Is ice-9/poe.scm broken?

Is ice-9/poe.scm broken?

[Alan Grover]

[-- Attachment #1.1: Type: text/plain, Size: 956 bytes --]

When I tried to use poe to cache some functions, I got a run-time error.
When I examined poe.scm to understand what I did wrong, I found code
that doesn't make sense under guile. Can anyone confirm that poe is
outdated, or I have a completely wrong idea of how to use it?

guile: 1.6.4

Here's how I tried to use it:

	(use-modules (ice-9 poe))
	(define (some-fn a) (list 'did-something a)) ; trivial function
	(define cached-fn (pure-funcq some-fn))

	(some-fn 'an-arg)
	(cached-fn 'an-arg) ; first time through (dies in here)
	(cached-fn 'an-arg) ; second time through

In poe.scm, it seems to be using the old idiom of '() being treated as
false:

	(or (and (and (not key) (not entry))
	... (eq? (car key) (car entry))

Which I think is supposed to be
	(or (and (and (not (pair? key)) (not (pair? entry)))
	... (eq? (car key) (car entry))

I re-wrote enough of poe.scm to get some things working.

-- 
Alan Grover
awgrover@mail.msen.com
+1.734.476.0969

[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 256 bytes --]

[-- Attachment #2: Type: text/plain, Size: 140 bytes --]

_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: Is ice-9/poe.scm broken?

[Thien-Thi Nguyen]

   From: Alan Grover <awgrover@mail.msen.com>
   Date: Mon, 08 Aug 2005 11:52:10 -0400

   I re-wrote enough of poe.scm to get some things working.

i likewise had to munge poe.scm for Guile 1.4.x.  see:

  http://www.glug.org/people/ttn/software/guile/
  http://www.glug.org/docbits/guile/1.4.x/Memoization.html

thi


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: Is ice-9/poe.scm broken?

[Kevin Ryde]

Alan Grover <awgrover@mail.msen.com> writes:
>
> I re-wrote enough of poe.scm to get some things working.

In the cvs head a while ago I changed funcq-assoc to the code below.
Dunno why I didn't do the same in 1.6 (maybe the hashx stuff has
problems too).  You're right that it doesn't work.


;; return true if lists X and Y are the same length and each element
;; is `eq?'
(define (eq?-list x y)
  (if (null? x)
      (null? y)
      (and (not (null? y))
           (eq? (car x) (car y))
           (eq?-list (cdr x) (cdr y)))))

(define (funcq-assoc arg-list alist)
  (if (null? alist)
      #f
      (if (eq?-list arg-list (caar alist))
          (car alist)
          (funcq-assoc arg-list (cdr alist)))))


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: Is ice-9/poe.scm broken?

[Marius Vollmer]

Kevin Ryde <user42@zip.com.au> writes:

> In the cvs head a while ago I changed funcq-assoc to the code below.
> Dunno why I didn't do the same in 1.6 (maybe the hashx stuff has
> problems too).  You're right that it doesn't work.

Could you try to fix this in 1.6 as well?

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: Is ice-9/poe.scm broken?

[Kevin Ryde]

Marius Vollmer <mvo@zagadka.de> writes:
>
> Could you try to fix this in 1.6 as well?

I put it in, I think it works.

On the merely dodgy as opposed to broken front, I suspect the hash
function should be using logxor, and that perfect-funcq doesn't need
to work the base-func into the hash, since the hash is private the
each func.


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user

Re: Is ice-9/poe.scm broken?

[Marius Vollmer]

Kevin Ryde <user42@zip.com.au> writes:

> Marius Vollmer <mvo@zagadka.de> writes:
>>
>> Could you try to fix this in 1.6 as well?
>
> I put it in, I think it works.

Thanks!

> On the merely dodgy as opposed to broken front, I suspect the hash
> function should be using logxor, and that perfect-funcq doesn't need
> to work the base-func into the hash, since the hash is private the
> each func.

Hmm, I don't really understand what you are saying here, but if you
feel strongly about changing it, go ahead!  (In HEAD :).

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405


_______________________________________________
Guile-user mailing list
Guile-user@gnu.org
http://lists.gnu.org/mailman/listinfo/guile-user