When <math.h> gets included abs is overloaded for type double and so abs(toffset*2)&0x1f attempts to do bitwise AND on a double, which is ill-formed. This causes a build failure with GCC 6, which now provides a conforming <math.h> header. The fix is to ensure abs(int) is used instead.
When
<math.h>
gets includedabs
is overloaded for typedouble
and soabs(toffset*2)&0x1f
attempts to do bitwise AND on adouble
, which is ill-formed. This causes a build failure with GCC 6, which now provides a conforming<math.h>
header. The fix is to ensureabs(int)
is used instead.