Re: access closure variable

In article
<4546207c-336a-4ab2-95c3-a1d8a46aaa09@xxxxxxxxxxxxxxxxxxxxxxxxxxx>,
Mirko.Vukovic@xxxxxxxxx wrote:

On Jul 8, 3:58 pm, Rainer Joswig <jos...@xxxxxxx> wrote:
In article
<73264c06-daa6-40ac-b480-a48cbbd24...@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,



Mirko.Vuko...@xxxxxxxxx wrote:
On Jul 7, 3:06 am, Rainer Joswig <jos...@xxxxxxx> wrote:
In article
<472fab77-164b-466b-9f5a-92b4cb845...@xxxxxxxxxxxxxxxxxxxxxxxxxxxx>,

gtasso <geta...@xxxxxxxxx> wrote:
Hello all,

consider this..

(setf foo '"hello world")

I could access the the value this way
(symbol-value'foo)

wonder if i could do something like this with variable setup in a
closure like this.

(let ((foo "hello world)
(bar "hello c.lang.lisp"))

(defun get-value (lst)
(dolist (e lst)
(print (symbol-valuee))))

and from outside let i issue

(get-value '(foo bar)) - > i get error variable unbound

Because you can't accesslexicalvariable values
withSYMBOL-VALUE.

I could write individual function for each variables in the closure
but I am looking for shorter way.

CL-USER 9 > (mapcar (let ((a 1) (b 2))
(flet ((get-value (sym)
(ecase sym
(a a)
(b b))))
#'get-value))
'(b a))
(2 1)

I could write the whole thing as an object and benefit from the
accessor and slot-value function but i guess i wanted it done with
closure instead :).

Many thanks.

regards,

George.

--http://lispm.dyndns.org/

This group is positively possessed by super-unnatural forces. So
often it happens that as I type a question, the answer just
materializes to me. And now, my question was answered several days
before it became even actual!

So, here is a macro with which I tried to automate the OP's quest for
automation. I am trying to re-learn to program without objects.

(in-package :my-utils)
(defmacro with-visible-lexs (visible-lexs &optional init-block &body
cases)
"Return function (case-based) with closure & with utility to view
lexical values"
`(let ,visible-lexs
,(when init-block init-block)
(lambda (command &optional args)
(case command
(prop
(ecase args
,@(loop for lex in visible-lexs
collect `(,lex ,lex))))
,@cases))))

Here is a sample session:

MY-UTILS-UNIT-TESTS> (setf *a* (foo 1 3))
#<CLOSURE (SB-C::&OPTIONAL-DISPATCH (LAMBDA #)) {100366F169}>
MY-UTILS-UNIT-TESTS> (apply *a* (list 'sum))
8
MY-UTILS-UNIT-TESTS> (apply *a* (list 'mult))
12
MY-UTILS-UNIT-TESTS> (apply *a* (list 'my-utils::prop 'c))
4

One thing I don't know how to deal with is that the macro uses the
symbol my-utils::prop, since that is where it sits. I don't know how
to go around that.

If anyone has a suggestion for a better name(s), by all means.
Inevitable improvements & suggestions, also (I have not really thought
out if it needs gensym - args & lex may need to be).

Mirko

Thanks Rainer,


Flavors uses

What is Flavors? Could not find it on cliki or the web.

http://en.wikipedia.org/wiki/Flavors_%28programming_language%29

Flavors was/is an early extension to Lisp for object-oriented programming.

ftp://publications.ai.mit.edu/ai-publications/pdf/AIM-602.pdf
http://www.softwarepreservation.org/projects/LISP/rutgers/clisp/flavors.pdf

It was oriented towards 'message passing'. Later developed into New Flavors,
which had some influence in the design of CLOS.


(send some-window :set-edges 10 10 40 40)

So, use keyword symbols. Use &rest args instead of optional args.

Then you have

(funcall *a* :prop 'c)

Cool. I keep forgetting about keywords and their package independence
(not cool).

Keywords are not really package independent. Keywords are in the package
"KEYWORD".

CL-USER 1 > (symbol-package :foo)
#<The KEYWORD package, 0/4 internal, 5302/8192 external>

CL-USER 2 > KEYWORD::FOO
:FOO

CL-USER 3 > KEYWORD:FOO
:FOO

CL-USER 4 > :FOO
:FOO

CL-USER 5 > ::FOO
:FOO

The reader will read :foo as a symbol in the keyword package.
The package is missing in front of the : and the symbol, so it will
be the KEYWORD package by default.


In good old style replace funcall with send:

(send *a* :prop 'c)

;-)

--http://lispm.dyndns.org/

"send"? I do not find it in hyperspec. Do you mean as an alias for
funcall?

Yes.


Thanks,

Mirko

--
http://lispm.dyndns.org/
.

Relevant Pages

  • Re: A Question about DEFPACKAGE syntax
    ... Lisp and for not-really-Common-Lisp ... that it's easier to type:foo than to type "FOO". ... "FOO" does not create superficial symbols in the keyword package ...
    (comp.lang.lisp)
  • Re: A Question about DEFPACKAGE syntax
    ... you give is a reason NOT to use strings. ... "FOO" does not create superficial symbols in the keyword package ... not creating those extra few symbols in the KEYWORD package is ...
    (comp.lang.lisp)
  • Re: Problem comparing symbols
    ... Since you have used three symbols not in the CL package ... CL-USER 1> ... symbols that you use as tags go in the keyword package. ... CL-USER> (foo:hello 'greetings) ...
    (comp.lang.lisp)
  • Re: whats the difference between : and :: ?
    ... Expanding a bit on the keyword part of your question: ... Keywords are all in a package named KEYWORD. ... a symbol:FOO as a symbol with the name FOO in the package KEYWORD. ...
    (comp.lang.lisp)
  • Re: from __future__ import absolute_import ?
    ... foo not in bar ... A path below the package level is generally a good means to shoot ... to represent a python "package" structure. ...
    (comp.lang.python)