bug#64508: Optimization of "-" function is incorrect in presence of GOOPS overloading

On Wed, 04 Mar 2026, Olivier Dion <[email protected]> wrote:
> On Wed, 04 Mar 2026, Werner LEMBERG <[email protected]> wrote:
>> Is there any progress in fixing this glitch?
[...]
> To something is happening when expanding: (primcall - (toplevel x))

So the culprit is in languages/tree-il/primitives.scm:

  (define-primitive-expander -
    (x) (- 0 x) ; <- here
    (x y) (- x y)
    (x y z ... last) (- (- x y . z) last))

introduced by cb28c08537790b49f7bc94f2f6b426497152bbe7

If I remove the 0, it seems to be okay and no tests are failing.  I
wonder why Andy decides to put this expansion.

There seems to be similar expansions for a couple of primitive generics:

  (define-primitive-expander zero? (x)
    (= x 0))

  (define-primitive-expander positive? (x)
    (> x 0))

  (define-primitive-expander negative? (x)
    (< x 0))

  (define-primitive-expander /
    (x) (/ 1 x) ; <- here
    (x y z ... last) (/ (/ x y . z) last))

I think it is fine to remove the expander for the `(x)' case for `-' and
`/', but not for the predicates.  Thoughts?

Thanks,
Olivier
-- 
Olivier Dion