LDK Splicing Meeting 10/12/2023

notes.md

Dual-Funding/Splicing Sync 10/12

Dual-Funding

• Duncan has been working within 2302 for v2 chan establishment
• Working with constructor has been working fine, and has done work up to tx_signatures
• Duncan left suggested changes in the interactive txs PR
• Precompute the serial ids instead of passing an entropy_source in interactive txs

What should we do for errors? Not for illegal state transitions, but for when our counterparty sends a message completely unexpected.

• we should send warning, and then disconnect instead of closing the channel since we do not want to do that for a splice use-case

• Duncan has been writing tests against interactive-txs at the channel level

• RBF

  • how would we want to expose this to the user?
  • we can use the bump-transaction event
  • method on ChannelManager we can call back into for adding additional inputs and outputs to pass to channel in order to commence a new interactive tx construction
  • if the channel is unconfirmed, we RBF on the funding tx, else we do it on the splice tx, otherwise, error

• RBF for dual-funding is pretty straightforward until we do splicing

• Do we need one monitor for each RBF?

  • We cannot have separate channel structs because they will have the same channel_id
  • Wilmer will follow-up with Duncan

• 2302 will not include RBF work, maybe merged for 119 instead of 118

• Dual-funding can be released as something experimental just like what was done for anchors

Splicing

• After BOLT 12, splicing is becoming the next priority
• Users have had a lot demand for this, and have started working around to get it
• Wilmer started work on the quiescence protocol
• Some testing needs to be added to take the PR out of draft, Jurvis and Wilmer to catch up with optout to find a sensible approach to that