Rewrite Py_ARITHMETIC_RIGHT_SHIFT so that it's valid for all signed

integer types T, not just those for which "unsigned T" is legal.
This commit is contained in:
Mark Dickinson 2009-03-20 23:16:14 +00:00
parent 2cef1a5409
commit f35f8044e4
1 changed files with 11 additions and 7 deletions

View File

@ -378,19 +378,23 @@ extern "C" {
* C doesn't define whether a right-shift of a signed integer sign-extends
* or zero-fills. Here a macro to force sign extension:
* Py_ARITHMETIC_RIGHT_SHIFT(TYPE, I, J)
* Return I >> J, forcing sign extension.
* Return I >> J, forcing sign extension. Arithmetically, return the
* floor of I/2**J.
* Requirements:
* I is of basic signed type TYPE (char, short, int, long, or long long).
* TYPE is one of char, short, int, long, or long long, although long long
* must not be used except on platforms that support it.
* J is an integer >= 0 and strictly less than the number of bits in TYPE
* (because C doesn't define what happens for J outside that range either).
* I should have signed integer type. In the terminology of C99, this can
* be either one of the five standard signed integer types (signed char,
* short, int, long, long long) or an extended signed integer type.
* J is an integer >= 0 and strictly less than the number of bits in the
* type of I (because C doesn't define what happens for J outside that
* range either).
* TYPE used to specify the type of I, but is now ignored. It's been left
* in for backwards compatibility with versions <= 2.6 or 3.0.
* Caution:
* I may be evaluated more than once.
*/
#ifdef SIGNED_RIGHT_SHIFT_ZERO_FILLS
#define Py_ARITHMETIC_RIGHT_SHIFT(TYPE, I, J) \
((I) < 0 ? ~((~(unsigned TYPE)(I)) >> (J)) : (I) >> (J))
((I) < 0 ? -1-((-1-(I)) >> (J)) : (I) >> (J))
#else
#define Py_ARITHMETIC_RIGHT_SHIFT(TYPE, I, J) ((I) >> (J))
#endif