On Wed, Oct 23, 2002 at 06:44:47PM -0400, Brian S. Julin wrote:
>
>Hi,
>
>As part of the SWAR implementation I could really use a "portable"
>uint64 typedef. So far as I can tell, the best thing to do in this
>case would be to *only* define uint64 as a type if the compiler can
>do logic operations (and, or, and shifts), math (add sub mul div)
>on a 64-bit type. It's OK if the compiler emulates this behavior,
>but if it doesn't the system is not considered to have a uint64,
>so we would not define it. Thus all code that uses it must sheild
>to only compile when the system supports it.
>
>So would the best course of action be to put a
>#define (or #undef) GG_HAVE_UINT64 in system.h, along with a conditional
>typedef of uint64?
see AC_TYPE_UNSIGNED_LONG_LONG
You then can wrap code whether or not you HAVE_UNSIGNED_LONG_LONG_INT
easily.
>
>Am I thinking straight? Comments?
>
>--
>Brian
