elliptic-curve: make FromEncodedPoint return a CtOption

[tarcieri]

Internally k256/p256 return a CtOption for this method, then convert to an Option.

It would be helpful when implementing other traits from the group crate, e.g. GroupEncoding, if this instead returned a CtOption.

[madninja]

I believe a corresponding change to ToCompactEncoding to return a CtOption instead of an Option would help with chaining key construction. The complexity in switching between CtOption and Option in the steps in: https://github.com/helium/helium-crypto-rs/pull/20/files#diff-522b3f5cfeb12570fd7f3eb9445fbe8a5151dadd10e4dc51ca79d2de9b51c9deR211-R230 doesn't smell right, does it.

I tried putting together a patch here but got lost in the elliptic_curves crate picking up that patch without newer commits creeping in

[tarcieri]

Yes, sorry, this got missed when group and the other APIs switched to CtOption.

Regarding that code example, there's an impl of From<CtOption<T>> for Option<T>, so you can use Option::from(ct_option) to do the conversion (or ct_option.into() if type inference permits it)

[madninja]

Yes, sorry, this got missed when group and the other APIs switched to CtOption.

I guess I'll await the change to trickle in since I couldn't get the parts to fit together right trying to make a PR for the two relevant crates.

Regarding that code example, there's an impl of From<CtOption<T>> for Option<T>, so you can use Option::from(ct_option) to do the conversion (or ct_option.into() if type inference permits it)

Ah yes! I missed that, updated. Thanks for the pointer

[tarcieri]

It's a breaking change, so we can't fix it until the next minor version bump

[madninja]

It's a breaking change, so we can't fix it until the next minor version bump

completely understand