On Wed, Jan 04, 2023 at 01:06:36PM +1100, Lloyd Fournier wrote:
> The advantage of using a covenant
> is that the channel would not have an expiry date and therefore be a first
> class citizen among channels.

I think the approach here would be:

 * receive funds on the in-potentiam address with 8000 block CSV
 * LSP tracks immediately
 * user's wallet wakes up, LSP reports address to user, user signs
   a funding tx to establish the channel, and the state 0 close tx
 * LSP considers it immediately active
 * LSP broadcasts the tx, targeting confirmation within 3 days
 * if the funding tx confirms promptly, you just have an ordinary channel
 * after 800 blocks, if the tx hasn't confirmed, LSP fee bumps or
   closes the channel (relying on the high-feerate unilateral close
   tx to do the fee bumping)

ie:

 day 0: someone -> in-potentiam address (payment made on-chain, confirmed)

 day 7: in-potentiam -> funding (wallet wakes up, tx signed and broadcast,
          not necessarily confirmed, channel active)

 day 12: in-potentiam -> funding (confirmed)

 day 9999: funding -> unilateral/cooperative close

or:

 day 0: someone -> in-potentiam address (payment made on-chain, confirmed)

 day 14: LSP forgets about in-potentiam utxo as its expiry is only 1000
         blocks away

 day 420: in-potentiam -> wherever (payment made on-chain by user)

So while the tx introspection approach you advocate *would* allow the
setup phase to skip the "expiry on day 14" restriction, I think the
*bigger* benefit is that you also wouldn't need the on-chain "in-potentiam
-> funding" transaction, but could instead just leave the in-potentiam
tx on chain indefinitely, until it was time to close the channel (which,
if it was a cooperative close, could just be a musig key path spend).

Either approach probably implies that you either have multiple channels
with your LSP (one for each in-potentiam payment you receive), or
that your single channel with your LSP is backed by multiple UTXOs
(maybe you choose an order for them, so that Alice owns 100% of the
balance in utxos 1..(k-1) and Bob owns 100% of the balance in utxos
(k+1)..n?). Otherwise you'd need an on-chain tx anyway to splice the new
funds into your existing channel; and that seems both annoying of itself,
and probably bad for privacy.

Cheers,
aj

_______________________________________________
Lightning-dev mailing list
Lightning-dev@lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/lightning-dev

Reply via email to