Add tree hash implementation for fixed bytes

[Nomad-Free-Talent]

Relevant to #22, I'm adding the new impl_for_fixed_bytes macro which can be used to implement the TreeHash trait to FixedBytes<N> struct.

It's not available for us to implement it like impl<const N: usize> for FixedBytes<N> because there's already a TreeHash implementation for B256 which is represented as FixedBytes<32>.

tree_hash/tree_hash/src/impls.rs

Lines 166 to 182 in e9708cd

	impl TreeHash for B256 {
	fn tree_hash_type() -> TreeHashType {
	TreeHashType::Vector
	}
	fn tree_hash_packed_encoding(&self) -> PackedEncoding {
	PackedEncoding::from_slice(self.as_slice())
	}
	fn tree_hash_packing_factor() -> usize {
	1
	}
	fn tree_hash_root(&self) -> Hash256 {
	*self
	}
	}

[michaelsproul]

I think we can probably make a generic impl that covers B256 as well.

[michaelsproul]

It's weird that we have Vector type and still specify a packed encoding. I think that code might be unreachable, need to check.

[Nomad-Free-Talent]

I think we can probably make a generic impl that covers B256 as well.

The implementation logics between FixedBytes<48> and B256, which can be assumed as FixedBytes<32>, are different each other.

It's weird that we have Vector type and still specify a packed encoding. I think that code might be unreachable, need to check.

The FixedBytes type is defined as a wrapper of [u8; N] type. I think, in this case, we should define an additional implementation logic.

[michaelsproul]

The implementation logics between FixedBytes<48> and B256, which can be assumed as FixedBytes<32>, are different each other.

They are different, yes. But I'm not convinced that they should be. I think that might be an oversight

[Nomad-Free-Talent]

I'm not convinced that they should be.

Do you think we can just implement the tree_hash implementation logic for FixedBytes<const N: usize>?
In that case, the original B256 related logic will be ignored.

[michaelsproul]

Yes, see:

• Generic FixedBytes implementation and bug fixes #28

I need one of my colleagues to review, but I think we should be Ok with a generic impl.

[michaelsproul]

I ended up adding the generic implementations in that PR, so I could see all the changes in one place. I can add you as a co-author if you would like?

[Nomad-Free-Talent]

I ended up adding the generic implementations in that PR, so I could see all the changes in one place. I can add you as a co-author if you would like?

Thank you for your help. Can we release the v0.9.1 with this change? I need the ssz_types crate with new tree_hash crate imported in it.
https://github.com/sigp/ssz_types/blob/b7f47d95d5cd926a2c299260838f2652a1f962c4/Cargo.toml#L14

And I'd be happy to be a co-author with you.

[michaelsproul]

Yes I think we can do a backwards-compatible v0.9.1 release, as soon as my PR is reviewed.

I've added a commit with you as co-author so you'll get some credit. I'll close this PR now in favour of:

• Generic FixedBytes implementation and bug fixes #28

[Nomad-Free-Talent]

Thank you for your help with this task.

[michaelsproul]

v0.9.1 is out now 🎉

[Nomad-Free-Talent]

v0.9.1 is out now 🎉

This is great. I'll create another PR on ssz_types repo with new tree_hash crate version.

[Nomad-Free-Talent]

I created a PR towards the ssz_types crate to use the tree_hash@0.9.1.
sigp/ssz_types#42