Java wrapper for Native Comparators

Summary:
This is an abstraction for working with custom Comparators implemented in native C++ code from Java. Native code must directly extend `rocksdb::Comparator`. When the native code comparator is compiled into the RocksDB codebase, you can then create a Java Class, and JNI stub to wrap it.

Useful if the C++/JNI barrier overhead is too much for your applications comparator performance.

An example is provided in `java/rocksjni/native_comparator_wrapper_test.cc` and `java/src/main/java/org/rocksdb/NativeComparatorWrapperTest.java`.
Closes https://github.com/facebook/rocksdb/pull/3334

Differential Revision: D7172605

Pulled By: miasantreble

fbshipit-source-id: e24b7eb267a3bcb6afa214e0379a1d5e8a2ceabe

CMakeLists.txt

@@ -34,15 +34,12 @@
 
 cmake_minimum_required(VERSION 2.8.12)
 project(rocksdb)
-enable_language(CXX)
-enable_language(C)
-enable_language(ASM)
 
 if(POLICY CMP0042)
   cmake_policy(SET CMP0042 NEW)
 endif()
 
-list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_LIST_DIR}/cmake/modules/")
+list(APPEND CMAKE_MODULE_PATH "${CMAKE_SOURCE_DIR}/cmake/modules/")
 
 option(WITH_JEMALLOC "build with JeMalloc" OFF)
 option(WITH_SNAPPY "build with SNAPPY" OFF)
@@ -132,19 +129,20 @@ endif()
 
 string(REGEX REPLACE "[^0-9a-f]+" "" GIT_SHA "${GIT_SHA}")
 
-set(SH_CMD "sh")
-execute_process(COMMAND
-  ${SH_CMD} -c "build_tools/version.sh full"
-  WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
-  OUTPUT_VARIABLE ROCKSDB_VERSION
-)
-string(STRIP "${ROCKSDB_VERSION}" ROCKSDB_VERSION)
-execute_process(COMMAND
-  ${SH_CMD} -c "build_tools/version.sh major"
-  WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
-  OUTPUT_VARIABLE ROCKSDB_VERSION_MAJOR
-)
-string(STRIP "${ROCKSDB_VERSION_MAJOR}" ROCKSDB_VERSION_MAJOR)
+if(NOT WIN32)
+  execute_process(COMMAND
+      "./build_tools/version.sh" "full"
+      WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
+      OUTPUT_VARIABLE ROCKSDB_VERSION
+  )
+  string(STRIP "${ROCKSDB_VERSION}" ROCKSDB_VERSION)
+  execute_process(COMMAND
+      "./build_tools/version.sh" "major"
+      WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
+      OUTPUT_VARIABLE ROCKSDB_VERSION_MAJOR
+  )
+  string(STRIP "${ROCKSDB_VERSION_MAJOR}" ROCKSDB_VERSION_MAJOR)
+endif()
 
 option(WITH_MD_LIBRARY "build with MD" ON)
 if(WIN32 AND MSVC)
@@ -180,18 +178,6 @@ else()
   endif()
 endif()
 
-include(CheckCCompilerFlag)
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
-  CHECK_C_COMPILER_FLAG("-maltivec" HAS_ALTIVEC)
-  if(HAS_ALTIVEC)
-    message(STATUS " HAS_ALTIVEC yes")
-    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -maltivec")
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -maltivec")
-    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mcpu=power8")
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mcpu=power8")
-  endif(HAS_ALTIVEC)
-endif(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
-
 option(PORTABLE "build a portable binary" OFF)
 option(FORCE_SSE42 "force building with SSE4.2, even when PORTABLE=ON" OFF)
 if(PORTABLE)
@@ -204,9 +190,7 @@ else()
   if(MSVC)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
   else()
-    if(NOT HAVE_POWER8)
-      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
-    endif()
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
   endif()
 endif()
 
@@ -602,6 +586,7 @@ set(SOURCES
         utilities/leveldb_options/leveldb_options.cc
         utilities/lua/rocks_lua_compaction_filter.cc
         utilities/memory/memory_util.cc
+        utilities/merge_operators/bytesxor.cc
         utilities/merge_operators/max.cc
         utilities/merge_operators/put.cc
         utilities/merge_operators/string_append/stringappend.cc
@@ -642,12 +627,6 @@ set_source_files_properties(
   endif()
 endif()
 
-if(HAVE_POWER8)
-  list(APPEND SOURCES
-    util/crc32c_ppc.c
-    util/crc32c_ppc_asm.S)
-endif(HAVE_POWER8)
-
 if(WIN32)
   list(APPEND SOURCES
     port/win/io_win.cc
@@ -742,7 +721,7 @@ if(NOT WIN32 OR ROCKSDB_INSTALL_ON_WINDOWS)
   set(package_config_destination ${CMAKE_INSTALL_LIBDIR}/cmake/rocksdb)
 
   configure_package_config_file(
-    ${CMAKE_CURRENT_LIST_DIR}/cmake/RocksDBConfig.cmake.in RocksDBConfig.cmake
+    ${CMAKE_SOURCE_DIR}/cmake/RocksDBConfig.cmake.in RocksDBConfig.cmake
     INSTALL_DESTINATION ${package_config_destination}
   )
 
@@ -972,18 +951,17 @@ if(WITH_TESTS)
 
   foreach(sourcefile ${TEST_EXES})
       get_filename_component(exename ${sourcefile} NAME_WE)
-      add_executable(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} ${sourcefile}
+      add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}
         $<TARGET_OBJECTS:testharness>)
-      set_target_properties(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX}
+      set_target_properties(${exename}${ARTIFACT_SUFFIX}
         PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
         EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
         EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
-        OUTPUT_NAME ${exename}${ARTIFACT_SUFFIX}
         )
-      target_link_libraries(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} gtest ${LIBS})
+      target_link_libraries(${exename}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} gtest ${LIBS})
       if(NOT "${exename}" MATCHES "db_sanity_test")
         add_test(NAME ${exename} COMMAND ${exename}${ARTIFACT_SUFFIX})
-        add_dependencies(check ${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX})
+        add_dependencies(check ${exename}${ARTIFACT_SUFFIX})
       endif()
   endforeach(sourcefile ${TEST_EXES})
 

HISTORY.md

@@ -1,9 +1,27 @@
 # Rocksdb Change Log
-## Unreleased
-### Public API Change
-* Iterator::SeekForPrev is now a pure virtual method. This is to prevent user who implement the Iterator interface fail to implement SeekForPrev by mistake.
+## 5.11.2 (02/24/2018)
 ### Bug Fixes
-* Fix `DisableFileDeletions()` followed by `GetSortedWalFiles()` to not return obsolete WAL files that `PurgeObsoleteFiles()` is going to delete.
+* Fix bug in iterator readahead causing blocks to incorrectly be considered truncated (corrupted).
+
+## 5.11.1 (02/22/2018)
+### New Features
+* Follow rsync-style naming convention for BackupEngine tempfiles. This enables some optimizations when run on GlusterFS.
+
+## 5.11.0 (01/08/2018)
+### Public API Change
+* Add `autoTune` and `getBytesPerSecond()` to RocksJava RateLimiter
+
+### New Features
+* Add a new histogram stat called rocksdb.db.flush.micros for memtable flush.
+* Add "--use_txn" option to use transactional API in db_stress.
+* Disable onboard cache for compaction output in Windows platform.
+* Improve the performance of iterators doing long range scans by using readahead.
+
+### Bug Fixes
+* Fix a stack-use-after-scope bug in ForwardIterator.
+* Fix builds on platforms including Linux, Windows, and PowerPC.
+* Fix buffer overrun in backup engine for DBs with huge number of files.
+* Fix a mislabel bug for bottom-pri compaction threads.
 * Fix DB::Flush() keep waiting after flush finish under certain condition.
 
 ## 5.10.0 (12/11/2017)
@@ -14,6 +32,8 @@
 * Provide lifetime hints when writing files on Linux. This reduces hardware write-amp on storage devices supporting multiple streams.
 * Add a DB stat, `NUMBER_ITER_SKIP`, which returns how many internal keys were skipped during iterations (e.g., due to being tombstones or duplicate versions of a key).
 * Add PerfContext counters, `key_lock_wait_count` and `key_lock_wait_time`, which measure the number of times transactions wait on key locks and total amount of time waiting.
+* Support dynamically changing `ColumnFamilyOptions::compaction_options_universal`.
+* Batch update stats at the end of each `Get`, rather than for each block cache access.
 
 ### Bug Fixes
 * Fix IOError on WAL write doesn't propagate to write group follower

README.md

@@ -1,8 +1,7 @@
 ## RocksDB: A Persistent Key-Value Store for Flash and RAM Storage
 
-[![Linux/Mac Build Status](https://travis-ci.org/facebook/rocksdb.svg?branch=master)](https://travis-ci.org/facebook/rocksdb)
-[![Windows Build status](https://ci.appveyor.com/api/projects/status/fbgfu0so3afcno78/branch/master?svg=true)](https://ci.appveyor.com/project/Facebook/rocksdb/branch/master)
-[![PPC64le Build Status](http://140.211.168.68:8080/buildStatus/icon?job=Rocksdb)](http://140.211.168.68:8080/job/Rocksdb)
+[![Build Status](https://travis-ci.org/facebook/rocksdb.svg?branch=master)](https://travis-ci.org/facebook/rocksdb)
+[![Build status](https://ci.appveyor.com/api/projects/status/fbgfu0so3afcno78/branch/master?svg=true)](https://ci.appveyor.com/project/Facebook/rocksdb/branch/master)
 
 
 RocksDB is developed and maintained by Facebook Database Engineering Team.

TARGETS

@@ -242,6 +242,7 @@ cpp_library(
         "utilities/leveldb_options/leveldb_options.cc",
         "utilities/lua/rocks_lua_compaction_filter.cc",
         "utilities/memory/memory_util.cc",
+        "utilities/merge_operators/bytesxor.cc",
         "utilities/merge_operators/max.cc",
         "utilities/merge_operators/put.cc",
         "utilities/merge_operators/string_append/stringappend.cc",

buckifier/buckify_rocksdb.py

@@ -36,7 +36,7 @@ def parse_src_mk(repo_path):
 # get all .cc / .c files
 def get_cc_files(repo_path):
     cc_files = []
-    for root, dirnames, filenames in os.walk(repo_path):
+    for root, dirnames, filenames in os.walk(repo_path):  # noqa: B007 T25377293 Grandfathered in
         root = root[(len(repo_path) + 1):]
         if "java" in root:
             # Skip java

buckifier/rocks_test_runner.sh

@@ -1,4 +1,5 @@
 #!/usr/bin/env bash
 # Create a tmp directory for the test to use
 TEST_DIR=$(mktemp -d /dev/shm/fbcode_rocksdb_XXXXXXX)
+# shellcheck disable=SC2068
 TEST_TMPDIR="$TEST_DIR" $@ && rm -rf "$TEST_DIR"

build_tools/cont_integration.sh

@@ -13,10 +13,12 @@ error=0
 
 function log {
   DATE=`date +%Y-%m-%d:%H:%M:%S`
+  # shellcheck disable=SC2068
   echo $DATE $@
 }
 
 function log_err {
+  # shellcheck disable=SC2145
   log "ERROR: $@ Error code: $error."
 }
 

build_tools/dependencies.sh

@@ -1,3 +1,4 @@
+# shellcheck disable=SC2148
 GCC_BASE=/mnt/gvfs/third-party2/gcc/8219ec1bcedf8ad9da05e121e193364de2cc4f61/5.x/centos6-native/c447969
 CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/64d8d58e3d84f8bde7a029763d4f5baf39d0d5b9/stable/centos6-native/6aaf4de
 LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/ba9be983c81de7299b59fe71950c664a84dcb5f8/5.x/gcc-5-glibc-2.23/339d858

build_tools/dependencies_4.8.1.sh

@@ -1,3 +1,4 @@
+# shellcheck disable=SC2148
 GCC_BASE=/mnt/gvfs/third-party2/gcc/cf7d14c625ce30bae1a4661c2319c5a283e4dd22/4.8.1/centos6-native/cc6c9dc
 CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/8598c375b0e94e1448182eb3df034704144a838d/stable/centos6-native/3f16ddd
 LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/d6e0a7da6faba45f5e5b1638f9edd7afc2f34e7d/4.8.1/gcc-4.8.1-glibc-2.17/8aac7fc

build_tools/make_package.sh

@@ -1,3 +1,4 @@
+# shellcheck disable=SC1113
 #/usr/bin/env bash
 
 set -e
@@ -28,12 +29,14 @@ function package() {
     if dpkg --get-selections | grep --quiet $1; then
       log "$1 is already installed. skipping."
     else
+      # shellcheck disable=SC2068
       apt-get install $@ -y
     fi
   elif [[ $OS = "centos" ]]; then
     if rpm -qa | grep --quiet $1; then
       log "$1 is already installed. skipping."
     else
+      # shellcheck disable=SC2068
       yum install $@ -y
     fi
   fi
@@ -52,6 +55,7 @@ function gem_install() {
   if gem list | grep --quiet $1; then
     log "$1 is already installed. skipping."
   else
+    # shellcheck disable=SC2068
     gem install $@
   fi
 }
@@ -125,4 +129,5 @@ function main() {
     include $LIB_DIR
 }
 
+# shellcheck disable=SC2068
 main $@

coverage/parse_gcov_output.py

@@ -72,7 +72,7 @@ def display_file_coverage(per_file_coverage, total_coverage):
     header_template = \
         "%" + str(max_file_name_length) + "s\t%s\t%s"
     separator = "-" * (max_file_name_length + 10 + 20)
-    print header_template % ("Filename", "Coverage", "Lines")
+    print header_template % ("Filename", "Coverage", "Lines")  # noqa: E999 T25377293 Grandfathered in
     print separator
 
     # -- Print body

db/column_family.cc

@@ -344,12 +344,13 @@ void SuperVersionUnrefHandle(void* ptr) {
   // When latter happens, we are in ~ColumnFamilyData(), no get should happen as
   // well.
   SuperVersion* sv = static_cast<SuperVersion*>(ptr);
-  if (sv->Unref()) {
-    sv->db_mutex->Lock();
-    sv->Cleanup();
-    sv->db_mutex->Unlock();
-    delete sv;
-  }
+  bool was_last_ref __attribute__((__unused__));
+  was_last_ref = sv->Unref();
+  // Thread-local SuperVersions can't outlive ColumnFamilyData::super_version_.
+  // This is important because we can't do SuperVersion cleanup here.
+  // That would require locking DB mutex, which would deadlock because
+  // SuperVersionUnrefHandle is called with locked ThreadLocalPtr mutex.
+  assert(!was_last_ref);
 }
 }  // anonymous namespace
 
