Compare fixnums and floats accurately in value<

Make `value<` compare fixnums and floats by value, as `<` does, instead
of coercing the fixnum to a float first which is what C would do.
This matters when the fixnum cannot be represented as a float. For
example, C would evaluate

  72057594037927935 < 72057594037927936.0

to false since the operands are converted to the same floating-point
number.

* src/fns.c (fixnum_float_cmp): New.
(value_cmp): Use it.
* test/src/fns-tests.el (fns-value<-ordered, fns-value<-unordered):
New test cases.

1 files changed, 19 insertions, 2 deletions

diff --git a/src/fns.c b/src/fns.c
index 45567fc183a..61270f0a59f 100644
--- a/src/fns.c
+++ b/src/fns.c
@@ -3012,6 +3012,23 @@ bool_vector_cmp (Lisp_Object a, Lisp_Object b)
   return (d & aw) ? 1 : -1;
 }
 
+/* Return -1 if a<b, 1 if a>b, 0 if a=b or if b is NaN.  */
+static inline int
+fixnum_float_cmp (EMACS_INT a, double b)
+{
+  double fa = (double)a;
+  if (fa == b)
+    {
+      /* This doesn't mean that a=b because the conversion may have
+         rounded, but b must be an integer that fits in an EMACS_INT
+         and we can compare in the integer domain instead.  */
+      EMACS_INT ib = b;         /* lossless conversion */
+      return a < ib ? -1 : a > ib;
+    }
+  else
+    return fa < b ? -1 : fa > b;   /* return 0 if b is NaN */
+}
+
 /* Return -1, 0 or 1 to indicate whether a<b, a=b or a>b in the sense of value<.
    In particular 0 does not mean equality in the sense of Fequal, only
    that the arguments cannot be ordered yet they can be compared (same
@@ -3035,7 +3052,7 @@ value_cmp (Lisp_Object a, Lisp_Object b, int maxdepth)
         if (FIXNUMP (b))
           return ia < XFIXNUM (b) ? -1 : 1;   /* we know that a≠b */
         if (FLOATP (b))
-          return ia < XFLOAT_DATA (b) ? -1 : ia > XFLOAT_DATA (b);
+          return fixnum_float_cmp (ia, XFLOAT_DATA (b));
         if (BIGNUMP (b))
           return -mpz_sgn (*xbignum_val (b));
       }
@@ -3176,7 +3193,7 @@ value_cmp (Lisp_Object a, Lisp_Object b, int maxdepth)
         if (FLOATP (b))
           return fa < XFLOAT_DATA (b) ? -1 : fa > XFLOAT_DATA (b);
         if (FIXNUMP (b))
-          return fa < XFIXNUM (b) ? -1 : fa > XFIXNUM (b);
+          return -fixnum_float_cmp (XFIXNUM (b), fa);
         if (BIGNUMP (b))
           {
             if (isnan (fa))