perf(stark-curve): no subgroup check on prime-order curve

[yelhousni]

No description provided.

[ivokub]

Looks good.

I'm trying to understand what was the previous behaviour - it seems that the method didn't depend on p at all and always returned true?

For long-term, should we add tests which try to pass non-valid points and check that the result is false?

[yelhousni]

I'm trying to understand what was the previous behaviour - it seems that the method didn't depend on p at all and always returned true?

Oh right! the res point was always the zero point which is always on curve and has z=0 in Jacobian coordinates. So the tests pass no matter what p is.. it should have been res.ScalarMultiplication(p, fr.Modulus()). But since the curve has a prime order we just check if p is on the curve (as for bn254/g1 and secp256k1).

For long-term, should we add tests which try to pass non-valid points and check that the result is false?

Some invalid test vectors here would definitely be useful !