Switching to fork of ggrs with some fixes I am testing for desync. Desync occurs when client receives a frame it should rollback and correct while also at max prediction threshold (requires 3+ clients to hit failure, as if only 2 after receiving remote input 2nd would not be at max prediction anymore).
Switching to fork of ggrs with some fixes I am testing for desync. Desync occurs when client receives a frame it should rollback and correct while also at max prediction threshold (requires 3+ clients to hit failure, as if only 2 after receiving remote input 2nd would not be at max prediction anymore).
Changes being used are here: https://github.com/MaxCWhitehead/ggrs/tree/prediction-error-rollback-test
So far I no longer reproduce desync in jumpy, and the test I implemented in that branch is now passing.