Look if someone doesn't like it they can run v0.13.0 of bitcoin core and have the witness data removed. I dunno this whole convo just seems like a gotcha for Knots because "something-something arbitrary data."
You know what they are trying to do. You know why a miner's arbitrary data is more valuable than someone broadcasting their transaction. I would like to stick to more substantive conversations like removing the OP_RETURN field from standardness.
That would accelerate dark mempools
I don't think so, but I don't care if it does. Mempools are stupid anyway. Libbitcoin has a better model for verification, and data stoarge. But Core and Knots are stuck with this architecture.
How is libbitcoin made aware of new txs?
You have a similar gossip except it pulls from the: fastest, most data-full, and responsive nodes around you. I think I saw 100 peer connections but again, it takes EVERYTHING as a transaction, indexes, then if that transaction is ever found in a block it validates. It doesn't filter as the transaction comes in.
That sounds like something knots people wouldn't want. Why accept everything? Isn't the idea of knots to filter things that are unwanted?
You take everything because disk is the cheapest resource (also frees up RAM for the actuall chain), and mining is what determines what goes in so, if you have literally EVERYTHING, you as a miner do the filtering.
Paired with Stratum V2 or Datum, this means there is a truly decentralized immutable way of CHOOSING what you want your proof of work to put on chain.
The chain cannot fit in RAM
Lol, I'm telling you that the architecture is wild but using self-validated "checkpoints" as they call them, you can fit the highest chain on RAM. (The guy was using 128G of RAM) but yes, they did.