@@ -966,6 +967,12 @@ void ColumnFamilyData::InstallSuperVersion(
       RecalculateWriteStallConditions(mutable_cf_options);
 
   if (old_superversion != nullptr) {
+    // Reset SuperVersions cached in thread local storage.
+    // This should be done before old_superversion->Unref(). That's to ensure
+    // that local_sv_ never holds the last reference to SuperVersion, since
+    // it has no means to safely do SuperVersion cleanup.
+    ResetThreadLocalSuperVersions();
+
     if (old_superversion->mutable_cf_options.write_buffer_size !=
         mutable_cf_options.write_buffer_size) {
       mem_->UpdateWriteBufferSize(mutable_cf_options.write_buffer_size);
@@ -981,9 +988,6 @@ void ColumnFamilyData::InstallSuperVersion(
       sv_context->superversions_to_free.push_back(old_superversion);
     }
   }
-
-  // Reset SuperVersions cached in thread local storage
-  ResetThreadLocalSuperVersions();
 }
 
 void ColumnFamilyData::ResetThreadLocalSuperVersions() {
@@ -995,10 +999,12 @@ void ColumnFamilyData::ResetThreadLocalSuperVersions() {
       continue;
     }
     auto sv = static_cast<SuperVersion*>(ptr);
-    if (sv->Unref()) {
-      sv->Cleanup();
-      delete sv;
-    }
+    bool was_last_ref __attribute__((__unused__));
+    was_last_ref = sv->Unref();
+    // sv couldn't have been the last reference because
+    // ResetThreadLocalSuperVersions() is called before
+    // unref'ing super_version_.
+    assert(!was_last_ref);
   }
 }
 
@@ -1060,12 +1066,10 @@ ColumnFamilySet::~ColumnFamilySet() {
   while (column_family_data_.size() > 0) {
     // cfd destructor will delete itself from column_family_data_
     auto cfd = column_family_data_.begin()->second;
-    bool last_ref __attribute__((__unused__)) = cfd->Unref();
-    assert(last_ref);
+    cfd->Unref();
     delete cfd;
   }
-  bool dummy_last_ref __attribute__((__unused__)) = dummy_cfd_->Unref();
-  assert(dummy_last_ref);
+  dummy_cfd_->Unref();
   delete dummy_cfd_;
 }
 

db/column_family_test.cc

@@ -1868,6 +1868,126 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
   }
 }
 
+// This test checks for automatic getting a conflict if there is a
+// manual which has not yet been scheduled.
+// The manual compaction waits in NotScheduled
+// We generate more files and then trigger an automatic compaction
+// This will wait because there is an unscheduled manual compaction.
+// Once the conflict is hit, the manual compaction starts and ends
+// Then another automatic will start and end.
+TEST_F(ColumnFamilyTest, SameCFManualAutomaticConflict) {
+  Open();
+  CreateColumnFamilies({"one"});
+  ColumnFamilyOptions default_cf, one;
+  db_options_.max_open_files = 20;  // only 10 files in file cache
+  db_options_.max_background_compactions = 3;
+
+  default_cf.compaction_style = kCompactionStyleLevel;
+  default_cf.num_levels = 3;
+  default_cf.write_buffer_size = 64 << 10;  // 64KB
+  default_cf.target_file_size_base = 30 << 10;
+  default_cf.max_compaction_bytes = default_cf.target_file_size_base * 1100;
+  BlockBasedTableOptions table_options;
+  table_options.no_block_cache = true;
+  default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options));
+
+  one.compaction_style = kCompactionStyleUniversal;
+
+  one.num_levels = 1;
+  // trigger compaction if there are >= 4 files
+  one.level0_file_num_compaction_trigger = 4;
+  one.write_buffer_size = 120000;
+
+  Reopen({default_cf, one});
+  // make sure all background compaction jobs can be scheduled
+  auto stop_token =
+      dbfull()->TEST_write_controler().GetCompactionPressureToken();
+
+  // SETUP column family "one" -- universal style
+  for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) {
+    PutRandomData(1, 10, 12000, true);
+    PutRandomData(1, 1, 10, true);
+    WaitForFlush(1);
+    AssertFilesPerLevel(ToString(i + 1), 1);
+  }
+  bool cf_1_1 = true;
+  bool cf_1_2 = true;
+  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::BackgroundCompaction()::Conflict",
+        "ColumnFamilyTest::ManualAutoCon:7"},
+       {"ColumnFamilyTest::ManualAutoCon:9",
+        "ColumnFamilyTest::ManualAutoCon:8"},
+       {"ColumnFamilyTest::ManualAutoCon:2",
+        "ColumnFamilyTest::ManualAutoCon:6"},
+       {"ColumnFamilyTest::ManualAutoCon:4",
+        "ColumnFamilyTest::ManualAutoCon:5"},
+       {"ColumnFamilyTest::ManualAutoCon:1",
+        "ColumnFamilyTest::ManualAutoCon:2"},
+       {"ColumnFamilyTest::ManualAutoCon:1",
+        "ColumnFamilyTest::ManualAutoCon:3"}});
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) {
+        if (cf_1_1) {
+          TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:4");
+          cf_1_1 = false;
+          TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:3");
+        } else if (cf_1_2) {
+          cf_1_2 = false;
+          TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:2");
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::RunManualCompaction:NotScheduled", [&](void* arg) {
+        InstrumentedMutex* mutex = static_cast<InstrumentedMutex*>(arg);
+        mutex->Unlock();
+        TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:9");
+        TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:7");
+        mutex->Lock();
+      });
+
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  rocksdb::port::Thread threads([&] {
+    CompactRangeOptions compact_options;
+    compact_options.exclusive_manual_compaction = false;
+    ASSERT_OK(
+        db_->CompactRange(compact_options, handles_[1], nullptr, nullptr));
+    TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:6");
+  });
+
+  TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:8");
+  WaitForFlush(1);
+
+  // Add more L0 files and force automatic compaction
+  for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) {
+    PutRandomData(1, 10, 12000, true);
+    PutRandomData(1, 1, 10, true);
+    WaitForFlush(1);
+    AssertFilesPerLevel(ToString(one.level0_file_num_compaction_trigger + i),
+                        1);
+  }
+
+  TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:5");
+  // Add more L0 files and force automatic compaction
+  for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) {
+    PutRandomData(1, 10, 12000, true);
+    PutRandomData(1, 1, 10, true);
+    WaitForFlush(1);
+  }
+  TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:1");
+
+  threads.join();
+  WaitForCompaction();
+  // VERIFY compaction "one"
+  ASSERT_LE(NumTableFilesAtLevel(0, 1), 3);
+
+  // Compare against saved keys
+  std::set<std::string>::iterator key_iter = keys_.begin();
+  while (key_iter != keys_.end()) {
+    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
+    key_iter++;
+  }
+}
+
 // In this test, we generate enough files to trigger automatic compactions.
 // The automatic compaction waits in NonTrivial:AfterRun
 // We generate more files and then trigger an automatic compaction

db/compaction.cc

@@ -405,23 +405,20 @@ void Compaction::Summary(char* output, int len) {
 uint64_t Compaction::OutputFilePreallocationSize() const {
   uint64_t preallocation_size = 0;
 
-  for (const auto& level_files : inputs_) {
-    for (const auto& file : level_files.files) {
-      preallocation_size += file->fd.GetFileSize();
+  if (max_output_file_size_ != port::kMaxUint64 &&
+      (cfd_->ioptions()->compaction_style == kCompactionStyleLevel ||
+       output_level() > 0)) {
+    preallocation_size = max_output_file_size_;
+  } else {
+    for (const auto& level_files : inputs_) {
+      for (const auto& file : level_files.files) {
+        preallocation_size += file->fd.GetFileSize();
+      }
     }
   }
-
-  if (max_output_file_size_ != port::kMaxUint64 &&
-      (immutable_cf_options_.compaction_style == kCompactionStyleLevel ||
-       output_level() > 0)) {
-    preallocation_size = std::min(max_output_file_size_, preallocation_size);
-  }
-
   // Over-estimate slightly so we don't end up just barely crossing
   // the threshold
-  // No point to prellocate more than 1GB.
-  return std::min(uint64_t{1073741824},
-                  preallocation_size + (preallocation_size / 10));
+  return preallocation_size + (preallocation_size / 10);
 }
 
 std::unique_ptr<CompactionFilter> Compaction::CreateCompactionFilter() const {

db/compaction_picker_test.cc

@@ -175,8 +175,6 @@ TEST_F(CompactionPickerTest, Level1Trigger) {
 }
 
 TEST_F(CompactionPickerTest, Level1Trigger2) {
-  mutable_cf_options_.target_file_size_base = 10000000000;
-  mutable_cf_options_.RefreshDerivedOptions(ioptions_);
   NewVersionStorage(6, kCompactionStyleLevel);
   Add(1, 66U, "150", "200", 1000000001U);
   Add(1, 88U, "201", "300", 1000000000U);
@@ -193,14 +191,13 @@ TEST_F(CompactionPickerTest, Level1Trigger2) {
   ASSERT_EQ(66U, compaction->input(0, 0)->fd.GetNumber());
   ASSERT_EQ(6U, compaction->input(1, 0)->fd.GetNumber());
   ASSERT_EQ(7U, compaction->input(1, 1)->fd.GetNumber());
-  ASSERT_EQ(uint64_t{1073741824}, compaction->OutputFilePreallocationSize());
 }
 
 TEST_F(CompactionPickerTest, LevelMaxScore) {
   NewVersionStorage(6, kCompactionStyleLevel);
   mutable_cf_options_.target_file_size_base = 10000000;
+  mutable_cf_options_.target_file_size_multiplier = 10;
   mutable_cf_options_.max_bytes_for_level_base = 10 * 1024 * 1024;
-  mutable_cf_options_.RefreshDerivedOptions(ioptions_);
   Add(0, 1U, "150", "200", 1000000U);
   // Level 1 score 1.2
   Add(1, 66U, "150", "200", 6000000U);
@@ -221,9 +218,6 @@ TEST_F(CompactionPickerTest, LevelMaxScore) {
   ASSERT_TRUE(compaction.get() != nullptr);
   ASSERT_EQ(1U, compaction->num_input_files(0));
   ASSERT_EQ(7U, compaction->input(0, 0)->fd.GetNumber());
-  ASSERT_EQ(mutable_cf_options_.target_file_size_base +
-                mutable_cf_options_.target_file_size_base / 10,
-            compaction->OutputFilePreallocationSize());
 }
 
 TEST_F(CompactionPickerTest, NeedsCompactionLevel) {
@@ -527,10 +521,9 @@ TEST_F(CompactionPickerTest, NeedsCompactionFIFO) {
 TEST_F(CompactionPickerTest, CompactionPriMinOverlapping1) {
   NewVersionStorage(6, kCompactionStyleLevel);
   ioptions_.compaction_pri = kMinOverlappingRatio;
-  mutable_cf_options_.target_file_size_base = 100000000000;
+  mutable_cf_options_.target_file_size_base = 10000000;
   mutable_cf_options_.target_file_size_multiplier = 10;
   mutable_cf_options_.max_bytes_for_level_base = 10 * 1024 * 1024;
-  mutable_cf_options_.RefreshDerivedOptions(ioptions_);
 
   Add(2, 6U, "150", "179", 50000000U);
   Add(2, 7U, "180", "220", 50000000U);
@@ -550,8 +543,6 @@ TEST_F(CompactionPickerTest, CompactionPriMinOverlapping1) {
   ASSERT_EQ(1U, compaction->num_input_files(0));
   // Pick file 8 because it overlaps with 0 files on level 3.
   ASSERT_EQ(8U, compaction->input(0, 0)->fd.GetNumber());
-  // Compaction input size * 1.1
-  ASSERT_GE(uint64_t{55000000}, compaction->OutputFilePreallocationSize());
 }
 
 TEST_F(CompactionPickerTest, CompactionPriMinOverlapping2) {

db/db_basic_test.cc

@@ -570,19 +570,19 @@ TEST_F(DBBasicTest, CompactBetweenSnapshots) {
 
 TEST_F(DBBasicTest, DBOpen_Options) {
   Options options = CurrentOptions();
-  Close();
-  Destroy(options);
+  std::string dbname = test::TmpDir(env_) + "/db_options_test";
+  ASSERT_OK(DestroyDB(dbname, options));
 
   // Does not exist, and create_if_missing == false: error
   DB* db = nullptr;
   options.create_if_missing = false;
-  Status s = DB::Open(options, dbname_, &db);
+  Status s = DB::Open(options, dbname, &db);
   ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != nullptr);
   ASSERT_TRUE(db == nullptr);
 
   // Does not exist, and create_if_missing == true: OK
   options.create_if_missing = true;
-  s = DB::Open(options, dbname_, &db);
+  s = DB::Open(options, dbname, &db);
   ASSERT_OK(s);
   ASSERT_TRUE(db != nullptr);
 
@@ -592,14 +592,14 @@ TEST_F(DBBasicTest, DBOpen_Options) {
   // Does exist, and error_if_exists == true: error
   options.create_if_missing = false;
   options.error_if_exists = true;
-  s = DB::Open(options, dbname_, &db);
+  s = DB::Open(options, dbname, &db);
   ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != nullptr);
   ASSERT_TRUE(db == nullptr);
 
   // Does exist, and error_if_exists == false: OK
   options.create_if_missing = true;
   options.error_if_exists = false;
-  s = DB::Open(options, dbname_, &db);
+  s = DB::Open(options, dbname, &db);
   ASSERT_OK(s);
   ASSERT_TRUE(db != nullptr);
 
