Closed hobnob closed 2 weeks ago
There's a lot of good stuff in here that I don't want to lose, however there are two points to have another think about:
Dice are supposed to be 'predicatably unpredicatable'. By salting the seed with the initial date, I feel like we lose a bit of that. Now, I don't know in practice if it makes a great deal of difference: Is anyone really testing that way? Seems unlikely, but I'm also not sure if adding the date will improve the issue being reported. If not then it's an arbitrary change we don't need.
Branch needs a rebase.
Done :)
Tests now run 3 lots of 200,000,000 rolls of dice and check the distribution of the results in order to test the assertion made in #750
There's a minor code cleanup here too, but nothing breaking (just re-using a method rather than having 2 similar methods)