Back to sw

[swasilyev]

Pasta curves don't have edwards form, so I found and replaced:

use ark_ec::twisted_edwards::{Affine, TECurveConfig};
-> use ark_ec::short_weierstrass::{Affine, SWCurveConfig};

TECurveConfig -> SWCurveConfig

ark_ed_on_bls12_381_bandersnatch::EdwardsAffine
-> ark_ed_on_bls12_381_bandersnatch::SWAffine

The question is. Do we have to restore the TE implementation fr bandersnatch? I would assume that our proofs are short-lived, so we can upgrade carefully?

[davxy]

This change would invalidate several JAM-related components, which is not something we can take lightly.
This involves changes to the GP, fuzzing tools, published and established test vectors, and implementors racing for the JAM price.

I imagine that is possible for both approaches to coexist? IIRC there was a PR quite a while ago.
For Bandersnatch, we need to continue using TE for now.