[Ecls-list] Problem with MOST-POSITIVE-DOUBLE-FLOAT and MSVC compiler

Goffioul Michael goffioul at imec.be
Tue May 3 04:02:58 UTC 2005


I discovered a very annoying problem with most-positive-double-float and the
MSVC compiler. Take a code like:

(let ((uplim (/ most-positive-double-float 5)))
  (defun f3 (x)
    (when (< x uplim)
      (print "ok"))))

The generated C code contains verbatim the value of most-positive-double-float,
which includes 307 digits. This triggers a fatal compiler error in MSVC. A
possible workaround is to use read-evaluation (-> #.(/ most-positive-double-float 5)).
Is there another solution?


More information about the ecl-devel mailing list