#2386 curious evaluation in ifactor.lisp

[Volker van Nek]

With clisp and sbcl:

(%i1) load("/home/volker/Maxima/git/maxima/src/ifactor.lisp")$

(%i2) :lisp (get-factor-list 15)

((5 1) (3 1))
(%i2) :lisp (let* ((fs (get-factor-list 14)) (f (car fs))) (incf (cadr f) 41))

42
(%i2) :lisp (get-factor-list 15)

((5 42) (3 1))

Each call to get-factor-list should cons a new list, so each return value should be referentially independent.

I tried to find the reason of this error and observed that in ifactor.lisp lines 223-226

 \(when \(> \(\* d d\) n\)
   \(push \`\(,n 1\) factors\)
   \(when $ifactor\_verbose  \(format t "small prime cofactor: ~A~%" n\)\)
   \(return-from get-small-factors \(values 1 factors\)\)\)

the backquote-comma-macro-line seems to cause this.

Using
(push (list n 1) factors)
or
(setq tmp 1)
(push `(,n ,tmp) factors)
instead the results are as expected.

I don't get the clue what's wrong with (push `(,n 1) factors).

If I create a new list foo with (copy-tree factors) inside of get-small-factors and do
(push `(,n 1) foo)
the results for foo are as expected.

So I guess the variable 'factors' has referential problems. Don't know why.

[Volker van Nek]

(setq foo (copy-tree factors))

isn't even necessary.

(setq foo factors)
or
(setf foo factors)
also works:

 \(when \(> \(\* d d\) n\)

(setf foo factors)
(print foo)
(push `(,n 1) foo)
(print foo)
(print factors)
(push `(,n 1) factors)
(print factors)

prints
(%i2) :lisp (get-factor-list 15)

((3 1))
((5 1) (3 1))
((3 1))
((5 42) (3 1)) <- print
((5 42) (3 1)) <- return value

[Robert Dodier]

I think you should just go ahead and change the code. Looks to me like you have identified a problem and a way to fix it, so I think you should ahead and fix it.

[Volker van Nek]

Hi Robert,

I have just checked if this problem is present with gcl on Windows and it's not. But with ccl (Windows agein). So it counts 3 : 1 (first noticed with sbcl and clisp on Linux).

The fact that gcl doesn't show this problem makes me wonder whether it's a Maxima or a Lisp problem.

I will follow your advice, change the code and close this report.

Best
Volker