Closed porcuquine closed 5 years ago
I actually meant to make this PR against our fork for now, but I'll leave it here as a very focused way to ask whether this approach will work. Our tests all pass, and this is an important optimization for us.
EDIT: I'm aware this breaks tests because it changes the number of constraints. I don't expect to have this merged, but is there a way to accomplish what we want that does the right thing and still works for you? If not, maybe we could parameterize initialization so we could use our own values without having to maintain a fork.
This allows us generation of Pedersen hashes with pre-images of 17 * 32 bytes.