Still need some polish but almost good

Rebased and squashed, this is now ready for review

Wrt #15 (comment) i wonder how far we should go with type safety. We could for example have rust into_prevout(&self, vout: u32) return the well-typed RevaultPrevout instead of a generalistic OutPoint. This would however incur a high complexity cost (we'd basically need our own boutique Transaction).

I think the current state is a good tradeoff (obviously), but i'm really interested in your opinions as well.

ACK ccb869d

Wrt #15 (comment) i wonder how far we should go with type safety. We could for example have rust into_prevout(&self, vout: u32) return the well-typed RevaultPrevout instead of a generalistic OutPoint. This would however incur a high complexity cost (we'd basically need our own boutique Transaction).

I think the current state is a good tradeoff (obviously), but i'm really interested in your opinions as well.

I think so too. I think type-safety here helps us guarantee that signatures are only generated for RevaultTransactions that conform to the protocol and avoids unnecessary exposure of private keys for inappropriate transactions.

What would be gained from having returning RevaultPrevOut instead of OutPoint?

What would be gained from having returning RevaultPrevOut instead of OutPoint?

Typically it'd have prevented my mistake in #15 (comment) . Which was:

// ..Or the spend_tx's change
let _sec_emer_tx = EmergencyTransaction::new(
       (VaultPrevout::new(spend_tx.into_prevout(1)), RBF_SEQUENCE),
  );

Instead of

// ..Or the spend_tx's change
let _sec_emer_tx = EmergencyTransaction::new(
       (VaultPrevout::new(spend_tx.into_prevout(0)), RBF_SEQUENCE),
  );

Sneaky, isn't it ? If into_prevout was strongly typed this would not have compiled at all. Anyways i think i can hack something similar using PSBTs, working on it now.