@@ -847,57 +847,6 @@ TEST_F(DBBasicTest, MmapAndBufferOptions) {
 }
 #endif
 
-class TestEnv : public EnvWrapper {
-  public:
-    explicit TestEnv(Env* base) : EnvWrapper(base) { };
-
-    class TestLogger : public Logger {
-      public:
-        using Logger::Logv;
-        virtual void Logv(const char *format, va_list ap) override { };
-      private:
-        virtual Status CloseImpl() override {
-          return Status::NotSupported();
-        }
-    };
-
-    virtual Status NewLogger(const std::string& fname,
-                             shared_ptr<Logger>* result) {
-      result->reset(new TestLogger());
-      return Status::OK();
-    }
-};
-
-TEST_F(DBBasicTest, DBClose) {
-  Options options = GetDefaultOptions();
-  std::string dbname = test::TmpDir(env_) + "/db_close_test";
-  ASSERT_OK(DestroyDB(dbname, options));
-
-  DB* db = nullptr;
-  options.create_if_missing = true;
-  options.env = new TestEnv(Env::Default());
-  Status s = DB::Open(options, dbname, &db);
-  ASSERT_OK(s);
-  ASSERT_TRUE(db != nullptr);
-
-  s = db->Close();
-  ASSERT_EQ(s, Status::NotSupported());
-
-  delete db;
-
-  // Provide our own logger and ensure DB::Close() does not close it
-  options.info_log.reset(new TestEnv::TestLogger());
-  options.create_if_missing = false;
-  s = DB::Open(options, dbname, &db);
-  ASSERT_OK(s);
-  ASSERT_TRUE(db != nullptr);
-
-  s = db->Close();
-  ASSERT_EQ(s, Status::OK());
-  delete db;
-  delete options.env;
-}
-
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/db_filesnapshot.cc

@@ -57,7 +57,6 @@ Status DBImpl::EnableFileDeletions(bool force) {
       ROCKS_LOG_INFO(immutable_db_options_.info_log, "File Deletions Enabled");
       should_purge_files = true;
       FindObsoleteFiles(&job_context, true);
-      bg_cv_.SignalAll();
     } else {
       ROCKS_LOG_WARN(
           immutable_db_options_.info_log,
@@ -142,18 +141,6 @@ Status DBImpl::GetLiveFiles(std::vector<std::string>& ret,
 }
 
 Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) {
-  {
-    // If caller disabled deletions, this function should return files that are
-    // guaranteed not to be deleted until deletions are re-enabled. We need to
-    // wait for pending purges to finish since WalManager doesn't know which
-    // files are going to be purged. Additional purges won't be scheduled as
-    // long as deletions are disabled (so the below loop must terminate).
-    InstrumentedMutexLock l(&mutex_);
-    while (disable_delete_obsolete_files_ > 0 &&
-           pending_purge_obsolete_files_ > 0) {
-      bg_cv_.Wait();
-    }
-  }
   return wal_manager_.GetSortedWalFiles(files);
 }
 

db/db_flush_test.cc

@@ -137,7 +137,7 @@ TEST_F(DBFlushTest, ManualFlushWithMinWriteBufferNumberToMerge) {
       {{"DBImpl::BGWorkFlush",
         "DBFlushTest::ManualFlushWithMinWriteBufferNumberToMerge:1"},
        {"DBFlushTest::ManualFlushWithMinWriteBufferNumberToMerge:2",
-        "DBImpl::FlushMemTableToOutputFile:BeforeInstallSV"}});
+        "FlushJob::WriteLevel0Table"}});
   SyncPoint::GetInstance()->EnableProcessing();
 
   ASSERT_OK(Put("key1", "value1"));

db/db_impl.cc

@@ -123,13 +123,13 @@ CompressionType GetCompressionFlush(
 namespace {
 void DumpSupportInfo(Logger* logger) {
   ROCKS_LOG_HEADER(logger, "Compression algorithms supported:");
-  for (auto& compression : OptionsHelper::compression_type_string_map) {
-    if (compression.second != kNoCompression &&
-        compression.second != kDisableCompressionOption) {
-      ROCKS_LOG_HEADER(logger, "\t%s supported: %d", compression.first.c_str(),
-                       CompressionTypeSupported(compression.second));
-    }
-  }
+  ROCKS_LOG_HEADER(logger, "\tSnappy supported: %d", Snappy_Supported());
+  ROCKS_LOG_HEADER(logger, "\tZlib supported: %d", Zlib_Supported());
+  ROCKS_LOG_HEADER(logger, "\tBzip supported: %d", BZip2_Supported());
+  ROCKS_LOG_HEADER(logger, "\tLZ4 supported: %d", LZ4_Supported());
+  ROCKS_LOG_HEADER(logger, "\tZSTDNotFinal supported: %d",
+                   ZSTDNotFinal_Supported());
+  ROCKS_LOG_HEADER(logger, "\tZSTD supported: %d", ZSTD_Supported());
   ROCKS_LOG_HEADER(logger, "Fast CRC32 supported: %s",
                    crc32c::IsFastCrc32Supported().c_str());
 }
@@ -141,7 +141,6 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
                const bool seq_per_batch)
     : env_(options.env),
       dbname_(dbname),
-      own_info_log_(options.info_log == nullptr),
       initial_db_options_(SanitizeOptions(dbname, options)),
       immutable_db_options_(initial_db_options_),
       mutable_db_options_(initial_db_options_),
@@ -178,7 +177,6 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
       num_running_flushes_(0),
       bg_purge_scheduled_(0),
       disable_delete_obsolete_files_(0),
-      pending_purge_obsolete_files_(0),
       delete_obsolete_files_last_run_(env_->NowMicros()),
       last_stats_dump_time_microsec_(0),
       next_job_id_(1),
@@ -214,8 +212,7 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
       // requires a custom gc for compaction, we use that to set use_custom_gc_
       // as well.
       use_custom_gc_(seq_per_batch),
-      preserve_deletes_(options.preserve_deletes),
-      closed_(false) {
+      preserve_deletes_(options.preserve_deletes) {
   env_->GetAbsolutePath(dbname, &db_absolute_path_);
 
   // Reserve ten files or so for other uses and give the rest to TableCache.
@@ -278,7 +275,7 @@ void DBImpl::CancelAllBackgroundWork(bool wait) {
   }
 }
 
-Status DBImpl::CloseImpl() {
+DBImpl::~DBImpl() {
   // CancelAllBackgroundWork called with false means we just set the shutdown
   // marker. After this we do a variant of the waiting and unschedule work
   // (to consider: moving all the waiting into CancelAllBackgroundWork(true))
@@ -294,8 +291,7 @@ Status DBImpl::CloseImpl() {
 
   // Wait for background work to finish
   while (bg_bottom_compaction_scheduled_ || bg_compaction_scheduled_ ||
-         bg_flush_scheduled_ || bg_purge_scheduled_ ||
-         pending_purge_obsolete_files_) {
+         bg_flush_scheduled_ || bg_purge_scheduled_) {
     TEST_SYNC_POINT("DBImpl::~DBImpl:WaitJob");
     bg_cv_.Wait();
   }
@@ -382,16 +378,8 @@ Status DBImpl::CloseImpl() {
 
   ROCKS_LOG_INFO(immutable_db_options_.info_log, "Shutdown complete");
   LogFlush(immutable_db_options_.info_log);
-
-  Status s = Status::OK();
-  if (immutable_db_options_.info_log && own_info_log_) {
-    s = immutable_db_options_.info_log->Close();
-  }
-  return s;
 }
 
-DBImpl::~DBImpl() { Close(); }
-
 void DBImpl::MaybeIgnoreError(Status* s) const {
   if (s->ok() || immutable_db_options_.paranoid_checks) {
     // No change needed
@@ -2059,7 +2047,8 @@ Status DBImpl::DeleteFile(std::string name) {
                       name.c_str());
       return Status::NotSupported("Delete only supported for archived logs");
     }
-    status = wal_manager_.DeleteFile(name, number);
+    status =
+        env_->DeleteFile(immutable_db_options_.wal_dir + "/" + name.c_str());
     if (!status.ok()) {
       ROCKS_LOG_ERROR(immutable_db_options_.info_log,
                       "DeleteFile %s failed -- %s.\n", name.c_str(),
@@ -2332,15 +2321,7 @@ Status DB::DestroyColumnFamilyHandle(ColumnFamilyHandle* column_family) {
   return Status::OK();
 }
 
-DB::~DB() {}
-
-Status DBImpl::Close() {
-  if (!closed_) {
-    closed_ = true;
-    return CloseImpl();
-  }
-  return Status::OK();
-}
+DB::~DB() { }
 
 Status DB::ListColumnFamilies(const DBOptions& db_options,
                               const std::string& name,

db/db_impl.h

@@ -468,8 +468,6 @@ class DBImpl : public DB {
   // belong to live files are posibly removed. Also, removes all the
   // files in sst_delete_files and log_delete_files.
   // It is not necessary to hold the mutex when invoking this method.
-  // If FindObsoleteFiles() was run, we need to also run
-  // PurgeObsoleteFiles(), even if disable_delete_obsolete_files_ is true
   void PurgeObsoleteFiles(const JobContext& background_contet,
                           bool schedule_only = false);
 
@@ -613,14 +611,10 @@ class DBImpl : public DB {
                      std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
                      const bool seq_per_batch);
 
-  virtual Status Close() override;
-
  protected:
   Env* const env_;
   const std::string dbname_;
   unique_ptr<VersionSet> versions_;
-  // Flag to check whether we allocated and own the info log file
-  bool own_info_log_;
   const DBOptions initial_db_options_;
   const ImmutableDBOptions immutable_db_options_;
   MutableDBOptions mutable_db_options_;
@@ -922,9 +916,6 @@ class DBImpl : public DB {
 
   uint64_t GetMaxTotalWalSize() const;
 
-  // Actual implementation of Close()
-  virtual Status CloseImpl();
-
   // table_cache_ provides its own synchronization
   std::shared_ptr<Cache> table_cache_;
 
@@ -953,8 +944,6 @@ class DBImpl : public DB {
   // (i.e. whenever a flush is done, even if it didn't make any progress)
   // * whenever there is an error in background purge, flush or compaction
   // * whenever num_running_ingest_file_ goes to 0.
-  // * whenever pending_purge_obsolete_files_ goes to 0.
-  // * whenever disable_delete_obsolete_files_ goes to 0.
   InstrumentedCondVar bg_cv_;
   // Writes are protected by locking both mutex_ and log_write_mutex_, and reads
   // must be under either mutex_ or log_write_mutex_. Since after ::Open,
@@ -1221,10 +1210,6 @@ class DBImpl : public DB {
   // without any synchronization
   int disable_delete_obsolete_files_;
 
-  // Number of times FindObsoleteFiles has found deletable files and the
-  // corresponding call to PurgeObsoleteFiles has not yet finished.
-  int pending_purge_obsolete_files_;
-
   // last time when DeleteObsoleteFiles with full scan was executed. Originaly
   // initialized with startup time.
   uint64_t delete_obsolete_files_last_run_;
@@ -1378,9 +1363,6 @@ class DBImpl : public DB {
   // is set to false.
   std::atomic<SequenceNumber> preserve_deletes_seqnum_;
   const bool preserve_deletes_;
-
-  // Flag to check whether Close() has been called on this DB
-  bool closed_;
 };
 
 extern Options SanitizeOptions(const std::string& db,

db/db_impl_compaction_flush.cc

@@ -134,7 +134,6 @@ Status DBImpl::FlushMemTableToOutputFile(
   }
 
   if (s.ok()) {
-    TEST_SYNC_POINT("DBImpl::FlushMemTableToOutputFile:BeforeInstallSV");
     InstallSuperVersionAndScheduleWork(cfd, &job_context->superversion_context,
                                        mutable_cf_options);
     if (made_progress) {
@@ -442,8 +441,7 @@ Status DBImpl::CompactFiles(
   }  // release the mutex
 
   // delete unnecessary files if any, this is done outside the mutex
-  if (job_context.HaveSomethingToClean() ||
-      job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+  if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
     // Have to flush the info logs before bg_compaction_scheduled_--
     // because if bg_flush_scheduled_ becomes 0 and the lock is
     // released, the deconstructor of DB can kick in and destroy all the
@@ -1072,7 +1070,6 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
   if (HasExclusiveManualCompaction()) {
     // only manual compactions are allowed to run. don't schedule automatic
     // compactions
-    TEST_SYNC_POINT("DBImpl::MaybeScheduleFlushOrCompaction:Conflict");
     return;
   }
 
@@ -1317,10 +1314,8 @@ void DBImpl::BackgroundCallFlush() {
     // created. Thus, we force full scan in FindObsoleteFiles()
     FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress());
     // delete unnecessary files if any, this is done outside the mutex
-    if (job_context.HaveSomethingToClean() ||
-        job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+    if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
       mutex_.Unlock();
-      TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:FilesFound");
       // Have to flush the info logs before bg_flush_scheduled_--
       // because if bg_flush_scheduled_ becomes 0 and the lock is
       // released, the deconstructor of DB can kick in and destroy all the
@@ -1400,8 +1395,7 @@ void DBImpl::BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
     FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress());
 
     // delete unnecessary files if any, this is done outside the mutex
-    if (job_context.HaveSomethingToClean() ||
-        job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+    if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
       mutex_.Unlock();
       // Have to flush the info logs before bg_compaction_scheduled_--
       // because if bg_flush_scheduled_ becomes 0 and the lock is
@@ -1520,7 +1514,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
                : m->manual_end->DebugString().c_str()));
     }
   } else if (!is_prepicked && !compaction_queue_.empty()) {
-    if (HasExclusiveManualCompaction()) {
+    if (HaveManualCompaction(compaction_queue_.front())) {
       // Can't compact right now, but try again later
       TEST_SYNC_POINT("DBImpl::BackgroundCompaction()::Conflict");
 

db/db_impl_files.cc

@@ -285,9 +285,6 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
   job_context->logs_to_free = logs_to_free_;
   job_context->log_recycle_files.assign(log_recycle_files_.begin(),
                                         log_recycle_files_.end());
-  if (job_context->HaveSomethingToDelete()) {
-    ++pending_purge_obsolete_files_;
-  }
   logs_to_free_.clear();
 }
 
@@ -345,12 +342,15 @@ void DBImpl::DeleteObsoleteFileImpl(int job_id, const std::string& fname,
 // files in sst_delete_files and log_delete_files.
 // It is not necessary to hold the mutex when invoking this method.
 void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
-  TEST_SYNC_POINT("DBImpl::PurgeObsoleteFiles:Begin");
   // we'd better have sth to delete
   assert(state.HaveSomethingToDelete());
 
