use nreverse unless you (I mean I) damn well know what is going on
    ... This is the code snippet: ... Never use NREVERSE unless you know that something /isn't/ going on; ... Lisp variables are, generally, pointers. ...
    (comp.lang.lisp)
  • Re: Fundamentals of Lisp efficiency?
    ... > tcp/ip stack in Lisp or any language, and don't plan on it, but I ... > efficiency, you pass around pointers, which is quite Lisp-friendly (at ... For socket-level programming, it's no big deal. ...
    (comp.lang.lisp)
  • Re: Bootstrapping ANSI CL
    ... This means that you can do tagged pointers without losing a single bit ... that hard to build a next-generation Lisp machine on an amd64 system. ... software to be compatible with the new addressing. ...
    (comp.lang.lisp)
  • Re: Walking a list
    ... | Lisp is ALL pointers. ... In Lisp, there's ... space as a generic Lisp reference without conflict [perhaps by ... many Lisps use such immediates for FIXNUMs, ...
    (comp.lang.lisp)