0.25.12-alpha
Please report bugs using the issue tracker at GitHub:
https://github.com/flokiorg/go-flokicoin/issues
This release introduces significant improvements to fee estimation logic, including atomic persistence to disk, better history tracking, and smarter handling of removal reasons.
Upgrade notes
- Fee estimator snapshots now live on disk as
fee_estimates.datunder the data dir (atomic write/read). Ensure the daemon can write there on shutdown; stale files older than ~60h are ignored on startup.
Highlights
- Fee estimation
- Tracks up to 1008 blocks of history and caps long-horizon queries instead of rejecting them, aligning with Bitcoin Core behavior (
mempool/estimatefee.go). - Records mempool removal reasons so unconfirmed drops (conflicts, reorgs, evictions, rejects) clear estimator state and reduce stale observations (
mempoolplumbing,netsync/manager.go,rpcserver.go). estimatesmartfeeacceptseconomicalorconservativemodes (defaults conservative) and reports invalid modes cleanly; result now includes an expliciterrorsfield (rpcserver.go).
- Tracks up to 1008 blocks of history and caps long-horizon queries instead of rejecting them, aligning with Bitcoin Core behavior (
- Persistence
- Fee estimator state is saved atomically to disk on shutdown and restored on startup when fresh; stale data surfaces an error and is skipped. Helpers and tests cover the binary format and max-age enforcement (
mempool/fee_persist*.go,server.go).
- Fee estimator state is saved atomically to disk on shutdown and restored on startup when fresh; stale data surfaces an error and is skipped. Helpers and tests cover the binary format and max-age enforcement (
Notable commits (planned batch)
- mempool: track removal reasons in fee estimation; widen history to 1008 blocks and cap queries.
- mempool: persist fee estimator to disk with atomic writes and staleness checks.
- Added aggregated nonce signing flow for MuSig2, including updated context handling and comprehensive tests.
Assets
Additional builds (ppc64, mips, *BSD, illumos/solaris) are available on the GitHub release page.