-  // FindObsoleteFiles() should've populated this so nonzero
-  assert(state.manifest_file_number != 0);
+  // this checks if FindObsoleteFiles() was run before. If not, don't do
+  // PurgeObsoleteFiles(). If FindObsoleteFiles() was run, we need to also
+  // run PurgeObsoleteFiles(), even if disable_delete_obsolete_files_ is true
+  if (state.manifest_file_number == 0) {
+    return;
+  }
 
   // Now, convert live list to an unordered map, WITHOUT mutex held;
   // set is slow.
@@ -533,13 +533,6 @@ void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
   wal_manager_.PurgeObsoleteWALFiles();
 #endif  // ROCKSDB_LITE
   LogFlush(immutable_db_options_.info_log);
-  InstrumentedMutexLock l(&mutex_);
-  --pending_purge_obsolete_files_;
-  assert(pending_purge_obsolete_files_ >= 0);
-  if (pending_purge_obsolete_files_ == 0) {
-    bg_cv_.SignalAll();
-  }
-  TEST_SYNC_POINT("DBImpl::PurgeObsoleteFiles:End");
 }
 
 void DBImpl::DeleteObsoleteFiles() {

db/db_iter.cc

@@ -217,6 +217,9 @@ class DBIter final: public Iterator {
         *prop = "Iterator is not valid.";
       }
       return Status::OK();
+    } else if (prop_name == "rocksdb.iterator.internal-key") {
+      *prop = saved_key_.GetUserKey().ToString();
+      return Status::OK();
     }
     return Status::InvalidArgument("Undentified property.");
   }
@@ -972,17 +975,6 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
   // assume there is at least one parseable key for this user key
   ParsedInternalKey ikey;
   FindParseableKey(&ikey, kForward);
-  assert(iter_->Valid());
-  assert(user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey()));
-
-  // In case read_callback presents, the value we seek to may not be visible.
-  // Seek for the next value that's visible.
-  while (!IsVisible(ikey.sequence)) {
-    iter_->Next();
-    FindParseableKey(&ikey, kForward);
-    assert(iter_->Valid());
-    assert(user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey()));
-  }
 
   if (ikey.type == kTypeDeletion || ikey.type == kTypeSingleDeletion ||
       range_del_agg_.ShouldDelete(

db/db_iterator_test.cc

@@ -9,7 +9,6 @@
 
 #include <functional>
 
-#include "db/db_iter.h"
 #include "db/db_test_util.h"
 #include "port/port.h"
 #include "port/stack_trace.h"
@@ -56,6 +55,7 @@ TEST_F(DBIteratorTest, IteratorProperty) {
   Options options = CurrentOptions();
   CreateAndReopenWithCF({"pikachu"}, options);
   Put(1, "1", "2");
+  Delete(1, "2");
   ReadOptions ropt;
   ropt.pin_data = false;
   {
@@ -65,9 +65,15 @@ TEST_F(DBIteratorTest, IteratorProperty) {
     ASSERT_NOK(iter->GetProperty("non_existing.value", &prop_value));
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("0", prop_value);
+    ASSERT_OK(iter->GetProperty("rocksdb.iterator.internal-key", &prop_value));
+    ASSERT_EQ("1", prop_value);
     iter->Next();
     ASSERT_OK(iter->GetProperty("rocksdb.iterator.is-key-pinned", &prop_value));
     ASSERT_EQ("Iterator is not valid.", prop_value);
+
+    // Get internal key at which the iteration stopped (tombstone in this case).
+    ASSERT_OK(iter->GetProperty("rocksdb.iterator.internal-key", &prop_value));
+    ASSERT_EQ("2", prop_value);
   }
   Close();
 }
@@ -2158,139 +2164,46 @@ TEST_F(DBIteratorTest, SkipStatistics) {
   ASSERT_EQ(skip_count, TestGetTickerCount(options, NUMBER_ITER_SKIP));
 }
 
-TEST_F(DBIteratorTest, ReadCallback) {
-  class TestReadCallback : public ReadCallback {
-   public:
-    explicit TestReadCallback(SequenceNumber last_visible_seq)
-        : last_visible_seq_(last_visible_seq) {}
+TEST_F(DBIteratorTest, SeekAfterHittingManyInternalKeys) {
+  Options options = CurrentOptions();
+  DestroyAndReopen(options);
+  ReadOptions ropts;
+  ropts.max_skippable_internal_keys = 2;
 
-    bool IsCommitted(SequenceNumber seq) override {
-      return seq <= last_visible_seq_;
-    }
+  Put("1", "val_1");
+  // Add more tombstones than max_skippable_internal_keys so that Next() fails.
+  Delete("2");
+  Delete("3");
+  Delete("4");
+  Delete("5");
+  Put("6", "val_6");
 
-   private:
-    SequenceNumber last_visible_seq_;
-  };
+  unique_ptr<Iterator> iter(db_->NewIterator(ropts));
+  iter->SeekToFirst();
 
-  ASSERT_OK(Put("foo", "v1"));
-  ASSERT_OK(Put("foo", "v2"));
-  ASSERT_OK(Put("foo", "v3"));
-  ASSERT_OK(Put("a", "va"));
-  ASSERT_OK(Put("z", "vz"));
-  SequenceNumber seq1 = db_->GetLatestSequenceNumber();
-  TestReadCallback callback1(seq1);
-  ASSERT_OK(Put("foo", "v4"));
-  ASSERT_OK(Put("foo", "v5"));
-  ASSERT_OK(Put("bar", "v7"));
-
-  SequenceNumber seq2 = db_->GetLatestSequenceNumber();
-  auto* cfd =
-      reinterpret_cast<ColumnFamilyHandleImpl*>(db_->DefaultColumnFamily())
-          ->cfd();
-  // The iterator are suppose to see data before seq1.
-  Iterator* iter =
-      dbfull()->NewIteratorImpl(ReadOptions(), cfd, seq2, &callback1);
-
-  // Seek
-  // The latest value of "foo" before seq1 is "v3"
-  iter->Seek("foo");
   ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("foo", iter->key());
-  ASSERT_EQ("v3", iter->value());
-  // "bar" is not visible to the iterator. It will move on to the next key
-  // "foo".
-  iter->Seek("bar");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("foo", iter->key());
-  ASSERT_EQ("v3", iter->value());
+  ASSERT_EQ(iter->key().ToString(), "1");
+  ASSERT_EQ(iter->value().ToString(), "val_1");
 
-  // Next
-  // Seek to "a"
-  iter->Seek("a");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("va", iter->value());
-  // "bar" is not visible to the iterator. It will move on to the next key
-  // "foo".
+  // This should fail as incomplete due to too many non-visible internal keys on
+  // the way to the next valid user key.
   iter->Next();
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("foo", iter->key());
-  ASSERT_EQ("v3", iter->value());
+  ASSERT_TRUE(!iter->Valid());
+  ASSERT_TRUE(iter->status().IsIncomplete());
 
-  // Prev
-  // Seek to "z"
-  iter->Seek("z");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("vz", iter->value());
-  // The previous key is "foo", which is visible to the iterator.
-  iter->Prev();
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("foo", iter->key());
-  ASSERT_EQ("v3", iter->value());
-  // "bar" is not visible to the iterator. It will move on to the next key "a".
-  iter->Prev();  // skipping "bar"
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("a", iter->key());
-  ASSERT_EQ("va", iter->value());
+  // Get the internal key at which Next() failed.
+  std::string prop_value;
+  ASSERT_OK(iter->GetProperty("rocksdb.iterator.internal-key", &prop_value));
+  ASSERT_EQ("4", prop_value);
 
-  // SeekForPrev
-  // The previous key is "foo", which is visible to the iterator.
-  iter->SeekForPrev("y");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("foo", iter->key());
-  ASSERT_EQ("v3", iter->value());
-  // "bar" is not visible to the iterator. It will move on to the next key "a".
-  iter->SeekForPrev("bar");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("a", iter->key());
-  ASSERT_EQ("va", iter->value());
+  // Create a new iterator to seek to the internal key.
+  unique_ptr<Iterator> iter2(db_->NewIterator(ropts));
+  iter2->Seek(prop_value);
+  ASSERT_TRUE(iter2->Valid());
+  ASSERT_OK(iter2->status());
 
-  delete iter;
-
-  // Prev beyond max_sequential_skip_in_iterations
-  uint64_t num_versions =
-      CurrentOptions().max_sequential_skip_in_iterations + 10;
-  for (uint64_t i = 0; i < num_versions; i++) {
-    ASSERT_OK(Put("bar", ToString(i)));
-  }
-  SequenceNumber seq3 = db_->GetLatestSequenceNumber();
-  TestReadCallback callback2(seq3);
-  ASSERT_OK(Put("bar", "v8"));
-  SequenceNumber seq4 = db_->GetLatestSequenceNumber();
-
-  // The iterator is suppose to see data before seq3.
-  iter = dbfull()->NewIteratorImpl(ReadOptions(), cfd, seq4, &callback2);
-  // Seek to "z", which is visible.
-  iter->Seek("z");
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("vz", iter->value());
-  // Previous key is "foo" and the last value "v5" is visible.
-  iter->Prev();
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("foo", iter->key());
-  ASSERT_EQ("v5", iter->value());
-  // Since the number of values of "bar" is more than
-  // max_sequential_skip_in_iterations, Prev() will ultimately fallback to
-  // seek in forward direction. Here we test the fallback seek is correct.
-  // The last visible value should be (num_versions - 1), as "v8" is not
-  // visible.
-  iter->Prev();
-  ASSERT_TRUE(iter->Valid());
-  ASSERT_OK(iter->status());
-  ASSERT_EQ("bar", iter->key());
-  ASSERT_EQ(ToString(num_versions - 1), iter->value());
-
-  delete iter;
+  ASSERT_EQ(iter2->key().ToString(), "6");
+  ASSERT_EQ(iter2->value().ToString(), "val_6");
 }
 
 }  // namespace rocksdb

db/db_options_test.cc

@@ -518,13 +518,8 @@ static void assert_candidate_files_empty(DBImpl* dbfull, const bool empty) {
   JobContext job_context(0);
   dbfull->FindObsoleteFiles(&job_context, false);
   ASSERT_EQ(empty, job_context.full_scan_candidate_files.empty());
-  dbfull->TEST_UnlockMutex();
-  if (job_context.HaveSomethingToDelete()) {
-    // fulfill the contract of FindObsoleteFiles by calling PurgeObsoleteFiles
-    // afterwards; otherwise the test may hang on shutdown
-    dbfull->PurgeObsoleteFiles(job_context);
-  }
   job_context.Clean();
+  dbfull->TEST_UnlockMutex();
 }
 
 TEST_F(DBOptionsTest, DeleteObsoleteFilesPeriodChange) {

db/db_test.cc

@@ -503,9 +503,7 @@ TEST_F(DBTest, DISABLED_VeryLargeValue) {
   ASSERT_OK(Put(key2, raw));
   dbfull()->TEST_WaitForFlushMemTable();
 
-#ifndef ROCKSDB_LITE
   ASSERT_EQ(1, NumTableFilesAtLevel(0));
-#endif  // !ROCKSDB_LITE
 
   std::string value;
   Status s = db_->Get(ReadOptions(), key1, &value);
@@ -2193,8 +2191,6 @@ class ModelDB : public DB {
     batch.Put(cf, k, v);
     return Write(o, &batch);
   }
-  using DB::Close;
-  virtual Status Close() override { return Status::OK(); }
   using DB::Delete;
   virtual Status Delete(const WriteOptions& o, ColumnFamilyHandle* cf,
                         const Slice& key) override {
@@ -5058,84 +5054,55 @@ TEST_F(DBTest, PromoteL0Failure) {
   status = experimental::PromoteL0(db_, db_->DefaultColumnFamily());
   ASSERT_TRUE(status.IsInvalidArgument());
 }
-
-// Github issue #596
-TEST_F(DBTest, CompactRangeWithEmptyBottomLevel) {
-  const int kNumLevels = 2;
-  const int kNumL0Files = 2;
-  Options options = CurrentOptions();
-  options.disable_auto_compactions = true;
-  options.num_levels = kNumLevels;
-  DestroyAndReopen(options);
-
-  Random rnd(301);
-  for (int i = 0; i < kNumL0Files; ++i) {
-    ASSERT_OK(Put(Key(0), RandomString(&rnd, 1024)));
-    Flush();
-  }
-  ASSERT_EQ(NumTableFilesAtLevel(0), kNumL0Files);
-  ASSERT_EQ(NumTableFilesAtLevel(1), 0);
-
-  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-  ASSERT_EQ(NumTableFilesAtLevel(0), 0);
-  ASSERT_EQ(NumTableFilesAtLevel(1), kNumL0Files);
-}
 #endif  // ROCKSDB_LITE
 
-TEST_F(DBTest, AutomaticConflictsWithManualCompaction) {
-  const int kNumL0Files = 50;
+// Github issue #596
+TEST_F(DBTest, HugeNumberOfLevels) {
   Options options = CurrentOptions();
-  options.level0_file_num_compaction_trigger = 4;
-  // never slowdown / stop
-  options.level0_slowdown_writes_trigger = 999999;
-  options.level0_stop_writes_trigger = 999999;
+  options.write_buffer_size = 2 * 1024 * 1024;         // 2MB
+  options.max_bytes_for_level_base = 2 * 1024 * 1024;  // 2MB
+  options.num_levels = 12;
   options.max_background_compactions = 10;
+  options.max_bytes_for_level_multiplier = 2;
+  options.level_compaction_dynamic_level_bytes = true;
   DestroyAndReopen(options);
 
-  // schedule automatic compactions after the manual one starts, but before it
-  // finishes to ensure conflict.
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
-      {{"DBImpl::BackgroundCompaction:Start",
-        "DBTest::AutomaticConflictsWithManualCompaction:PrePuts"},
-       {"DBTest::AutomaticConflictsWithManualCompaction:PostPuts",
-        "DBImpl::BackgroundCompaction:NonTrivial:AfterRun"}});
-  std::atomic<int> callback_count(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::MaybeScheduleFlushOrCompaction:Conflict",
-      [&](void* arg) { callback_count.fetch_add(1); });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  Random rnd(301);
+  for (int i = 0; i < 300000; ++i) {
+    ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024)));
+  }
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+}
+
+TEST_F(DBTest, AutomaticConflictsWithManualCompaction) {
+  Options options = CurrentOptions();
+  options.write_buffer_size = 2 * 1024 * 1024;         // 2MB
+  options.max_bytes_for_level_base = 2 * 1024 * 1024;  // 2MB
+  options.num_levels = 12;
+  options.max_background_compactions = 10;
+  options.max_bytes_for_level_multiplier = 2;
+  options.level_compaction_dynamic_level_bytes = true;
+  DestroyAndReopen(options);
 
