Raymond Toy pushed to branch issue-504-read-denormals-with-rounding at cmucl / cmucl

Commits:

• 06ea24af
  by Raymond Toy at 2026-05-26T07:22:37-07:00

  Change declaration of exp arg
  
  In `scale-float-maybe-underflow`, the exponent (`exp`) isn't limited
  to the double-float-exponent range.  It can basically be any fixnum.
  This can be seen with `(scale-float 1d0 -9999)`.
  `scale-float-maybe-underflow` sees -9999, well outside the range of
  double-float-exponent type.
  

1 changed file:

• src/code/float.lisp

Changes:

 ;;;

 (defun scale-float-maybe-underflow (x exp)

   (declare (type (or single-float double-float) x)

-	   (type kernel:double-float-exponent exp))

+	   (fixnum exp))

   (multiple-value-bind (sig old-exp)

       (integer-decode-float x)

     (let* ((digits (float-digits x))