Bug fixes and extra features?

[nanaknihal]

I've been using this library for a couple projects and have added various features and bug fixes in a fork: https://github.com/nanaknihal/babyjubjub-rs
If you're interested, I will make a PR. First I'd have to refactor some code because I made some unnecessary breaking changes (e.g., using a different hash) that need to be changed back before I make the PR. Please let me know if that would be helpful and I will make the PR.

List of features I have added

• Point.on_curve(&self) -> bool and Point.in_subgroup() -> bool methods to check that Points are safe to operate on for certain constructions, e.g. ElGamal decryption.
• ElGamal encryption and decryption
• DLEQ proofs
• Method for taking negative of point,
• Fl : same as Fr but uses the subgroup's order instead of r
• Serialization and Deserialization of Fr and Fl into JSON or BigInts. These aren't the most elegant or efficient (using strings as intermediate representation) but they work

List of fixes I can make

• Fixing bug in Point.mul_scar by only allowing a BigUInt. Currently, it takes BigInt, and actually gives the wrong answer when the inputs is <0. Cleanest way might be to force BigUInt as input which would require breaking changes. Easiest way without breaking would be just to panic if the input is negative.
• Fixing bug in ORDER being stored as an Fr, when it is larger than r. It should probably rather be stored as a BigInt.

I'm happy to merge these features, and let me know if so which bug fixes I should add as well. The Point.mul_scalar can be fixed number of different ways; if we're fine with panicking that would be an easy fix!

[arnaucube]

Sounds good! The initial objective of this repo was to have a rust implementation compatible with circom's EdDSA over BabyJubJub version from circomlib. Also this was done in a moment where arkworks was not a thing yet (nowadays I would use directly arkworks). So, as long as we keep compatibility with circom lib, all improvements and new features are more than welcome ^^

[nanaknihal]

Finally done in #8 (except the mul_scalar negative input check). Sorry for the delay -- I had changed the blake hash function because I was having trouble compiling on my M1 mac, and couldn't merge until I had a chance to get the original codebase to run. Turns, out I just was missing the flag --features aarch64 for cargo build|test|run which should hopefully help for anyone else running into issues on an Apple Silicon.