   Random rnd(301);
-  for (int i = 0; i < 2; ++i) {
-    // put two keys to ensure no trivial move
-    for (int j = 0; j < 2; ++j) {
-      ASSERT_OK(Put(Key(j), RandomString(&rnd, 1024)));
-    }
-    ASSERT_OK(Flush());
+  for (int i = 0; i < 300000; ++i) {
+    ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024)));
   }
-  std::thread manual_compaction_thread([this]() {
-      CompactRangeOptions croptions;
-      croptions.exclusive_manual_compaction = true;
-      ASSERT_OK(db_->CompactRange(croptions, nullptr, nullptr));
-    });
-
-  TEST_SYNC_POINT("DBTest::AutomaticConflictsWithManualCompaction:PrePuts");
-  for (int i = 0; i < kNumL0Files; ++i) {
-    // put two keys to ensure no trivial move
-    for (int j = 0; j < 2; ++j) {
-      ASSERT_OK(Put(Key(j), RandomString(&rnd, 1024)));
-    }
-    ASSERT_OK(Flush());
-  }
-  TEST_SYNC_POINT("DBTest::AutomaticConflictsWithManualCompaction:PostPuts");
 
+  std::atomic<int> callback_count(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::BackgroundCompaction()::Conflict",
+      [&](void* arg) { callback_count.fetch_add(1); });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  CompactRangeOptions croptions;
+  croptions.exclusive_manual_compaction = false;
+  ASSERT_OK(db_->CompactRange(croptions, nullptr, nullptr));
   ASSERT_GE(callback_count.load(), 1);
-  for (int i = 0; i < 2; ++i) {
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  for (int i = 0; i < 300000; ++i) {
     ASSERT_NE("NOT_FOUND", Get(Key(i)));
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  manual_compaction_thread.join();
-  dbfull()->TEST_WaitForCompact();
 }
 
 // Github issue #595
@@ -5355,15 +5322,12 @@ class WriteStallListener : public EventListener {
  public:
   WriteStallListener() : condition_(WriteStallCondition::kNormal) {}
   void OnStallConditionsChanged(const WriteStallInfo& info) override {
-    MutexLock l(&mutex_);
     condition_ = info.condition.cur;
   }
   bool CheckCondition(WriteStallCondition expected) {
-    MutexLock l(&mutex_);
     return expected == condition_;
   }
  private:
-  port::Mutex mutex_;
   WriteStallCondition condition_;
 };
 
@@ -5692,6 +5656,50 @@ TEST_F(DBTest, PauseBackgroundWorkTest) {
   // now it's done
   ASSERT_TRUE(done.load());
 }
+
+// Keep spawning short-living threads that create an iterator and quit.
+// Meanwhile in another thread keep flushing memtables.
+// This used to cause a deadlock.
+TEST_F(DBTest, ThreadLocalPtrDeadlock) {
+  std::atomic<int> flushes_done{0};
+  std::atomic<int> threads_destroyed{0};
+  auto done = [&] {
+    return flushes_done.load() > 10;
+  };
+
+  std::thread flushing_thread([&] {
+    for (int i = 0; !done(); ++i) {
+      ASSERT_OK(db_->Put(WriteOptions(), Slice("hi"),
+                         Slice(std::to_string(i).c_str())));
+      ASSERT_OK(db_->Flush(FlushOptions()));
+      int cnt = ++flushes_done;
+      fprintf(stderr, "Flushed %d times\n", cnt);
+    }
+  });
+
+  std::vector<std::thread> thread_spawning_threads(10);
+  for (auto& t: thread_spawning_threads) {
+    t = std::thread([&] {
+      while (!done()) {
+        {
+          std::thread tmp_thread([&] {
+            auto it = db_->NewIterator(ReadOptions());
+            delete it;
+          });
+          tmp_thread.join();
+        }
+        ++threads_destroyed;
+      }
+    });
+  }
+
+  for (auto& t: thread_spawning_threads) {
+    t.join();
+  }
+  flushing_thread.join();
+  fprintf(stderr, "Done. Flushed %d times, destroyed %d threads\n",
+          flushes_done.load(), threads_destroyed.load());
+}
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/db_test2.cc

@@ -2430,60 +2430,6 @@ TEST_F(DBTest2, ReadCallbackTest) {
   }
 }
 
-#ifndef ROCKSDB_LITE
-
-TEST_F(DBTest2, LiveFilesOmitObsoleteFiles) {
-  // Regression test for race condition where an obsolete file is returned to
-  // user as a "live file" but then deleted, all while file deletions are
-  // disabled.
-  //
-  // It happened like this:
-  //
-  // 1. [flush thread] Log file "x.log" found by FindObsoleteFiles
-  // 2. [user thread] DisableFileDeletions, GetSortedWalFiles are called and the
-  //    latter returned "x.log"
-  // 3. [flush thread] PurgeObsoleteFiles deleted "x.log"
-  // 4. [user thread] Reading "x.log" failed
-  //
-  // Unfortunately the only regression test I can come up with involves sleep.
-  // We cannot set SyncPoints to repro since, once the fix is applied, the
-  // SyncPoints would cause a deadlock as the repro's sequence of events is now
-  // prohibited.
-  //
-  // Instead, if we sleep for a second between Find and Purge, and ensure the
-  // read attempt happens after purge, then the sequence of events will almost
-  // certainly happen on the old code.
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
-      {"DBImpl::BackgroundCallFlush:FilesFound",
-       "DBTest2::LiveFilesOmitObsoleteFiles:FlushTriggered"},
-      {"DBImpl::PurgeObsoleteFiles:End",
-       "DBTest2::LiveFilesOmitObsoleteFiles:LiveFilesCaptured"},
-  });
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::PurgeObsoleteFiles:Begin",
-      [&](void* arg) { env_->SleepForMicroseconds(1000000); });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
-
-  Put("key", "val");
-  FlushOptions flush_opts;
-  flush_opts.wait = false;
-  db_->Flush(flush_opts);
-  TEST_SYNC_POINT("DBTest2::LiveFilesOmitObsoleteFiles:FlushTriggered");
-
-  db_->DisableFileDeletions();
-  VectorLogPtr log_files;
-  db_->GetSortedWalFiles(log_files);
-  TEST_SYNC_POINT("DBTest2::LiveFilesOmitObsoleteFiles:LiveFilesCaptured");
-  for (const auto& log_file : log_files) {
-    ASSERT_OK(env_->FileExists(LogFileName(dbname_, log_file->LogNumber())));
-  }
-
-  db_->EnableFileDeletions();
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-}
-
-#endif  // ROCKSDB_LITE
-
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/db_test_util.cc

@@ -867,6 +867,7 @@ size_t DBTestBase::CountLiveFiles() {
   db_->GetLiveFilesMetaData(&metadata);
   return metadata.size();
 }
+#endif  // ROCKSDB_LITE
 
 int DBTestBase::NumTableFilesAtLevel(int level, int cf) {
   std::string property;
@@ -927,7 +928,6 @@ std::string DBTestBase::FilesPerLevel(int cf) {
   result.resize(last_non_zero_offset);
   return result;
 }
-#endif  // !ROCKSDB_LITE
 
 size_t DBTestBase::CountFiles() {
   std::vector<std::string> files;
@@ -997,7 +997,6 @@ void DBTestBase::MoveFilesToLevel(int level, int cf) {
   }
 }
 
-#ifndef ROCKSDB_LITE
 void DBTestBase::DumpFileCounts(const char* label) {
   fprintf(stderr, "---\n%s:\n", label);
   fprintf(stderr, "maxoverlap: %" PRIu64 "\n",
@@ -1009,7 +1008,6 @@ void DBTestBase::DumpFileCounts(const char* label) {
     }
   }
 }
-#endif  // !ROCKSDB_LITE
 
 std::string DBTestBase::DumpSSTableList() {
   std::string property;

db/db_test_util.h

@@ -863,13 +863,13 @@ class DBTestBase : public testing::Test {
   size_t TotalLiveFiles(int cf = 0);
 
   size_t CountLiveFiles();
+#endif  // ROCKSDB_LITE
 
   int NumTableFilesAtLevel(int level, int cf = 0);
 
   double CompressionRatioAtLevel(int level, int cf = 0);
 
   int TotalTableFiles(int cf = 0, int levels = -1);
-#endif  // ROCKSDB_LITE
 
   // Return spread of files per level
   std::string FilesPerLevel(int cf = 0);
@@ -897,9 +897,7 @@ class DBTestBase : public testing::Test {
 
   void MoveFilesToLevel(int level, int cf = 0);
 
-#ifndef ROCKSDB_LITE
   void DumpFileCounts(const char* label);
-#endif  // ROCKSDB_LITE
 
   std::string DumpSSTableList();
 

db/external_sst_file_test.cc

@@ -1940,55 +1940,6 @@ TEST_F(ExternalSSTFileTest, IngestBehind) {
   size_t kcnt = 0;
   VerifyDBFromMap(true_data, &kcnt, false);
 }
-
-TEST_F(ExternalSSTFileTest, SkipBloomFilter) {
-  Options options = CurrentOptions();
-
-  BlockBasedTableOptions table_options;
-  table_options.filter_policy.reset(NewBloomFilterPolicy(10));
-  table_options.cache_index_and_filter_blocks = true;
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-
-
-  // Create external SST file and include bloom filters
-  options.statistics = rocksdb::CreateDBStatistics();
-  DestroyAndReopen(options);
-  {
-    std::string file_path = sst_files_dir_ + "sst_with_bloom.sst";
-    SstFileWriter sst_file_writer(EnvOptions(), options);
-    ASSERT_OK(sst_file_writer.Open(file_path));
-    ASSERT_OK(sst_file_writer.Put("Key1", "Value1"));
-    ASSERT_OK(sst_file_writer.Finish());
-
-    ASSERT_OK(
-        db_->IngestExternalFile({file_path}, IngestExternalFileOptions()));
-
-    ASSERT_EQ(Get("Key1"), "Value1");
-    ASSERT_GE(
-        options.statistics->getTickerCount(Tickers::BLOCK_CACHE_FILTER_ADD), 1);
-  }
-
-  // Create external SST file but skip bloom filters
-  options.statistics = rocksdb::CreateDBStatistics();
-  DestroyAndReopen(options);
-  {
-    std::string file_path = sst_files_dir_ + "sst_with_no_bloom.sst";
-    SstFileWriter sst_file_writer(EnvOptions(), options, nullptr, true,
-                                  Env::IOPriority::IO_TOTAL,
-                                  true /* skip_filters */);
-    ASSERT_OK(sst_file_writer.Open(file_path));
-    ASSERT_OK(sst_file_writer.Put("Key1", "Value1"));
-    ASSERT_OK(sst_file_writer.Finish());
-
-    ASSERT_OK(
-        db_->IngestExternalFile({file_path}, IngestExternalFileOptions()));
-
-    ASSERT_EQ(Get("Key1"), "Value1");
-    ASSERT_EQ(
-        options.statistics->getTickerCount(Tickers::BLOCK_CACHE_FILTER_ADD), 0);
-  }
-}
-
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/job_context.h

@@ -81,11 +81,8 @@ struct SuperVersionContext {
 struct JobContext {
   inline bool HaveSomethingToDelete() const {
     return full_scan_candidate_files.size() || sst_delete_files.size() ||
-           log_delete_files.size() || manifest_delete_files.size();
-  }
-
-  inline bool HaveSomethingToClean() const {
-    return memtables_to_free.size() > 0 || logs_to_free.size() > 0 ||
+           log_delete_files.size() || manifest_delete_files.size() ||
+           memtables_to_free.size() > 0 || logs_to_free.size() > 0 ||
            superversion_context.HaveSomethingToDelete();
   }
 

db/log_reader.h

@@ -54,6 +54,7 @@ class Reader {
   // The Reader will start reading at the first record located at physical
   // position >= initial_offset within the file.
   Reader(std::shared_ptr<Logger> info_log,
+  // @lint-ignore TXT2 T25377293 Grandfathered in
 	 unique_ptr<SequentialFileReader>&& file,
          Reporter* reporter, bool checksum, uint64_t initial_offset,
          uint64_t log_num);

db/malloc_stats.cc

@@ -16,12 +16,7 @@
 namespace rocksdb {
 
 #ifdef ROCKSDB_JEMALLOC
-#ifdef __FreeBSD__
-#include <malloc_np.h>
-#define je_malloc_stats_print malloc_stats_print
-#else
 #include "jemalloc/jemalloc.h"
-#endif
 
 typedef struct {
   char* cur;

db/memtable.cc

@@ -236,14 +236,6 @@ int MemTable::KeyComparator::operator()(const char* prefix_len_key,
   return comparator.Compare(a, key);
 }
 
-void MemTableRep::InsertConcurrently(KeyHandle handle) {
-#ifndef ROCKSDB_LITE
-    throw std::runtime_error("concurrent insert not supported");
-#else
-    abort();
-#endif
-  }
-
 Slice MemTableRep::UserKey(const char* key) const {
   Slice slice = GetLengthPrefixedSlice(key);
   return Slice(slice.data(), slice.size() - 8);

db/wal_manager.cc

@@ -39,15 +39,6 @@ namespace rocksdb {
 
 #ifndef ROCKSDB_LITE
 
-Status WalManager::DeleteFile(const std::string& fname, uint64_t number) {
-  auto s = env_->DeleteFile(db_options_.wal_dir + "/" + fname);
-  if (s.ok()) {
-    MutexLock l(&read_first_record_cache_mutex_);
-    read_first_record_cache_.erase(number);
-  }
-  return s;
-}
-
 Status WalManager::GetSortedWalFiles(VectorLogPtr& files) {
   // First get sorted files in db dir, then get sorted files from archived
   // dir, to avoid a race condition where a log file is moved to archived

db/wal_manager.h

@@ -49,8 +49,6 @@ class WalManager {
 
   void ArchiveWALFile(const std::string& fname, uint64_t number);
 
