Meeting Notes 2021 11 29

Target release for 0.0.104

#1054 (ariard)
#1106 (TheBlueMatt)
#129 (?)
#216 (?)
#240 (?)
#580 (?)
#615 (?)
#630 (?)
#784 (?)
#993 (?)
#996 (?)
#1069 (?)
#1111 (?)
#1126 (?)
#1129 (?)
#1164 (?)
#1174 (?)
A few are just requests for review
Only thing that really needs to be fixed is 1164 (all paths failed thing), matt todo soon
Whenever thats done we’ll cut 104, and fix minor issues along the way
Antoine: would be useful for someone to go thru the spec and check for issues/inconsistencies in LDK
1056
- Auto close if no fees updated for a long period of time
- Matt: thinking from a user pov, if feerate spikes 5x on a monday, peer disconnected, im prob still ok w the old feerate if i have 0 htlcs. Should we only do the autoclose if htlcs are pending?
- Antoine: we do that in the latest version

Matt: chatted w rene, easiest way to model available bandwidth on a channel is to model it as a min and max estimates for a channel’s liquidity that gets continually updated on success/failure
- Doesn’t rly capture nodes being offline
- Seems trivial, could go in this direction, no pressure to do it immediately, but seems simple

Ariard: Unknown what tibo has in mind for next step
High level issue: if u want to dlc over ln, every channel you’ll update the commit tx, you have to re-sign the dlc contract txs, and you might have like 10,000. Quite slow.
Every time htlc shared on same commit tx as dlc one, might have to re-sign for every pending dlc across your channel
ongoing conversation on the ML: https://mailmanlists.org/pipermail/dlc-dev/2021-November/000091.html and https://github.com/Tibo-lg/lightning-rfc/pull/1/files
Rn, no channel jamming mitigation across the network, also an issue
One solution is sighash_anyprevout, bc it moves txid out of children txs, so can change states and don’t need to re-sign the children. Thats nice but we don’t know when sighash_anyprevout will be deployed
Lloyd, nadav, chris, are also involved in these meetings
Currently just trying to get out the first specifications
Matt: we can always spend time on custom channel scripts, got other stuff to do now tho

Video meeting last week
There’s a recurring disagreement about how to view feature bits, comes up too often
PR 1179 is an important first step to 0-conf

Devrandom working on best way to get onchain validation support. Given utxo is on or off chain. Have been looking a utreexo and compact block filters for this
Recently, devrandom decided maybe we can just use spv w/ a heuristic about the hashrate that we observe. So as long as we see hashrate is proceeding apace, we can presume we’re not being eclipsed and perhaps build a simpler version of that immediately and move on
- Not necessarily long term but nice initial step
Everything going pretty well, got some anchors advances merged recently

Buncha papercuts crop up on the java side
Arik: ran into some issues getting ram (the guy from apple who wants to contribute to ldk-swift) have the generator compile on his machine. Considering it was a mac, i was surprised since it should’ve been identical to my setup. We did get it to work in the end, so we’ll be able to work together fixing a bunch of issues after a couple days of me working on lnsync stuff. In the interim, i wanna figure out why there was this discrepancy.

We’re a ways out of sync here
Haven’t had a chance to work on it
We really just need to automate the website to be something that gets updated in code instead of being a separate html document, but that’s a whole other project
Conor mentioned adding a quickstart guide to the website using polar
#95 and #96 need review
community: public board for feature request/milestones tracking, see https://lightningdevkit.slack.com/archives/CSYT185MY/p1637600110131600