feature: add e2hs mode to portal-bridge

[njgheorghita]

What was wrong?

With the introduction of the new e2hs file format, we need a bridge mode to support these files.

This is a very basic bridge mode, that won't really work unless you have all necessary e2hs files in a local test-e2hs dir, inside the root level of the trin dir. But the intention is that this will be updated to use a centralized server that hosts all e2hs files, once they are generated and made available.

The existing BridgeMode type has turned into a bit of a frankenstein, as a result of us trying to push too much into a single cli arg. After this new bridge mode matures, I expect a fair amount of legacy bridge code to be deleted, that's why I decided to create new, isolated cli args for the e2hs mode to use, rather than try and squeeze them into the old BridgeMode type.

There's another trade-off to consider. I opted to require the user to provide the entire block range, rather than using a live el provider to detect the head of the chain. I'm not married to this, but I think it's easier if a user just wants to casually run the bridge, without an infura / pandaops account.

But both of these decisions are fairly trivial, and can be changed as new use-cases for the bridge are implemented.

Usage

"--mode e2hs --e2hs-range 1000-10000 --e2hs-randomize": randomize the order in which epochs from block range are gossiped

How was it fixed?

• added e2hs mode to portal-bridge

To-Do

• Clean up commit history and use conventional commits.

[KolbyML]

looks good. I think it is fine if we iterate on this like you explained in the PR's description.

[KolbyML]

Suggested change

	fn validate_block_tuple(&self, block_tuple: &BlockTuple) -> anyhow::Result<()> {
	self.header_validator
	.validate_header_with_proof(&block_tuple.header_with_proof.header_with_proof)?;
	let receipts = &block_tuple.receipts.receipts;
	let receipts_root = receipts.root();
	ensure!(
	receipts_root
	== block_tuple
	.header_with_proof
	.header_with_proof
	.header
	.receipts_root,
	"Receipts root mismatch"
	);
	let body = &block_tuple.body.body;
	body.validate_against_header(&block_tuple.header_with_proof.header_with_proof.header)?;
	Ok(())
	}
	fn validate_block_tuple(&self, block_tuple: &BlockTuple) -> anyhow::Result<()> {
	let header_with_proof = &block_tuple.header_with_proof.header_with_proof;
	self.header_validator
	.validate_header_with_proof(header_with_proof)?;
	let body = &block_tuple.body.body;
	body.validate_against_header(&header_with_proof.header)?;
	let receipts = &block_tuple.receipts.receipts;
	ensure!(
	receipts.root()
	== header_with_proof
	.header
	.receipts_root,
	"Receipts root mismatch"
	);
	Ok(())
	}

I think ordering it like this would be a bit more readable. Also I don't think we need the extra variable receipts_root

Idk up to you

[KolbyML]

Suggested change

	let header_number = block_tuple
	let block_number = block_tuple

I think this is more clear

[KolbyML]

Suggested change

	let header_hash = block_tuple
	let block_hash = block_tuple

I think this is more clear

[KolbyML]

We are doing this 3 times with 3 different variable names

• block_number
• number
• header_number

Could we only do it once, then just pass it as a variable and use block_number as I think that is the standard terminology.

[KolbyML]

Suggested change

	match block_range {
	Some(_) => {
	info!("Gossiping block range: {block_range:?} from epoch {epoch_index}",);
	}
	None => {
	info!("Gossiping entire epoch {epoch_index}");
	}
	}
	match block_range {
	Some(_) => info!("Gossiping block range: {block_range:?} from epoch {epoch_index}"),
	None => info!("Gossiping entire epoch {epoch_index}"),
	}

Perhaps we could condense it now like this?

[KolbyML]

Suggested change

	error!(
	"Failed to gossip history content key: {:?} - {:?}",
	hwp_by_number_content_key, err
	);
	error!("Failed to gossip history content key: {hwp_by_number_content_key:?} - {err:?}");

[KolbyML]

Suggested change

	error!(
	"Failed to gossip history content key: {:?} - {:?}",
	body_content_key, err
	);
	error!("Failed to gossip history content key: {body_content_key:?} - {err:?}");

[KolbyML]

Suggested change

	error!(
	"Failed to gossip history content key: {:?} - {:?}",
	receipts_content_key, err
	);
	error!("Failed to gossip history content key: {receipts_content_key:?} - {err:?}");

[KolbyML]

Suggested change

	if parts.len() != 2 {
	return Err(anyhow!("Invalid block range format"));
	}
	ensure!(parts.len() == 2, "Invalid block range format");

[KolbyML]

Suggested change

const EPOCH_SIZE: u64 = 8192;

Isn't this a pretty common constant couldn't we import this from somewhere else?