-  Status DeleteFile(const std::string& fname, uint64_t number);
-
   Status TEST_ReadFirstRecord(const WalFileType type, const uint64_t number,
                               SequenceNumber* sequence) {
     return ReadFirstRecord(type, number, sequence);

db/write_batch.cc

@@ -15,12 +15,10 @@
 //    kTypeValue varstring varstring
 //    kTypeDeletion varstring
 //    kTypeSingleDeletion varstring
-//    kTypeRangeDeletion varstring varstring
 //    kTypeMerge varstring varstring
 //    kTypeColumnFamilyValue varint32 varstring varstring
-//    kTypeColumnFamilyDeletion varint32 varstring
-//    kTypeColumnFamilySingleDeletion varint32 varstring
-//    kTypeColumnFamilyRangeDeletion varint32 varstring varstring
+//    kTypeColumnFamilyDeletion varint32 varstring varstring
+//    kTypeColumnFamilySingleDeletion varint32 varstring varstring
 //    kTypeColumnFamilyMerge varint32 varstring varstring
 //    kTypeBeginPrepareXID varstring
 //    kTypeEndPrepareXID
@@ -146,12 +144,6 @@ WriteBatch::WriteBatch(const std::string& rep)
       max_bytes_(0),
       rep_(rep) {}
 
-WriteBatch::WriteBatch(std::string&& rep)
-    : save_points_(nullptr),
-      content_flags_(ContentFlags::DEFERRED),
-      max_bytes_(0),
-      rep_(std::move(rep)) {}
-
 WriteBatch::WriteBatch(const WriteBatch& src)
     : save_points_(src.save_points_),
       wal_term_point_(src.wal_term_point_),

docs/Gemfile.lock

@@ -105,13 +105,13 @@ GEM
       rb-fsevent (>= 0.9.3)
       rb-inotify (>= 0.9.7)
     mercenary (0.3.6)
-    mini_portile2 (2.3.0)
+    mini_portile2 (2.1.0)
     minima (2.0.0)
     minitest (5.9.1)
     multipart-post (2.0.0)
     net-dns (0.8.0)
-    nokogiri (1.8.1)
-      mini_portile2 (~> 2.3.0)
+    nokogiri (1.6.8.1)
+      mini_portile2 (~> 2.1.0)
     octokit (4.4.1)
       sawyer (~> 0.7.0, >= 0.5.3)
     pathutil (0.14.0)

env/env.cc

@@ -73,18 +73,9 @@ RandomAccessFile::~RandomAccessFile() {
 WritableFile::~WritableFile() {
 }
 
-Logger::~Logger() { Close(); }
-
-Status Logger::Close() {
-  if (!closed_) {
-    closed_ = true;
-    return CloseImpl();
-  }
-  return Status::OK();
+Logger::~Logger() {
 }
 
-Status Logger::CloseImpl() { return Status::OK(); }
-
 FileLock::~FileLock() {
 }
 

env/env_hdfs.cc

@@ -277,16 +277,6 @@ class HdfsLogger : public Logger {
   HdfsWritableFile* file_;
   uint64_t (*gettid_)();  // Return the thread id for the current thread
 
-  virtual Status CloseImpl() {
-    ROCKS_LOG_DEBUG(mylog, "[hdfs] HdfsLogger closed %s\n",
-                    file_->getName().c_str());
-    Status s = file_->Close();
-    if (mylog != nullptr && mylog == this) {
-      mylog = nullptr;
-    }
-    return s;
-  }
-
  public:
   HdfsLogger(HdfsWritableFile* f, uint64_t (*gettid)())
       : file_(f), gettid_(gettid) {
@@ -295,6 +285,12 @@ class HdfsLogger : public Logger {
   }
 
   virtual ~HdfsLogger() {
+    ROCKS_LOG_DEBUG(mylog, "[hdfs] HdfsLogger closed %s\n",
+                    file_->getName().c_str());
+    delete file_;
+    if (mylog != nullptr && mylog == this) {
+      mylog = nullptr;
+    }
   }
 
   virtual void Logv(const char* format, va_list ap) {

env/posix_logger.h

@@ -24,7 +24,6 @@
 #endif
 
 #include <atomic>
-#include "env/io_posix.h"
 #include "monitoring/iostats_context_imp.h"
 #include "rocksdb/env.h"
 #include "util/sync_point.h"
@@ -33,15 +32,6 @@ namespace rocksdb {
 
 class PosixLogger : public Logger {
  private:
-  virtual Status CloseImpl() override {
-    int ret;
-
-    ret = fclose(file_);
-    if (ret) {
-      return IOError("Unable to close log file", "", ret);
-    }
-    return Status::OK();
-  }
   FILE* file_;
   uint64_t (*gettid_)();  // Return the thread id for the current thread
   std::atomic_size_t log_size_;
@@ -61,7 +51,9 @@ class PosixLogger : public Logger {
         last_flush_micros_(0),
         env_(env),
         flush_pending_(false) {}
-  virtual ~PosixLogger() { Close(); }
+  virtual ~PosixLogger() {
+    fclose(file_);
+  }
   virtual void Flush() override {
     TEST_SYNC_POINT("PosixLogger::Flush:Begin1");
     TEST_SYNC_POINT("PosixLogger::Flush:Begin2");

hdfs/setup.sh

@@ -1,3 +1,4 @@
+# shellcheck disable=SC2148
 export USE_HDFS=1
 export LD_LIBRARY_PATH=$JAVA_HOME/jre/lib/amd64/server:$JAVA_HOME/jre/lib/amd64:/usr/lib/hadoop/lib/native
 

include/rocksdb/db.h

@@ -163,12 +163,6 @@ class DB {
                      const std::vector<ColumnFamilyDescriptor>& column_families,
                      std::vector<ColumnFamilyHandle*>* handles, DB** dbptr);
 
-  // Close the DB by releasing resources, closing files etc. This should be
-  // called before calling the desctructor so that the caller can get back a
-  // status in case there are any errors. Regardless of the return status, the
-  // DB must be freed
-  virtual Status Close() { return Status::OK(); }
-
   // ListColumnFamilies will open the DB specified by argument name
   // and return the list of all column families in that DB
   // through column_families argument. The ordering of
@@ -979,7 +973,7 @@ class DB {
   // the file can fit in, and ingest the file into this level (2). A file that
   // have a key range that overlap with the memtable key range will require us
   // to Flush the memtable first before ingesting the file.
-  // In the second mode we will always ingest in the bottom most level (see
+  // In the second mode we will always ingest in the bottom mode level (see
   // docs to IngestExternalFileOptions::ingest_behind).
   //
   // (1) External SST files can be created using SstFileWriter

include/rocksdb/env.h

@@ -819,12 +819,9 @@ class Logger {
   size_t kDoNotSupportGetLogFileSize = (std::numeric_limits<size_t>::max)();
 
   explicit Logger(const InfoLogLevel log_level = InfoLogLevel::INFO_LEVEL)
-      : closed_(false), log_level_(log_level) {}
+      : log_level_(log_level) {}
   virtual ~Logger();
 
-  // Close the log file. Must be called before destructor
-  virtual Status Close();
-
   // Write a header to the log file with the specified format
   // It is recommended that you log all header information at the start of the
   // application. But it is not enforced.
@@ -855,8 +852,6 @@ class Logger {
   // No copying allowed
   Logger(const Logger&);
   void operator=(const Logger&);
-  virtual Status CloseImpl();
-  bool closed_;
   InfoLogLevel log_level_;
 };
 

include/rocksdb/iterator.h

@@ -51,7 +51,7 @@ class Iterator : public Cleanable {
   // Position at the last key in the source that at or before target
   // The iterator is Valid() after this call iff the source contains
   // an entry that comes at or before target.
-  virtual void SeekForPrev(const Slice& target) = 0;
+  virtual void SeekForPrev(const Slice& target) {}
 
   // Moves to the next entry in the source.  After this call, Valid() is
   // true iff the iterator was not positioned at the last entry in the source.
@@ -97,6 +97,9 @@ class Iterator : public Cleanable {
   // Property "rocksdb.iterator.super-version-number":
   //   LSM version used by the iterator. The same format as DB Property
   //   kCurrentSuperVersionNumber. See its comment for more information.
+  // Property "rocksdb.iterator.internal-key":
+  //   Get the user-key portion of the internal key at which the iteration
+  //   stopped.
   virtual Status GetProperty(std::string prop_name, std::string* prop);
 
  private:

include/rocksdb/memtablerep.h

@@ -96,7 +96,13 @@ class MemTableRep {
 
   // Like Insert(handle), but may be called concurrent with other calls
   // to InsertConcurrently for other handles
-  virtual void InsertConcurrently(KeyHandle handle);
+  virtual void InsertConcurrently(KeyHandle handle) {
+#ifndef ROCKSDB_LITE
+    throw std::runtime_error("concurrent insert not supported");
+#else
+    abort();
+#endif
+  }
 
   // Returns true iff an entry that compares equal to key is in the collection.
   virtual bool Contains(const char* key) const = 0;

include/rocksdb/sst_file_writer.h

@@ -72,18 +72,16 @@ class SstFileWriter {
   SstFileWriter(const EnvOptions& env_options, const Options& options,
                 ColumnFamilyHandle* column_family = nullptr,
                 bool invalidate_page_cache = true,
-                Env::IOPriority io_priority = Env::IOPriority::IO_TOTAL,
-                bool skip_filters = false)
+                Env::IOPriority io_priority = Env::IOPriority::IO_TOTAL)
       : SstFileWriter(env_options, options, options.comparator, column_family,
-                      invalidate_page_cache, io_priority, skip_filters) {}
+                      invalidate_page_cache, io_priority) {}
 
   // Deprecated API
   SstFileWriter(const EnvOptions& env_options, const Options& options,
                 const Comparator* user_comparator,
                 ColumnFamilyHandle* column_family = nullptr,
                 bool invalidate_page_cache = true,
-                Env::IOPriority io_priority = Env::IOPriority::IO_TOTAL,
-                bool skip_filters = false);
+                Env::IOPriority io_priority = Env::IOPriority::IO_TOTAL);
 
   ~SstFileWriter();
 

include/rocksdb/table.h

@@ -217,11 +217,6 @@ struct BlockBasedTableOptions {
   // This option only affects newly written tables. When reading exising tables,
   // the information about version is read from the footer.
   uint32_t format_version = 2;
-
-  // Store index blocks on disk in compressed format. Changing this option to
-  // false  will avoid the overhead of decompression if index blocks are evicted
-  // and read back
-  bool enable_index_compression = true;
 };
 
 // Table Properties that are specific to block-based table properties.

include/rocksdb/types.h

@@ -25,7 +25,7 @@ enum EntryType {
   kEntryOther,
 };
 
-// <user key, sequence number, and entry type> tuple.
+// <user key, seqeence number and entry type> tuple.
 struct FullKey {
   Slice user_key;
   SequenceNumber sequence;

include/rocksdb/utilities/db_ttl.h

@@ -60,10 +60,6 @@ class DBWithTTL : public StackableDB {
                      DBWithTTL** dbptr, std::vector<int32_t> ttls,
                      bool read_only = false);
 
- virtual void SetTtl(int32_t ttl) = 0;
-
- virtual void SetTtl(ColumnFamilyHandle *h, int32_t ttl) = 0;
-
  protected:
   explicit DBWithTTL(DB* db) : StackableDB(db) {}
 };

include/rocksdb/utilities/stackable_db.h

@@ -23,19 +23,18 @@ class StackableDB : public DB {
   explicit StackableDB(DB* db) : db_(db) {}
 
   // StackableDB take shared ownership of the underlying db.
-  explicit StackableDB(std::shared_ptr<DB>& db)
-      : db_(db.get()), shared_db_(db) {}
+  explicit StackableDB(std::shared_ptr<DB> db)
+      : db_(db.get()), shared_db_ptr_(db) {}
 
   ~StackableDB() {
-    if (shared_db_ == nullptr) {
+    if (shared_db_ptr_ == nullptr) {
       delete db_;
     } else {
-      assert(shared_db_.get() == db_);
+      assert(shared_db_ptr_.get() == db_);
     }
+    db_ = nullptr;
   }
 
-  virtual Status Close() override { return db_->Close(); }
-
   virtual DB* GetBaseDB() {
     return db_;
   }
@@ -384,7 +383,7 @@ class StackableDB : public DB {
 
  protected:
   DB* db_;
-  std::shared_ptr<DB> shared_db_;
+  std::shared_ptr<DB> shared_db_ptr_;
 };
 
 } //  namespace rocksdb

include/rocksdb/version.h

@@ -5,8 +5,8 @@
 #pragma once
 
 #define ROCKSDB_MAJOR 5
-#define ROCKSDB_MINOR 10
-#define ROCKSDB_PATCH 0
+#define ROCKSDB_MINOR 11
+#define ROCKSDB_PATCH 2
 
 // Do not use these. We made the mistake of declaring macros starting with
 // double underscore. Now we have to live with our choice. We'll deprecate these

include/rocksdb/write_batch.h

@@ -315,7 +315,6 @@ class WriteBatch : public WriteBatchBase {
 
   // Constructor with a serialized string object
   explicit WriteBatch(const std::string& rep);
-  explicit WriteBatch(std::string&& rep);
 
