Use alloy instead of primitive-types and ethereum-types

[0xaatif]

obsoletes 0xPolygonZero/eth-tx-proof#22 as per 0xPolygonZero/zk_evm#157 (comment)

• rewrite ::rpc
• add ::rpc::Compat
• trim deps

Reviewers

• this is the most risky change:

  zero-bin/rpc/src/lib.rs

  Lines 172 to 185 in 3a5b21b

  	let alloy::primitives::Bloom(alloy::primitives::FixedBytes(src)) = self;
  	// have u8 * 256
  	// want U256 * 8
  	// (no unsafe, no unstable)
  	let mut chunks = src.chunks_exact(32);
  	let dst = core::array::from_fn(|_ix| {
  	// This is a bit spicy because we're going from an uninterpeted array of bytes
  	// to wide integers, but we trust this `From` impl to do the right thing
  	__compat_primitive_types::U256::from(
  	<[u8; 32]>::try_from(chunks.next().unwrap()).unwrap(),
  	)
  	});
  	assert_eq!(chunks.len(), 0);
  	dst

• Unsure how to test this change at all, since this branch and the target branch both fail CI in the same way:
  • target: https://github.com/0xPolygonZero/zero-bin/actions/runs/9205710549/job/25321965551#step:3:717
  • this:

[muursh]

Does anyone happen to know why they haven't/aren't releasing this? It makes me slightly uncomfortable with unreleased code but I don't think we really have another option here. We've been looking at alloy in the agglayer too but haven't made the switch yet for this exact reason. But... I think we just bite the bullet here.

[muursh]

I'm going to gently push for more tests here, I guess that's not a massive surprise :)

[Nashtare]

Not much to add, looks good to me overall, thanks!

[Nashtare]

@0xaatif you mentioned having trouble testing this branch because of some failure. Has this been resolved since?

[muursh]

@0xaatif you mentioned having trouble testing this branch because of some failure. Has this been resolved since?

#79

[Nashtare]

I think the error quoted in the CI is independent of the linker error. It was complaining about Error: invalid type: integer 2024, expected struct ProofWithPublicInputs at line 1 column 5

[cpubot]

I think the error quoted in the CI is independent of the linker error. It was complaining about Error: invalid type: integer 2024, expected struct ProofWithPublicInputs at line 1 column 5

I believe so as well. Resolving the linker error resolved the compile error, but the test returns this error now. Not sure how long this has been broken, but seems unrelated

[muursh]

I mean I'm open to believing it really exists if someone can actually make it happen

[Nashtare]

Not sure how long this has been broken, but seems unrelated

I don't know how this failure is reproducible, the script is passing fine on my laptop.
However, I think this branch hasn't been tested properly before merging. Compilation with the test_only feature is broken.

[muursh]

Compilation with the test_only feature is broken.

@0xaatif