try to fix panic at CompressedRistretto from_slice

[HaoXuan40404]

Hello, I found that the function "from_silce" in src/ristretto.rs will happen a panic when receiving a slice of incorrect length. Maybe this unreasonable. And then I try to fix it.

Test main.rs like this

extern crate curve25519_dalek;
use curve25519_dalek::ristretto;
use curve25519_dalek::scalar::Scalar;

fn main() {
    let test_u8 = [1, 2, 3, 4];
    let test_point = ristretto::CompressedRistretto::from_slice(&test_u8);
    println!("test_point = {:?}", test_point);
}

Then I got this:

thread 'main' panicked at 'assertion failed: `(left == right)`
  left: `32`,
 right: `0`: destination and source slices have different lengths', src/libcore/slice/mod.rs:2119:9
stack backtrace:

[tarcieri]

I think the correct solution to avoiding a panic here is to have a fallible API, not zero-padding. Your “solution” is still panicking because copy_from_slice requires that the two slice lengths are equal, and you’re not slicing the temporary buffer to match the input length. But even if you did, this approach can’t handle an overlength input, nor does zero padding actually help in this context.

[HaoXuan40404]

So maybe return ”result” would be better?

[HaoXuan40404]

Ok, so this fallible API is in expected, right?

[tarcieri]

Yes, although it’d be a breaking API change.

If @hdevalence is ok with an MSRV bump, you could add a new fallible API by adding a TryFrom impl perhaps.

[HaoXuan40404]

Ok, thank you