pow_vartime([n, 0, 0, 0] => pow_vartime([n]

[zhiqiangxu]

pow_vartime([n, 0, 0, 0] == pow_vartime([n] always holds, so it seems there's no point in wasting resource computing pow_vartime([n, 0, 0, 0] instead of pow_vartime([n].

[str4d]

This is probably left over from when pow_vartime was on FieldExt and took a fixed-length array (IIRC). Generally +1 on the PR, but I'm on PTO and am not reviewing Zcash PRs currently.

[zhiqiangxu]

This is probably left over from when pow_vartime was on FieldExt and took a fixed-length array (IIRC). Generally +1 on the PR, but I'm on PTO and am not reviewing Zcash PRs currently.

No hurry:)

BTW, it seems pow_vartime and pow are almost the same, is the difference really that big or another left over to be unified?

UPDATE

I see what constant time means here now: it doesn't mean it's faster, just that the time is constant no matter which branch is chosen, to anti timing attack.

E.g:

            fn conditional_select(a: &Self, b: &Self, choice: Choice) -> Self {
                // if choice = 0, mask = (-0) = 0000...0000
                // if choice = 1, mask = (-1) = 1111...1111
                let mask = -(choice.unwrap_u8() as to_signed_int!($t)) as $t;
                a ^ (mask & (a ^ b))
            }