#3120 freel not checking arguments

[Leo Butler]

Reported on mailing list

(%i1) sign(rat([2,3]));

Maxima encountered a Lisp error:

 The value 1 is not of type LIST.

Automatically continuing.
To enable the Lisp debugger set *debugger-hook* to nil.

This patches freel

(defmfun freel (l var)
  (cond ((and (listp l) (listp (cdr l)))
     (do ((l l (cdr l))) ((null l) t)
       (cond ((not (free (car l) var)) (return nil)))))
    (t
     t)))

Which gives

(%i2) sign(rat([2,3]));
(%o2)                                 pnz

[Leo Butler]

Patch applied in commit 22f4fd0.

[Dan Gildea]

This commit causes two failures in the test suite:

    taylor(rat(x*(x - 1)), x, 0, 5)
    taylor(rat(factor(1 + x)), x, 0, 1)

[Dan Gildea]

Fixed by [691cbc]

[Robert Dodier]

Reopening this report -- commit [22f4fd0] appears to break tests 23 and 129 in rtest_taylor by changing the return value of RATDISREP in some cases. RATDISREP is supposed to return a simple MPLUS expression but with commit 22f4, it returns an MRAT expression.

For the record the tests are:

taylor(rat(x*(x-1)),x,0,5);
x^2 - x$

taylor(rat(factor(1+x)),x,0,1);
1+x$

[Leo Butler]

Commit 691cbc resolved those testsuite errors.

FWIW, I believe that the exit condition in that commit when l is an atom should punt to free, rather than equal. E.g.

~~~~
defmfun freel (l var)
(do ((l l (cdr l))) ((atom l) (or (null l) (free l var)))
(unless (free (car l) var) (return nil))))
~~~~

[Dan Gildea]

You are right, it should call free. Fixed in [09d160].
I have applied the fix to the 5.38 branch.