release 0.28.1

[apoelstra]

We need a new rust-secp release to deal with the new rust-secp-sys release.

[apoelstra]

cc @Kixunil can you quickly ACK this?

[Kixunil]

This shouldn't be required for the sys crate since it's not semver-breaking. (And while we did strictly change the API it was broken in the first place so I believe non-semver-breaking version was correct.)

[TheBlueMatt]

The changes between 60a5e36 and 940e56f LGTM, but are also only a version bump.

ACK 940e56f

[Kixunil]

ACK 940e56f

[apoelstra]

@Kixunil the change in -sys was semver-breaking because we changed the signature of a publicly-exposed FFI function. rust-secp assumed that signature, and was therefore broken by it.

[apoelstra]

I do think this semver break is okay because it was a security-sensitive fix to an objectively wrong signature, which is why I'm updating rust-secp rather than reverting the change...but it is a semver break.

[apoelstra]

Oops, need re-ack. I forgot to update the lockfiles.

[sanket1729]

utACK 0ed5a55

[Kixunil]

the change in -sys was semver-breaking

If it was then it should've been 0.10, not 0.9. :) But I claim it isn't since it was already broken. (Maybe subtly broken; this is in line with Rusts policy of fixing soundenss bugs which I believe is a great one.)

And since we didn't bump semver-majoor version this bump is not required but neither harmful. (And serialized signature is nice...)

[Kixunil]

ACK 0ed5a55

[apoelstra]

And since we didn't bump semver-majoor version this bump is not required but neither harmful. (And serialized signature is nice...)

It is required because we broke compilation of rust-secp :).

[apoelstra]

Tagged, published, and yanked 0.28.0.

[Kixunil]

Oh, I see now, didn't realize the previous change updated stuff in non-sys crate too.