blob: bd15e2d80ec086fbfd8a67fa3023bf2aafc87f97 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
|
#include "util/index.h"
#include <cstddef>
#include <limits>
namespace CVC4 {
static_assert(sizeof(Index) <= sizeof(size_t),
"Index cannot be larger than size_t");
static_assert(!std::numeric_limits<Index>::is_signed,
"Index must be unsigned");
/* Discussion: Why is Index a uint32_t instead of size_t (or uint_fast32_t)?
*
* size_t is a more appropriate choice than uint32_t as the choice is dictated
* by uniqueness in arrays and vectors. These correspond to size_t. However, the
* using size_t with a sizeof == 8 on 64 bit platforms is noticeably slower.
* (Limited testing suggests a ~1/16 of running time.) Interestingly,
* uint_fast32_t also has a sizeof == 8 on x86_64.
*/
}/* CVC4 namespace */
|