[libcxx-commits] [libcxx] linear_congruential_engine: add using more precision to prevent overflow (PR #81583)

[via libcxx-commits]

================
@@ -38,12 +38,12 @@ int main(int, char**)
 
     // m might overflow. The overflow is not OK and result will be in bounds
     // so we should use Schrage's algorithm
-    typedef std::linear_congruential_engine<T, (1ull << 32), 0, (1ull << 63) + 1> E2;
+    typedef std::linear_congruential_engine<T, 0x100000000ull, 0, (1ull << 63) + 1ull> E2;
----------------
LRFLEW wrote:

No *specific* reason. Mostly, I was trying to make this code a little cleaner and consistent. I figured that the `(1ull << 63) - 0x1ffffffffull` I added a few lines down would be longer and less clear if written as `(1ull << 63) - (1ull << 33) + 1ull`, and if I'm writing the 31-bit hex literal there, then I might as well do it here as well. I don't mind changing this back if that's what's preferred, though.

https://github.com/llvm/llvm-project/pull/81583