   WriteBatch(const WriteBatch& src);
   WriteBatch(WriteBatch&& src) noexcept;

java/CMakeLists.txt

@@ -24,6 +24,7 @@ set(JNI_NATIVE_SOURCES
         rocksjni/lru_cache.cc
         rocksjni/memtablejni.cc
         rocksjni/merge_operator.cc
+        rocksjni/native_comparator_wrapper_test.cc
         rocksjni/options.cc
         rocksjni/options_util.cc
         rocksjni/ratelimiterjni.cc
@@ -87,6 +88,8 @@ set(NATIVE_JAVA_CLASSES
         org.rocksdb.LRUCache
         org.rocksdb.MemTableConfig
         org.rocksdb.MergeOperator
+        org.rocksdb.NativeComparatorWrapper
+        org.rocksdb.NativeComparatorWrapperTest.NativeStringComparatorWrapper
         org.rocksdb.NativeLibraryLoader
         org.rocksdb.Options
         org.rocksdb.OptionsUtil
@@ -122,6 +125,8 @@ set(NATIVE_JAVA_CLASSES
         org.rocksdb.WriteBatchTestInternalHelper
         org.rocksdb.WriteBatchWithIndex
         org.rocksdb.WriteOptions
+        org.rocksdb.util.CapturingWriteBatchHandler
+        org.rocksdb.util.WriteBatchGetter
 )
 
 include(FindJava)
@@ -177,6 +182,7 @@ add_jar(
   src/main/java/org/rocksdb/CompactionStyle.java
   src/main/java/org/rocksdb/Comparator.java
   src/main/java/org/rocksdb/ComparatorOptions.java
+  src/main/java/org/rocksdb/ComparatorType.java
   src/main/java/org/rocksdb/CompressionOptions.java
   src/main/java/org/rocksdb/CompressionType.java
   src/main/java/org/rocksdb/DBOptions.java
@@ -203,6 +209,7 @@ add_jar(
   src/main/java/org/rocksdb/MergeOperator.java
   src/main/java/org/rocksdb/MutableColumnFamilyOptions.java
   src/main/java/org/rocksdb/MutableColumnFamilyOptionsInterface.java
+  src/main/java/org/rocksdb/NativeComparatorWrapper.java
   src/main/java/org/rocksdb/NativeLibraryLoader.java
   src/main/java/org/rocksdb/Options.java
   src/main/java/org/rocksdb/OptionsUtil.java
@@ -244,11 +251,14 @@ add_jar(
   src/main/java/org/rocksdb/WriteOptions.java
   src/test/java/org/rocksdb/BackupEngineTest.java
   src/test/java/org/rocksdb/IngestExternalFileOptionsTest.java
+  src/test/java/org/rocksdb/NativeComparatorWrapperTest.java
   src/test/java/org/rocksdb/PlatformRandomHelper.java
   src/test/java/org/rocksdb/RocksDBExceptionTest.java
   src/test/java/org/rocksdb/RocksMemoryResource.java
   src/test/java/org/rocksdb/SnapshotTest.java
   src/test/java/org/rocksdb/WriteBatchTest.java
+  src/test/java/org/rocksdb/util/CapturingWriteBatchHandler.java
+  src/test/java/org/rocksdb/util/WriteBatchGetter.java
   INCLUDE_JARS ${JAVA_TESTCLASSPATH}
 )
 

java/Makefile

@@ -1,6 +1,7 @@
 NATIVE_JAVA_CLASSES = org.rocksdb.AbstractCompactionFilter\
 	org.rocksdb.AbstractCompactionFilterFactory\
 	org.rocksdb.AbstractSlice\
+	org.rocksdb.AbstractTransactionNotifier\
 	org.rocksdb.BackupEngine\
 	org.rocksdb.BackupableDBOptions\
 	org.rocksdb.BlockBasedTableConfig\
@@ -29,6 +30,9 @@ NATIVE_JAVA_CLASSES = org.rocksdb.AbstractCompactionFilter\
 	org.rocksdb.Logger\
 	org.rocksdb.LRUCache\
 	org.rocksdb.MergeOperator\
+	org.rocksdb.NativeComparatorWrapper\
+	org.rocksdb.OptimisticTransactionDB\
+	org.rocksdb.OptimisticTransactionOptions\
 	org.rocksdb.Options\
 	org.rocksdb.OptionsUtil\
 	org.rocksdb.PlainTableConfig\
@@ -45,6 +49,10 @@ NATIVE_JAVA_CLASSES = org.rocksdb.AbstractCompactionFilter\
 	org.rocksdb.Slice\
 	org.rocksdb.SstFileWriter\
 	org.rocksdb.Statistics\
+	org.rocksdb.Transaction\
+	org.rocksdb.TransactionDB\
+	org.rocksdb.TransactionDBOptions\
+	org.rocksdb.TransactionOptions\
 	org.rocksdb.TransactionLogIterator\
 	org.rocksdb.TtlDB\
 	org.rocksdb.VectorMemTableConfig\
@@ -57,6 +65,7 @@ NATIVE_JAVA_CLASSES = org.rocksdb.AbstractCompactionFilter\
 	org.rocksdb.WBWIRocksIterator
 
 NATIVE_JAVA_TEST_CLASSES = org.rocksdb.RocksDBExceptionTest\
+    org.rocksdb.NativeComparatorWrapperTest.NativeStringComparatorWrapper\
     org.rocksdb.WriteBatchTest\
     org.rocksdb.WriteBatchTestInternalHelper
 
@@ -104,7 +113,11 @@ JAVA_TESTS = org.rocksdb.BackupableDBOptionsTest\
 	org.rocksdb.MergeTest\
 	org.rocksdb.MixedOptionsTest\
 	org.rocksdb.MutableColumnFamilyOptionsTest\
+	org.rocksdb.NativeComparatorWrapperTest\
 	org.rocksdb.NativeLibraryLoaderTest\
+	org.rocksdb.OptimisticTransactionTest\
+	org.rocksdb.OptimisticTransactionDBTest\
+	org.rocksdb.OptimisticTransactionOptionsTest\
 	org.rocksdb.OptionsUtilTest\
 	org.rocksdb.OptionsTest\
 	org.rocksdb.PlainTableConfigTest\
@@ -120,6 +133,10 @@ JAVA_TESTS = org.rocksdb.BackupableDBOptionsTest\
 	org.rocksdb.SliceTest\
 	org.rocksdb.SnapshotTest\
 	org.rocksdb.SstFileWriterTest\
+	org.rocksdb.TransactionTest\
+	org.rocksdb.TransactionDBTest\
+	org.rocksdb.TransactionOptionsTest\
+	org.rocksdb.TransactionDBOptionsTest\
 	org.rocksdb.TransactionLogIteratorTest\
 	org.rocksdb.TtlDBTest\
 	org.rocksdb.StatisticsTest\
@@ -209,6 +226,20 @@ column_family_sample: java
 	java $(JAVA_ARGS) -Djava.library.path=target -cp $(MAIN_CLASSES):$(SAMPLES_MAIN_CLASSES) RocksDBColumnFamilySample /tmp/rocksdbjni
 	$(AM_V_at)@rm -rf /tmp/rocksdbjni
 
+transaction_sample: java
+	$(AM_V_GEN)mkdir -p $(SAMPLES_MAIN_CLASSES)
+	$(AM_V_at)javac -cp $(MAIN_CLASSES) -d $(SAMPLES_MAIN_CLASSES) $(SAMPLES_MAIN_SRC)/TransactionSample.java
+	$(AM_V_at)@rm -rf /tmp/rocksdbjni
+	java -ea -Xcheck:jni -Djava.library.path=target -cp $(MAIN_CLASSES):$(SAMPLES_MAIN_CLASSES) TransactionSample /tmp/rocksdbjni
+	$(AM_V_at)@rm -rf /tmp/rocksdbjni
+
+optimistic_transaction_sample: java
+	$(AM_V_GEN)mkdir -p $(SAMPLES_MAIN_CLASSES)
+	$(AM_V_at)javac -cp $(MAIN_CLASSES) -d $(SAMPLES_MAIN_CLASSES) $(SAMPLES_MAIN_SRC)/OptimisticTransactionSample.java
+	$(AM_V_at)@rm -rf /tmp/rocksdbjni
+	java -ea -Xcheck:jni -Djava.library.path=target -cp $(MAIN_CLASSES):$(SAMPLES_MAIN_CLASSES) OptimisticTransactionSample /tmp/rocksdbjni
+	$(AM_V_at)@rm -rf /tmp/rocksdbjni
+
 resolve_test_deps:
 	test -d "$(JAVA_TEST_LIBDIR)" || mkdir -p "$(JAVA_TEST_LIBDIR)"
 	test -s "$(JAVA_JUNIT_JAR)" || cp $(MVN_LOCAL)/junit/junit/4.12/junit-4.12.jar $(JAVA_TEST_LIBDIR) || curl -k -L -o $(JAVA_JUNIT_JAR) $(SEARCH_REPO_URL)junit/junit/4.12/junit-4.12.jar

java/jdb_bench.sh

@@ -1,3 +1,4 @@
+# shellcheck disable=SC2148
 PLATFORM=64
 if [ `getconf LONG_BIT` != "64" ]
 then
@@ -7,4 +8,5 @@ fi
 ROCKS_JAR=`find target -name rocksdbjni*.jar`
 
 echo "Running benchmark in $PLATFORM-Bit mode."
+# shellcheck disable=SC2068
 java -server -d$PLATFORM -XX:NewSize=4m -XX:+AggressiveOpts -Djava.library.path=target -cp "${ROCKS_JAR}:benchmark/target/classes" org.rocksdb.benchmark.DbBenchmark $@

java/rocksjni/columnfamilyhandle.cc

@@ -3,8 +3,8 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).
 //
-// This file implements the "bridge" between Java and C++ and enables
-// calling c++ rocksdb::Iterator methods from Java side.
+// This file implements the "bridge" between Java and C++ for
+// rocksdb::ColumnFamilyHandle.
 
 #include <stdio.h>
 #include <stdlib.h>
@@ -13,14 +13,56 @@
 #include "include/org_rocksdb_ColumnFamilyHandle.h"
 #include "rocksjni/portal.h"
 
+/*
+ * Class:     org_rocksdb_ColumnFamilyHandle
+ * Method:    getName
+ * Signature: (J)[B
+ */
+jbyteArray Java_org_rocksdb_ColumnFamilyHandle_getName(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* cfh = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jhandle);
+  std::string cf_name = cfh->GetName();
+  return rocksdb::JniUtil::copyBytes(env, cf_name);
+}
+
+/*
+* Class:     org_rocksdb_ColumnFamilyHandle
+* Method:    getID
+* Signature: (J)I
+*/
+jint Java_org_rocksdb_ColumnFamilyHandle_getID(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* cfh = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jhandle);
+  const int32_t id = cfh->GetID();
+  return static_cast<jint>(id);
+}
+
+/*
+ * Class:     org_rocksdb_ColumnFamilyHandle
+ * Method:    getDescriptor
+ * Signature: (J)Lorg/rocksdb/ColumnFamilyDescriptor;
+ */
+jobject Java_org_rocksdb_ColumnFamilyHandle_getDescriptor(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* cfh = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jhandle);
+  rocksdb::ColumnFamilyDescriptor desc;
+  rocksdb::Status s = cfh->GetDescriptor(&desc);
+  if (s.ok()) {
+    return rocksdb::ColumnFamilyDescriptorJni::construct(env, &desc);
+  } else {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+    return nullptr;
+  }
+}
+
 /*
  * Class:     org_rocksdb_ColumnFamilyHandle
  * Method:    disposeInternal
  * Signature: (J)V
  */
 void Java_org_rocksdb_ColumnFamilyHandle_disposeInternal(
-    JNIEnv* env, jobject jobj, jlong handle) {
-  auto* cfh = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(handle);
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* cfh = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jhandle);
   assert(cfh != nullptr);
   delete cfh;
 }

java/rocksjni/compaction_filter_factory.cc

@@ -34,4 +34,5 @@ void Java_org_rocksdb_AbstractCompactionFilterFactory_disposeInternal(
   auto* ptr_sptr_cff =
       reinterpret_cast<std::shared_ptr<rocksdb::CompactionFilterFactoryJniCallback> *>(jhandle);
   delete ptr_sptr_cff;
+// @lint-ignore TXT4 T25377293 Grandfathered in
 }

java/rocksjni/comparator.cc

@@ -14,6 +14,7 @@
 
 #include "include/org_rocksdb_Comparator.h"
 #include "include/org_rocksdb_DirectComparator.h"
+#include "include/org_rocksdb_NativeComparatorWrapper.h"
 #include "rocksjni/comparatorjnicallback.h"
 #include "rocksjni/portal.h"
 
@@ -49,4 +50,16 @@ jlong Java_org_rocksdb_DirectComparator_createNewDirectComparator0(
       new rocksdb::DirectComparatorJniCallback(env, jobj, copt);
   return reinterpret_cast<jlong>(c);
 }
+
+/*
+ * Class:     org_rocksdb_NativeComparatorWrapper
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_NativeComparatorWrapper_disposeInternal(
+    JNIEnv* env, jobject jobj, jlong jcomparator_handle) {
+  auto* comparator =
+      reinterpret_cast<rocksdb::Comparator*>(jcomparator_handle);
+  delete comparator;
+}
 // </editor-fold>

java/rocksjni/ingest_external_file_options.cc

@@ -146,4 +146,5 @@ void Java_org_rocksdb_IngestExternalFileOptions_disposeInternal(
   auto* options =
       reinterpret_cast<rocksdb::IngestExternalFileOptions*>(jhandle);
   delete options;
+// @lint-ignore TXT4 T25377293 Grandfathered in
 }

java/rocksjni/jnicallback.cc

@@ -49,4 +49,5 @@ JniCallback::~JniCallback() {
 
   releaseJniEnv(attached_thread);
 }
+// @lint-ignore TXT4 T25377293 Grandfathered in
 }  // namespace rocksdb

java/rocksjni/jnicallback.h

@@ -25,4 +25,5 @@ namespace rocksdb {
   };
 }
 
+// @lint-ignore TXT4 T25377293 Grandfathered in
 #endif  // JAVA_ROCKSJNI_JNICALLBACK_H_

java/rocksjni/native_comparator_wrapper_test.cc

@@ -0,0 +1,50 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include <jni.h>
+#include <string>
+
+#include "rocksdb/comparator.h"
+#include "rocksdb/slice.h"
+
+#include "include/org_rocksdb_NativeComparatorWrapperTest_NativeStringComparatorWrapper.h"
+
+namespace rocksdb {
+
+class NativeComparatorWrapperTestStringComparator
+    : public Comparator {
+
+  const char* Name() const {
+    return "NativeComparatorWrapperTestStringComparator";
+  }
+
+  int Compare(
+      const Slice& a, const Slice& b) const {
+    return a.ToString().compare(b.ToString());
+  }
+
+  void FindShortestSeparator(
+      std::string* start, const Slice& limit) const {
+    return;
+  }
+
+  void FindShortSuccessor(
+      std::string* key) const {
+    return;
+  }
+};
+}  // end of rocksdb namespace
+
+/*
+ * Class:     org_rocksdb_NativeComparatorWrapperTest_NativeStringComparatorWrapper
+ * Method:    newStringComparator
+ * Signature: ()J
+ */
+jlong Java_org_rocksdb_NativeComparatorWrapperTest_00024NativeStringComparatorWrapper_newStringComparator(
+    JNIEnv* env , jobject jobj) {
+  auto* comparator =
+      new rocksdb::NativeComparatorWrapperTestStringComparator();
+  return reinterpret_cast<jlong>(comparator);
+}

