Closed adevaykin closed 5 years ago
Good catch, I'd assumed it was working in my tests because I never test changing the seed, seems like a good test to add to prevent a regression. It looks like some of the unit tests are affected by this change since it (correctly) changes the random number seed. If you could fix the tests on your fork that would be great, otherwise I'll pull your changes into a branch here and fix up the tests myself.
Should be fixed now.
Thanks! I've merged your changes to master.
get_random_seed() returned bool _has_rand_seed casted to uint64_t, which led to STL assertion with "invalid random seed" when compiled with MSVC in debug mode.