Closed mossprescott closed 6 years ago
Also note this is a breaking change to the API; most of the interesting functions have lost a parameter and moved from IO to MonadIO and MonadRandom constraints.
Please advise if the PR should bump the version and what to.
This was motivated by what appeared to be a bug in crypto-random that led to duplicate nonces being generated.
Switching to MonadRandom significantly cleans up the code and also gives some additional flexibility in where the randomness comes from.
Note: this essentially recapitulates PRs #16 and #17 against the current state. I will comment on a couple of spots where this differs from those PRs.