Add keccak services/requests/messages

[viquezclaudio]

Functionality to request and relay keccak256 proofs

Pull request checklist

• All tests pass. The project builds and runs.
• I have resolved any merge conflicts.
• I have resolved all clippy and rustfmt warnings.

[pkg-pr-new]

Open in StackBlitz

npm i https://pkg.pr.new/nimiq/core-rs-albatross/@nimiq/core@3580

commit: 90efdcd

[jsdanielh]

Why aren't we using digest?

[viquezclaudio]

Because Hasher::digest only takes a single &[u8], so it can't hash two byte slices in sequence. We need to hash self || other without allocating.

[jsdanielh]

Ahh, sorry. Yes, I meant chain

[jsdanielh]

Same as above

[jsdanielh]

Same as above

[jsdanielh]

Same as above

[jsdanielh]

Isn't it better to use chain to avoid importing std::io::Write?

[jsdanielh]

I was thinking on chain from the hasher. Something like:
Blake2bHasher::default().chain(self.as_bytes()).chain(other.as_bytes()).finish();

[viquezclaudio]

Ok there is a problem with chain because doing something like:
Blake2bHasher::default().chain(self.as_bytes()).chain(other.as_bytes()).finish().into(),

gives a compilation error because chain<T: SerializeContent>(self, h: &T) uses a type parameter T that is Sized by default. self.as_bytes() is &[u8], so T is inferred as [u8], which is unsized, so the compiler errors.

The fix would be to use the hasher's write API, as it currently is

[jsdanielh]

Ah yes, but then you could implement SerializeContent for AnyHash:

impl SerializeContent for AnyHash {
    fn serialize_content<W: io::Write, H: HashOutput>(&self, writer: &mut W) -> io::Result<()> {
        writer.write_all(self.as_bytes())?;
        Ok(())
    }
}

And then:

Blake2bHasher::default()
    .chain(&self)
    .chain(&other)
    .finish()
    .into(),

[jsdanielh]

LGTM