Fix RNG used by cosmetics editor to use same RNG method as rando #5979
Conversation
| state = seed; | ||
| } | ||
|
|
||
| uint32_t ShipUtils_next32() { |
There was a problem hiding this comment.
ideally random.cpp core functions would change to taking state as parameter, then these endpoints with globals could pass those into the core function
There was a problem hiding this comment.
Yeah if we want to shift Rando over to these funcs as well that's what I would do, trying to decide if that's in-scope for right now or not. I guess if I'm doing this on develop instead of dev-copper I might as well go all the way.
| #if !defined(__SWITCH__) && !defined(__WIIU__) | ||
| uint64_t seed = static_cast<uint64_t>(std::random_device{}()); | ||
| #else | ||
| uint64_t seed = static_cast<uint64_t>(std::hash<std::string>{}(std::to_string(rand()))); |
There was a problem hiding this comment.
| uint64_t seed = static_cast<uint64_t>(std::hash<std::string>{}(std::to_string(rand()))); | |
| uint64_t seed = static_cast<uint64_t>(rand()); |
Beyond this PR's scope, but we should go ahead & rely on rand without the string-then-hash song & dance
There was a problem hiding this comment.
If this was on dev-copper instead of dev I'd say it's probably out of scope but since its on dev I might as well go ahead and make the bigger changes, so I'm thinking this is maybe not quite out of scope after all.
2 participants
Apparently,
std::random_deviceis not properly non-deterministic on certain situations where rng hardware is not accessible or something like that. I recently bumped into this with a Linux distro on my gaming desktop. So I essentially copied the randomizer rng to ShipUtils so I could use it for the Cosmetics Editor (withstd::random_deviceit was always generating every color to the same dark blue color). It is a copy of the same functionality, it does not use the same seeds/state variables so randomizing cosmetics won't effect randomizer seed generation and vice/versa.Build Artifacts