Repository navigation
Bump LDK to v0.3.0-rc4 - #1231
Open
benthecarman wants to merge 1 commit into
Open
benthecarman wants to merge 1 commit into
benthecarman wants to merge 1 commit into
Conversation
|
👋 Thanks for assigning @tnull as a reviewer! |
Collaborator
|
CI failure should be fixed by https://git.rust-bitcoin.org/lightningdevkit/rust-lightning/pulls/5575 |
benthecarman
marked this pull request as ready for review
November 11, 2026 19:34
benthecarman
force-pushed
the
update-ldk-0.3-rc4
branch
from
November 11, 2026 19:38
3770094 to
249f6bc
Compare
tnull
reviewed
Nov 12, 2026
tnull
left a comment
Collaborator
There was a problem hiding this comment.
Looks good I think, but can't move forward with this as v0.3-rc4 is broken.
Contributor
Author
|
It's on your merged commit that fixes |
tnull
reviewed
Nov 17, 2026
tnull
left a comment
Collaborator
There was a problem hiding this comment.
Claude review:
Findings
- High: zero-conf splices silently stop being zero-conf (src/config.rs:849, and everywhere UserConfig is built).
- rc4 adds inbound_splice_min_depth and outbound_splice_min_depth, both defaulting to 6, and no longer derives splice depth from the channel's minimum_depth.
- The PR fills in the defaults, so splices on channels with trusted_peers_0conf or LSP-trusted peers now need 6 confirmations before splice_locked. On rc3 they locked immediately.
- Existing channels are affected too, since persisted configs without the new fields read as 6.
- superseded_zero_conf_splice_round_keeps_its_payment_at_close (tests/integration_tests_rust.rs:4383) expects ChannelReady right after the splice with no blocks mined, so I expect it to time out. I could not confirm this from CI: the run on this head has one failure (build (macos-latest,
beta)) and many cancelled jobs, and I can't read the logs.- The PR should either set the knobs to 0 where ldk-node treats the peer as 0conf-trusted, or state the behaviour change in the changelog and fix the test. Setting either knob to 0 disables splice RBF on that channel.
- Low: update_channel_config will reset the new knobs (src/config.rs:840-851). From for LdkChannelConfig now uses ..LdkChannelConfig::default(), so any config round-trip resets the splice depths to 6. This is harmless today, but becomes a bug as soon as finding 1 is fixed by
setting non-default values.- Low: no test for the new LSPS2 delay cap (src/liquidity/service/lsps2.rs:496-503).
- The logic looks right: it caps the correct handshake limit, does the min in u32 as LDK's docs ask, and has a sensible fallback for old persisted events.
- It is a behaviour change, because the service never enforced max_client_to_self_delay before. A service configured below the client's delay (144 for LDK clients) will now reject every JIT open, with the usual silent failure and the HTLC held until timeout.
- Every integration test uses 1024, so neither the accept nor the reject path is exercised.
- Nit: the commit message and PR description are slightly off. "Tip of the 0.3 branch" is stale: the branch is now at 3ed13ee3, which adds the re-gating of inbound v2 channel opens and a rustfmt bump. The tag is v0.3-rc4, not v0.3.0-rc4.
Pin every LDK crate to commit 3ed13ee3 on the 0.3 branch, i.e., the v0.3-rc4 tag plus the fix for the Esplora build of lightning-transaction-sync and the re-gating of inbound v2 channel opens. LSPS2 OpenChannel events now carry the client to_self_delay limit agreed upon in the selected opening fee parameters, which the service must enforce itself. We now cap the channel's their_to_self_delay handshake limit accordingly, falling back to our configured limit for events persisted before the field existed. LDK's ChannelConfig gained inbound and outbound splice minimum-depth knobs, which default to six confirmations and no longer follow the channel's own minimum depth. We adopt that default, so splices on zero-conf channels are no longer zero-conf. The test for superseded zero-conf splice rounds, which can no longer occur, is replaced by one covering the new behavior, and the test for a second splice on a zero-conf channel now waits for each splice to confirm. As we don't expose the knobs, our ChannelConfig is applied as a partial update so that it doesn't reset them. LDK now bounds the RBF adjustment of a prior splice contribution by the bump's own maximum feerate rather than the prior round's, so a bump above the prior maximum is paid from the existing inputs where they suffice. The tests relying on such a bump to pull in a new coin now size the first round so its inputs cannot pay the bump. bitcoin-payment-instructions is moved to a fork commit pinned to the same LDK revision so both resolve to a single copy of LDK. Developed with assistance from Claude Code.
benthecarman
force-pushed
the
update-ldk-0.3-rc4
branch
from
November 17, 2026 20:54
81878f8 to
6d99f4d
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Bumps LDK to v0.3.0-rc4, pinned to
3ed13ee3on the0.3branch: the v0.3-rc4 tag plus the Esplora build fix (lightningdevkit/rust-lightning#5575) and the re-gating of inbound v2 channel opens.OpenChannelnow carriesmax_client_to_self_delay, which we enforce by cappingtheir_to_self_delayon the channel we open.ChannelConfigis now applied as a partial update so it doesn't reset the knobs.