I published my first rust library yesterday: https://crates.io/crates/bitcoin-nostr-relay

You can test it out in this playground environment:

https://github.com/vnprc/tx-relay-playground

I think this approach solves the filter debate by dismantling the bitcoin core architecture into separate components. Bitcoin core should be used only for blocks and blockchain consensus.

It is bad architecture to combine TX relay and block relay into a single binary with a single peer-to-peer network. It opens the node to many DOS attack vectors. It open the developers to social attack vectors. And it forces unnecessary constraints on bitcoin transactions (mempool policy) that severely limit what developers can accomplish with off-chain protocols like lightning.

Why continue down this path when alternatives are available?

I intend to continue developing permissionless software solutions to enable sovereign pool operators to meet and exceed the capabilities of large, closed source pool operators.