H1: init.cpp now detects crashed migrations (MIGRATION_INCOMPLETE marker)
and retries instead of opening a partial RocksDB. Refuses to start if
the marker persists after migration attempt.
H2: LevelDB ExistsRaw now returns false for keys deleted in the active
batch, matching ReadRaw and the RocksDB backend. Fixes latent
cross-backend consensus split in intra-batch spend checks.
H3: All RocksDB close paths now go through close_rocksdb() which
destroys CF handles before deleting the DB. Fixes RocksDB assertion
/ UB on shutdown and version-reset.
M4: Migration marker write is now flushed + verified (refuses to start
migration if marker can't be written).
CF routing permanently disabled: GetCF() always returns nullptr (default
column family). The read path (NewIterator, LoadBlockIndex) only iterates
the default CF, so writes routed to per-prefix CFs were invisible to scans.
This is why -chaindb=rocksdb compiled clean but was never runtime-valid.
Existing CF-enabled DBs still open (handles retained for cleanup) but no
routing occurs. CF-aware iteration is a future follow-up.
txdb-factory: RocksDB is now the default backend (was still leveldb).
Tests updated for RocksDB-as-default expectations.
From Claude's ROCKSDB-T010-REVIEW-2026-07-01 audit on E:\repos\triangles.
The CF Open overload had the same DB** vs unique_ptr<DB>* API drift
as the non-CF version, but was calling rocksdb::DB::Open directly
instead of through the SFINAE wrapper. On MSYS2 MinGW (Windows CI)
the unique_ptr-only overload causes a compile error. Added
OpenRocksDBCF with the same int/long SFINAE pattern.
BIP152 Compact Blocks (main.cpp, net.cpp, protocol.h):
- SipHash-2-4 short IDs (48-bit) for transaction identification
- Compact block relay with mempool reconstruction
- Merkle root verification before acceptance
- Graceful fallback to full block on any mismatch
- Collision detection for ambiguous short IDs
RocksDB Column Families (txdb-rocksdb.cpp/h):
- 5 CFs: default, blockindex, txindex, utxo, addrindex
- Per-CF tuning: UTXO optimized for point lookups, addrindex for scans
- Backward-compatible: falls back to default CF for pre-migration data
- Prefix-based routing in ReadRaw/WriteRaw/EraseRaw/ExistsRaw
Fork Detector (main.cpp, net.cpp, net.h):
- Background thread checks local tip vs peer median every 60s post-IBD
- Alerts on divergence > forkthreshold (default 5 blocks)
- Optional auto-rebuild trigger on severe divergence
Cross-Network Tor↔I2P Discovery (net.cpp, init.cpp):
- I2P seed addresses loaded into addrman alongside onion seeds
- Address relay bridges .onion and .b32.i2p between networks
- IsI2PAddr/IsOnionAddr helpers for network-type detection
Configurable Outbound Connections (net.cpp, init.cpp):
- -maxoutboundconnections flag (range 4-32, default 8)
Mempool Fee-Priority Boost (miner.cpp):
- 2x fee weight in PoS block assembly for higher staking rewards
SAM v3 Direct Streaming (i2p/i2p_embedded.cpp/h):
- CI2PSamSocket class with full SAM v3 protocol
- SESSION CREATE + STREAM CONNECT handshake
- Factory method on CI2PEmbedded for native I2P connections
- SAM bridge readiness check in bootstrap loop
- txdb-factory.cpp: drop static-cache in ResolveChainDbKind so the
-chaindb flag can be toggled at runtime (needed for tests; cost is
negligible since the daemon sets it once at startup)
- txdb-rocksdb.cpp: fix ExistsRaw to honor pending-batch delete markers.
Previously a key erased inside an open batch was still reported as
existing because the underlying DB hadn't been updated yet. Mirror
ReadRaw's correct behavior: a delete marker shadows the DB value.
- chaindb_runtime_tests.cpp: per-test fresh handle via close-reopen
dance so the static g_rocksdb singleton doesn't leak state between
cases. Tests filter framework keys (length-prefixed 'version' and
'dbformat') from iterator walks. block_index test fixed to Seek()
not Seek("blockindex") since the serialized keys start with the
length byte 0x0a.
- snapshotnet_tests.cpp, chaindb_runtime_tests.cpp: include wallet.h,
ui_interface.h, uint256.h, checkpoints.h as needed for linker; add
BOOST_TEST_MODULE decl; define global stubs (pwalletMain,
uiInterface, fConfChange, etc.) so wallet.cpp link succeeds.
Result: test_snapshotnet + test_chaindb_runtime both pass with zero
errors. Found and fixed a real production bug in ExistsRaw along
the way.
