Traversing over block data is a very common task and, at the same time, more complex that it should be. Reasons are:

• Dealing with era-specific details adds a lot of branching in consumer code
• Block data layout makes it harder to iterate over independent transactions

The new pallas-traverse lib attempts to abstract away that complexity by creating a multi-era layer on top of the ledger primitives. By providing a MultiEraBlock and a MultiEraTx, the lib consumer can code against a single structure regardless of the era.

This is an example of how to iterate over Tx of blocks from multiple eras (see `examples/block-decode):

use pallas::ledger::traverse::MultiEraBlock;

fn main() {
    let blocks = vec![
        include_str!("blocks/byron.block"),
        include_str!("blocks/shelley.block"),
        include_str!("blocks/mary.block"),
        include_str!("blocks/allegra.block"),
        include_str!("blocks/alonzo.block"),
    ];

    for block_str in blocks.iter() {
        let cbor = hex::decode(block_str).expect("invalid hex");

        let block = MultiEraBlock::decode(&cbor).expect("invalid cbor");

        println!("{} {}", block.slot(), block.hash());

        for tx in block.tx_iter() {
            println!("{:?}", tx);
        }
    }
}

The api surface of the lib is still limited, but the goal is to evolve it so that it becomes a single point of entry for "traversing" over any block data. At some point, this will be a replacement of the ad-hoc "crawling" mechanism that we have in Oura.