%description:
Test double-to-long conversions:
- fraction part gets truncated

The rest is inconsistent:
- when converting a negative double to unsigned long, the result may be
  (MAX_ULONG-x) or zero (newer gcc [4.0>=?] yields zero)
- if doesn't fit into a long, result is MAX_LONG (gcc), or something else (msvc)

%activity:

#define PRINT(x) {d=(double)x; l=(long)d; ul=(unsigned long)d; EV << "d=" << d << ", l=" << l << ", ul=" << ul << endl;}

double d;
long l;
unsigned long ul;

PRINT(3.01)
PRINT(3.99)

PRINT(-3.01)
PRINT(-3.99)

/// overflows seemingly produce inconsistent results on linux/gcc and msvc :(
// PRINT(1e100)
// PRINT(-1e100

%contains-regex: stdout
d=3.01, l=3, ul=3
d=3.99, l=3, ul=3
d=-3.01, l=-3, ul=(0|4294967293|18446744073709551613)
d=-3.99, l=-3, ul=(0|4294967293|18446744073709551613)