After LoadSnapshot, the daemon has headers + UTXOs but the raw block
bodies haven't been downloaded yet — they'll arrive via P2P as the
node syncs past the snapshot tip. LoadBlockIndex's verification
loop tries to read the last 50 block bodies from disk and fails
with 'OpenBlockFile failed' because the data isn't on disk yet.
Add fLoadedFromSnapshot global, set true at the end of successful
LoadSnapshot. In both txdb-leveldb.cpp and txdb-rocksdb.cpp LoadBlockIndex
verification loops, when ReadFromDisk fails AND fLoadedFromSnapshot is
true, log a warning and continue (the UTXO set itself was already
content-hash verified during LoadSnapshot, so we have strong evidence
the chain state is correct).
For non-snapshot chains (full blk0001.dat downloaded, normal IBD), the
ReadFromDisk failure remains a fatal error as before.
Combined with the prior fix in utxosnapshot.cpp that sets
fSerializeChainTrust=true before writes, the full snapshot path now
works end-to-end on a fresh datadir.
Persisted hashNext can be stale or zeroed by crash-interrupted reorgs, breaking GetKernelStakeModifier()'s forward walk and silently rejecting valid new PoS blocks ('check kernel failed', hashProof=0). On DNS3 every one of 2,201,458 links was zeroed on disk. Root cause of the 2026-04-24 chain halt. Rebuild the in-memory links from pindexBest at every startup.
1. rocksdb::WriteBatch::Handler typeinfo missing on Ubuntu's librocksdb-dev.
Both SecMsgBatchScanner (smessage.cpp) and CRocksBatchScanner
(txdb-rocksdb.cpp) inherited from Handler to scan an active WriteBatch
for pending writes/deletes; that subclass-based scan fails to link
because Ubuntu's package hides the parent's typeinfo. Replaced both
scanners with a parallel std::map<std::string, std::optional<std::string>>
maintained alongside each WriteBatch — Put adds a value entry, Delete
adds a nullopt entry, ScanBatch becomes an O(log n) map lookup. Same
semantics, no Handler dependency.
2. macOS Homebrew's RocksDB 10.x removed the raw DB** overload of
DB::Open; only std::unique_ptr<DB>* remains. txdb-rocksdb.cpp called
the raw form, breaking the macOS build. Added the same SFINAE Open
wrapper used in smessage.cpp (commit 4265343) that picks whichever
overload the linked rocksdb actually has.
3. CSignal<>'s SignalState::slots member collided with Qt's `#define slots`
to empty, stripping the member name in any TU that pulls in <QtCore>
(e.g. moc-generated files that include util_signal.h transitively).
Renamed to slot_map.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
The secure-messaging store (smsgDB) used the LevelDB API directly. Mass-
mapped to the equivalent RocksDB types: leveldb::DB/Status/WriteBatch/
Iterator/Slice/ReadOptions/WriteOptions/WriteBatch::Handler -> rocksdb::*.
The RocksDB API surface for our usage is binary-compatible — pure namespace
substitution, no semantic changes. Consumers in rpcsmessage.cpp and
qt/messagemodel.cpp updated to match.
RocksDB now becomes a hard build dependency (was optional behind
BUILD_ROCKSDB). The chain-DB rocksdb backend is consequently always
available; -chaindb=leveldb remains the default until the Phase-4
LevelDB retirement. Removed the BUILD_ROCKSDB cmake option, the
#ifdef BUILD_ROCKSDB guards in txdb*, and the runtime error path
that triggered when the flag was off.
CI updated: librocksdb-dev (Ubuntu), mingw-w64-x86_64-rocksdb (MSYS2),
and rocksdb (Homebrew) added to all build jobs.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Migration from boost to std-library equivalents and removal of unreachable
code paths. Touches infrastructure only — no consensus rule or wallet
serialization changes.
Dead code removed:
- IRC bootstrap (irc.cpp/h, 417 lines): orphan from pre-Tor era, no callers.
- Alert system (alert.cpp/h + sendalert RPC + Qt UI signal, ~500 lines):
retired post-V5 fork; old peers' alert messages now hit the unknown-cmd
default branch, logged + ignored.
- Legacy P2P handlers in main.cpp: "checkpoint" (already a no-op stub since
V5 fork master-key removal), "checkorder"/"reply" (2010-era Receive-by-IP
feature), plus their unused supporting structures (CRequestTracker,
PushRequest overloads, mapRequests/cs_mapRequests, mapReuseKey).
- Unreachable RPCs clearwallettransactions and scanforalltxns (~175 lines):
defined in rpcwallet.cpp but never registered in the dispatch table.
- Stale -alertnotify CLI help text (option was advertised but never wired).
boost::filesystem -> std::filesystem (C++17):
- 30 source files, 5 headers. namespace fs = boost::filesystem swapped to
namespace fs = std::filesystem; boost::filesystem::ifstream/ofstream
replaced with std::ifstream/ofstream (path-aware in C++17);
fs::system_complete -> fs::absolute; boost::filesystem::filesystem_error
-> std::filesystem::filesystem_error.
