TITLE: Float versus double

RESPONSE: leni@kwi.com (Leni Mayo), 2 Sep 93

Section 2.5.3 of the ARM states "The type of a floating constant is
double unless explicitly specified by a suffix."

This is relevant when writing overloaded functions having floating
point arguments.  Here's a code sample where the compilation error was
a surprise:

#include <iostream.h>

void print(short s) { cout << "s: " << s << endl; }
void print(float f) { cout << "f: " << f << endl; }

main()
{
    print(0.8);
}

"float.cxx", line 8: error: ambiguous call: print ( double )
"float.cxx", line 8: choice of print()s:
"float.cxx", line 8: print(short );
"float.cxx", line 8: print(float );

The error is explained in section 13.2 of the ARM, particularly pages 322-3.

I infer a rule-of-thumb: "Use double arguments unless there is
a particular reason to use float".