--- Log opened Tue Jul 05 00:00:12 2022
05:56 -!- andytosh1 is now known as andytoshi
10:17 -!- evanlinjin [~root@gateway/tor-sasl/evanlinjin] has quit [Remote host closed the connection]
10:17 -!- evanlinjin [~root@gateway/tor-sasl/evanlinjin] has joined #bitcoin-rust
19:08 < ariard> BlueMatt[m]: hey do we have data on the success rate of first-routing attempt of HTLC sends? and do you think we can make it equal to 99.9999% in the future?
19:09 < BlueMatt[m]> you mean like without any knowledge of channel reliability, just based on the network graph?
19:09 < BlueMatt[m]> i could look something up but there's no way in hell we're ever gonna get that to 6 nines lol
19:10 < ariard> what do you mean by channel reliability, like probing?
19:10 < BlueMatt[m]> or at least having sent many payments to learn which channels are reliable
19:10 < BlueMatt[m]> if you accept downloading scoring data from a third party you could probably do okay, with trampoline probably quite good, but you always get a lot of failures that are just "recipient has no liquieity"
19:10 < ariard> okay good to know, we're finishing some channel jamming research, and one of the conclusion for upfront fee sounds should be paid at HTLC forward, though if rate failure too high sounds not fair for HTLC senders
19:10 < BlueMatt[m]> liquidity
19:11 < ariard> okay "recipient has no liquidity" too high
19:11 < BlueMatt[m]> right, yea, you just see a lot of failures in practice
19:11 < ariard> maybe if we assume perfect knowledge of the balance, like the hops could announce their balance
19:11 < ariard> though you will still have issues with information propagation
19:12 < BlueMatt[m]> yea, i mean sure if you broke everyone's privacy maybe you could do okay
19:13 < ariard> maybe if we have 99% only, and the routing fees far in the future are medium, that might an "acceptable" loss
19:14 < BlueMatt[m]> i think the only future in which lightning has any privacy is one where we accept that payment failure rate will be kinda high and we have to retry a lot
19:14 < BlueMatt[m]> there are ways to make the retry less painful too, eg the sender over paying and making the recipient ask for something mixed into the preimage
19:14 < ariard> hmmmm, so attaching an upfront fee to any retry not good
19:16 < ariard> well the recipient mixing into the preimage that's still a risk for the routing hop, introducing an upfront fee the purpose would be to hedge against that, i think
19:17 < ariard> okay, i'll mind the payment failure rate gonna stay high, even if i don't like the answer for a practical upfront fee scheme :/
19:18 < BlueMatt[m]> right i dont mean as a protection against jamming, i mean as a way to allow you to send more htlcs than the payment is for
19:18 < BlueMatt[m]> "over-commit" the payment and then only let the recipient claim the right amount
20:39 -!- z9z0b3t1c [~z9z0b3t1c@145.224.65.3] has quit [Ping timeout: 244 seconds]
20:45 -!- z9z0b3t1c [z9z0b3t1c@gateway/vpn/protonvpn/z9z0b3t1c] has joined #bitcoin-rust
23:35 -!- z9z0b3t1_ [z9z0b3t1c@gateway/vpn/protonvpn/z9z0b3t1c] has joined #bitcoin-rust
23:38 -!- z9z0b3t1c [z9z0b3t1c@gateway/vpn/protonvpn/z9z0b3t1c] has quit [Ping timeout: 255 seconds]
--- Log closed Wed Jul 06 00:00:13 2022