- Build system: dropped Boost::filesystem from link libs and Boost
components; PCH includes updated.
- Added explicit <filesystem> includes where types were previously
available only transitively (db.h, rpcblockchain.cpp).
boost::thread -> std::thread (12 files):
- sync.h CCriticalSection/CWaitableCriticalSection now alias
std::recursive_mutex/std::mutex. boost::unique_lock and
boost::condition_variable / boost::mutex::scoped_lock swapped to std
equivalents; sync.cpp boost::thread_specific_ptr -> thread_local
std::unique_ptr.
- init.cpp boost::thread_group rewritten as std::vector<std::thread> with
manual join loop. boost::thread::hardware_concurrency ->
std::thread::hardware_concurrency.
- main.cpp/wallet.cpp -blocknotify/-walletnotify shell-out threads now use
std::thread(...).detach() — fixes a latent bug where modern boost::thread
destructor would call std::terminate on the joinable thread.
- util.cpp NewThread now catches std::system_error.
- No interruption_point/interrupt usage anywhere — pure mechanical swap.
boost::chrono / boost::posix_time -> std::chrono (3 of 5 files):
- util.h: MilliSleep, GetTimeMillis, GetTimeMicros rewritten on std::chrono
(system_clock for epoch math, sleep_for for delays).
- snapshotnet.cpp: sleep_for swapped.
- DoS_tests.cpp: timing harness uses steady_clock.
- Skipped: rpcdump.cpp (boost::posix_time::time_input_facet has no clean
std::get_time equivalent) and qt/qtipcserver.cpp (locked to
boost::posix_time by boost::interprocess::message_queue::timed_receive).
Other housekeeping:
- Dropped unnecessary "using namespace boost;" from txdb-leveldb.cpp,
txdb-rocksdb.cpp, walletdb.cpp, db.cpp (verified no unqualified boost
names in those TUs).
- Removed unused extern declaration for clearwallettransactions.
Build fixes for non-unity builds on MinGW64/GCC 15:
- net.cpp: dropped stale #include "irc.h".
- addrman.cpp + main.cpp: explicit <cmath> include for sqrt/pow (was
arriving transitively via boost headers).
- rpcblockchain.cpp + init.cpp: defensive #undef STRICT/ADVISORY/PERMISSIVE
since windows.h macros collide with the Checkpoints:: enum values when
std headers reorder include flow.
- tor_embed_hooks.cpp: triangles_tor_check_interrupted now polls fShutdown
instead of boost::this_thread::interruption_requested (we never used
boost interruption — the hook was always effectively a no-op).
- snapshotnet.cpp: fs::remove error handle uses std::error_code.
- serialize.h: added <ios> for std::ios::badbit/failbit (was relying on
transitive include via boost).
Note: unity builds currently fail on this branch due to std::byte (C++17)
colliding with COM 'byte' typedef from shlobj.h when 'using namespace std;'
from earlier files in the unity slice leaks into util.cpp's parse of
shlobj.h. Build with -DENABLE_UNITY_BUILD=OFF (the default).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds CRocksTxDB, the second concrete backend for CTxDBBase. Mirrors
CTxDB (LevelDB) one-for-one with rocksdb:: substitutions: same key
serialization (inherited from CTxDBBase), same active-batch semantics,
same LoadBlockIndex flow including the dbformat v3 chain-trust upgrade.
Build flag BUILD_ROCKSDB defaults OFF, so the existing LevelDB build is
untouched — RocksDB headers are only included when the flag is on, and
the entire .cpp file is wrapped in #ifdef BUILD_ROCKSDB.
Build system:
* Top-level option(BUILD_ROCKSDB ... OFF)
* find_package(RocksDB CONFIG) with pkg-config fallback
* Conditional list(APPEND CORE_SOURCES txdb-rocksdb.cpp)
* Conditional target_link_libraries(... RocksDB::rocksdb)
Data layout: RocksDB lives under <datadir>/rocksdb/, separate from
<datadir>/txleveldb/, so both backends can coexist for migration and
parity testing.
Acknowledged debt: LoadBlockIndex is duplicated between CTxDB and
CRocksTxDB. Will be extracted into CTxDBBase once the iterator and
batch abstractions are proven across both backends (M1.4 or later).
Validated: default-OFF build still compiles cleanly. The BUILD_ROCKSDB=ON
path is NOT compile-validated yet — RocksDB isn't installed on this dev
machine. The code is straight namespace substitution from the working
LevelDB backend; whoever first enables the flag should report any
header/API drift between rocksdb releases.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Krystie
sami7777