Re: g++4.0 diagnostic on casting a static const member initializer to a double

[Jonathan Wakely <cow at compsoc dot man dot ac dot uk>]

On Mon, Jan 31, 2005 at 10:16:03AM -0800, Mark Mitchell wrote:

> Fariborz Jahanian wrote:
> >g++ 4.0 in FSF issues a diagnostic on the following test case:
> >
> >template <class T> struct S {
> >  static const double scd = (double)24;
> >};
> >
> >test1.C:2: error: a cast to a type other than an integral or enumeration 
> >type cannot appear in a constant-expression
> >
> >Rewriting the above to:
> >
> >template <class T> struct S {
> >  static const double scd;
> >};
> >
> >template <class T>
> >const double S<T>::scd = (double)24;
> >
> >Compiles. Event though the diagnostic is clear in what it is saying, but 
> >it seems to be a g++ bug
> >one way or the other.
> 
> No, the input program is invalid.  From [class.static.data]:
> 
> "If  a  static  data  member  is of const integral or const enumeration 
> type, its declaration in the class definition can specify a 
> constant-initializer   which   shall   be   an   integral  constant 
> expression(_expr.const_)."
> 
> The diagnostic could be better, but it is indeed accurate; if there's an 
> initializer it must be integral constant-expression, and a cast to 
> "double" cannot appear in such an expression.

There seems to be a (possibly undocumented) GCC extension involved here.

[class.static.data] should also prevent this compiling, but unless you
use -pedantic both 3.4 and 4.0 allow it:

    struct S {
        static const double scd = 24.0;
    };

I didn't realise this worked until seeing the diagnostic quoted above.
I thought that used to fail with 3.4, was I wrong?

jon


-- 
"A well-written program is its own heaven
 A poorly written program is its own hell"
	- The Tao of Programming