java/rocksjni/optimistic_transaction_db.cc

@@ -0,0 +1,267 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the "bridge" between Java and C++
+// for rocksdb::TransactionDB.
+
+#include <jni.h>
+
+#include "include/org_rocksdb_OptimisticTransactionDB.h"
+
+#include "rocksdb/options.h"
+#include "rocksdb/utilities/optimistic_transaction_db.h"
+#include "rocksdb/utilities/transaction.h"
+
+#include "rocksjni/portal.h"
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    open
+ * Signature: (JLjava/lang/String;)J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionDB_open__JLjava_lang_String_2(
+    JNIEnv* env, jclass jcls, jlong joptions_handle, jstring jdb_path) {
+  const char* db_path = env->GetStringUTFChars(jdb_path, nullptr);
+  if (db_path == nullptr) {
+      // exception thrown: OutOfMemoryError
+      return 0;
+  }
+
+  auto* options = reinterpret_cast<rocksdb::Options*>(joptions_handle);
+  rocksdb::OptimisticTransactionDB* otdb = nullptr;
+  rocksdb::Status s =
+      rocksdb::OptimisticTransactionDB::Open(*options, db_path, &otdb);
+  env->ReleaseStringUTFChars(jdb_path, db_path);
+
+  if (s.ok()) {
+    return reinterpret_cast<jlong>(otdb);
+  } else {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+    return 0;
+  }
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    open
+ * Signature: (JLjava/lang/String;[[B[J)[J
+ */
+jlongArray Java_org_rocksdb_OptimisticTransactionDB_open__JLjava_lang_String_2_3_3B_3J(
+    JNIEnv* env, jclass jcls, jlong jdb_options_handle, jstring jdb_path,
+    jobjectArray jcolumn_names, jlongArray jcolumn_options_handles) {
+  const char* db_path = env->GetStringUTFChars(jdb_path, nullptr);
+  if (db_path == nullptr) {
+      // exception thrown: OutOfMemoryError
+      return nullptr;
+  }
+
+  std::vector<rocksdb::ColumnFamilyDescriptor> column_families;
+  const jsize len_cols = env->GetArrayLength(jcolumn_names);
+  if (len_cols > 0) {
+    if (env->EnsureLocalCapacity(len_cols) != 0) {
+      // out of memory
+      env->ReleaseStringUTFChars(jdb_path, db_path);
+      return nullptr;
+    }
+
+    jlong* jco =
+        env->GetLongArrayElements(jcolumn_options_handles, nullptr);
+    if(jco == nullptr) {
+        // exception thrown: OutOfMemoryError
+        env->ReleaseStringUTFChars(jdb_path, db_path);
+        return nullptr;
+    }
+
+    for (int i = 0; i < len_cols; i++) {
+        const jobject jcn = env->GetObjectArrayElement(jcolumn_names, i);
+        if (env->ExceptionCheck()) {
+            // exception thrown: ArrayIndexOutOfBoundsException
+            env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+            env->ReleaseStringUTFChars(jdb_path, db_path);
+            return nullptr;
+        }
+
+        const jbyteArray jcn_ba = reinterpret_cast<jbyteArray>(jcn);
+        const jsize jcf_name_len = env->GetArrayLength(jcn_ba);
+        if (env->EnsureLocalCapacity(jcf_name_len) != 0) {
+          // out of memory
+          env->DeleteLocalRef(jcn);
+          env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+          env->ReleaseStringUTFChars(jdb_path, db_path);
+          return nullptr;
+        }
+
+        jbyte* jcf_name = env->GetByteArrayElements(jcn_ba, nullptr);
+        if (jcf_name == nullptr) {
+            // exception thrown: OutOfMemoryError
+            env->DeleteLocalRef(jcn);
+            env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+            env->ReleaseStringUTFChars(jdb_path, db_path);
+            return nullptr;
+        }
+
+        const std::string cf_name(reinterpret_cast<char *>(jcf_name), jcf_name_len);
+        const rocksdb::ColumnFamilyOptions* cf_options =
+            reinterpret_cast<rocksdb::ColumnFamilyOptions*>(jco[i]);
+        column_families.push_back(
+            rocksdb::ColumnFamilyDescriptor(cf_name, *cf_options));
+
+        env->ReleaseByteArrayElements(jcn_ba, jcf_name, JNI_ABORT);
+        env->DeleteLocalRef(jcn);
+    }
+    env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+  }
+
+  auto* db_options = reinterpret_cast<rocksdb::DBOptions*>(jdb_options_handle);
+  std::vector<rocksdb::ColumnFamilyHandle*> handles;
+  rocksdb::OptimisticTransactionDB* otdb = nullptr;
+  const rocksdb::Status s = rocksdb::OptimisticTransactionDB::Open(*db_options,
+      db_path, column_families, &handles, &otdb);
+
+  env->ReleaseStringUTFChars(jdb_path, db_path);
+
+  // check if open operation was successful
+  if (s.ok()) {
+    const jsize resultsLen = 1 + len_cols;  // db handle + column family handles
+    std::unique_ptr<jlong[]> results =
+        std::unique_ptr<jlong[]>(new jlong[resultsLen]);
+    results[0] = reinterpret_cast<jlong>(otdb);
+    for (int i = 1; i <= len_cols; i++) {
+      results[i] = reinterpret_cast<jlong>(handles[i - 1]);
+    }
+
+    jlongArray jresults = env->NewLongArray(resultsLen);
+    if (jresults == nullptr) {
+        // exception thrown: OutOfMemoryError
+        return nullptr;
+    }
+    env->SetLongArrayRegion(jresults, 0, resultsLen, results.get());
+    if (env->ExceptionCheck()) {
+        // exception thrown: ArrayIndexOutOfBoundsException
+        return nullptr;
+    }
+    return jresults;
+  }
+
+  rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+  return nullptr;
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    beginTransaction
+ * Signature: (JJ)J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionDB_beginTransaction__JJ(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_options_handle) {
+  auto* optimistic_txn_db =
+      reinterpret_cast<rocksdb::OptimisticTransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  rocksdb::Transaction* txn =
+      optimistic_txn_db->BeginTransaction(*write_options);
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    beginTransaction
+ * Signature: (JJJ)J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionDB_beginTransaction__JJJ(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_options_handle,
+    jlong joptimistic_txn_options_handle) {
+  auto* optimistic_txn_db =
+      reinterpret_cast<rocksdb::OptimisticTransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  auto* optimistic_txn_options =
+      reinterpret_cast<rocksdb::OptimisticTransactionOptions*>(
+      joptimistic_txn_options_handle);
+  rocksdb::Transaction* txn =
+      optimistic_txn_db->BeginTransaction(*write_options,
+      *optimistic_txn_options);
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    beginTransaction_withOld
+ * Signature: (JJJ)J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionDB_beginTransaction_1withOld__JJJ(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_options_handle,
+    jlong jold_txn_handle) {
+  auto* optimistic_txn_db =
+      reinterpret_cast<rocksdb::OptimisticTransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  auto* old_txn =
+      reinterpret_cast<rocksdb::Transaction*>(
+      jold_txn_handle);
+  rocksdb::OptimisticTransactionOptions optimistic_txn_options;
+  rocksdb::Transaction* txn =
+      optimistic_txn_db->BeginTransaction(*write_options,
+          optimistic_txn_options, old_txn);
+
+  // RocksJava relies on the assumption that
+  // we do not allocate a new Transaction object
+  // when providing an old_optimistic_txn
+  assert(txn == old_txn);
+
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    beginTransaction_withOld
+ * Signature: (JJJJ)J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionDB_beginTransaction_1withOld__JJJJ(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_options_handle,
+    jlong joptimistic_txn_options_handle, jlong jold_txn_handle) {
+  auto* optimistic_txn_db =
+      reinterpret_cast<rocksdb::OptimisticTransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  auto* optimistic_txn_options =
+      reinterpret_cast<rocksdb::OptimisticTransactionOptions*>(
+        joptimistic_txn_options_handle);
+  auto* old_txn =
+      reinterpret_cast<rocksdb::Transaction*>(
+      jold_txn_handle);
+  rocksdb::Transaction* txn =
+      optimistic_txn_db->BeginTransaction(*write_options,
+      *optimistic_txn_options, old_txn);
+
+  // RocksJava relies on the assumption that
+  // we do not allocate a new Transaction object
+  // when providing an old_optimisic_txn
+  assert(txn == old_txn);
+
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    getBaseDB
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionDB_getBaseDB(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* optimistic_txn_db =
+      reinterpret_cast<rocksdb::OptimisticTransactionDB*>(jhandle);
+  return reinterpret_cast<jlong>(optimistic_txn_db->GetBaseDB());
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionDB
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_OptimisticTransactionDB_disposeInternal(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  delete reinterpret_cast<rocksdb::OptimisticTransactionDB*>(jhandle);
+}

java/rocksjni/optimistic_transaction_options.cc

@@ -0,0 +1,72 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the "bridge" between Java and C++
+// for rocksdb::OptimisticTransactionOptions.
+
+#include <jni.h>
+
+#include "include/org_rocksdb_OptimisticTransactionOptions.h"
+
+#include "rocksdb/comparator.h"
+#include "rocksdb/utilities/optimistic_transaction_db.h"
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionOptions
+ * Method:    newOptimisticTransactionOptions
+ * Signature: ()J
+ */
+jlong Java_org_rocksdb_OptimisticTransactionOptions_newOptimisticTransactionOptions(
+    JNIEnv* env, jclass jcls) {
+  rocksdb::OptimisticTransactionOptions* opts =
+      new rocksdb::OptimisticTransactionOptions();
+  return reinterpret_cast<jlong>(opts);
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionOptions
+ * Method:    isSetSnapshot
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_OptimisticTransactionOptions_isSetSnapshot(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts =
+      reinterpret_cast<rocksdb::OptimisticTransactionOptions*>(jhandle);
+  return opts->set_snapshot;
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionOptions
+ * Method:    setSetSnapshot
+ * Signature: (JZ)V
+ */
+void Java_org_rocksdb_OptimisticTransactionOptions_setSetSnapshot(JNIEnv* env,
+    jobject jobj, jlong jhandle, jboolean jset_snapshot) {
+  auto* opts =
+      reinterpret_cast<rocksdb::OptimisticTransactionOptions*>(jhandle);
+  opts->set_snapshot = jset_snapshot;
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionOptions
+ * Method:    setComparator
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_OptimisticTransactionOptions_setComparator(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jcomparator_handle) {
+  auto* opts =
+      reinterpret_cast<rocksdb::OptimisticTransactionOptions*>(jhandle);
+  opts->cmp = reinterpret_cast<rocksdb::Comparator*>(jcomparator_handle);
+}
+
+/*
+ * Class:     org_rocksdb_OptimisticTransactionOptions
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_OptimisticTransactionOptions_disposeInternal(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  delete reinterpret_cast<rocksdb::OptimisticTransactionOptions*>(jhandle);
+}

java/rocksjni/options.cc

@@ -146,19 +146,33 @@ void Java_org_rocksdb_Options_setComparatorHandle__JI(
 /*
  * Class:     org_rocksdb_Options
  * Method:    setComparatorHandle
- * Signature: (JJZ)V
+ * Signature: (JJB)V
  */
-void Java_org_rocksdb_Options_setComparatorHandle__JJZ(
+void Java_org_rocksdb_Options_setComparatorHandle__JJB(
     JNIEnv* env, jobject jobj, jlong jopt_handle, jlong jcomparator_handle,
-    jboolean is_direct) {
-  auto* opt = reinterpret_cast<rocksdb::Options*>(jopt_handle);
-  if(is_direct) {
-    opt->comparator =
-        reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jcomparator_handle);
-  } else {
-    opt->comparator =
-        reinterpret_cast<rocksdb::ComparatorJniCallback*>(jcomparator_handle);
+    jbyte jcomparator_type) {
+  rocksdb::Comparator *comparator = nullptr;
+  switch(jcomparator_type) {
+      // JAVA_COMPARATOR
+      case 0x0:
+        comparator =
+            reinterpret_cast<rocksdb::ComparatorJniCallback*>(jcomparator_handle);
+        break;
+
+      // JAVA_DIRECT_COMPARATOR
+      case 0x1:
+        comparator =
+            reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jcomparator_handle);
+        break;
+
+      // JAVA_NATIVE_COMPARATOR_WRAPPER
+      case 0x2:
+        comparator =
+            reinterpret_cast<rocksdb::Comparator*>(jcomparator_handle);
+        break;
   }
+  auto* opt = reinterpret_cast<rocksdb::Options*>(jopt_handle);
+  opt->comparator = comparator;
 }
 
 /*
@@ -2960,19 +2974,33 @@ void Java_org_rocksdb_ColumnFamilyOptions_setComparatorHandle__JI(
 /*
  * Class:     org_rocksdb_ColumnFamilyOptions
  * Method:    setComparatorHandle
- * Signature: (JJZ)V
+ * Signature: (JJB)V
  */
-void Java_org_rocksdb_ColumnFamilyOptions_setComparatorHandle__JJZ(
+void Java_org_rocksdb_ColumnFamilyOptions_setComparatorHandle__JJB(
     JNIEnv* env, jobject jobj, jlong jopt_handle, jlong jcomparator_handle,
-    jboolean is_direct) {
-  auto* opt = reinterpret_cast<rocksdb::ColumnFamilyOptions*>(jopt_handle);
-  if(is_direct) {
-    opt->comparator =
-        reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jcomparator_handle);
-  } else {
-    opt->comparator =
-        reinterpret_cast<rocksdb::ComparatorJniCallback*>(jcomparator_handle);
+    jbyte jcomparator_type) {
+  rocksdb::Comparator *comparator = nullptr;
+  switch(jcomparator_type) {
+      // JAVA_COMPARATOR
+      case 0x0:
+        comparator =
+            reinterpret_cast<rocksdb::ComparatorJniCallback*>(jcomparator_handle);
+        break;
+
+      // JAVA_DIRECT_COMPARATOR
+      case 0x1:
+        comparator =
+            reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jcomparator_handle);
+        break;
+
+      // JAVA_NATIVE_COMPARATOR_WRAPPER
+      case 0x2:
+        comparator =
+            reinterpret_cast<rocksdb::Comparator*>(jcomparator_handle);
+        break;
   }
+  auto* opt = reinterpret_cast<rocksdb::ColumnFamilyOptions*>(jopt_handle);
+  opt->comparator = comparator;
 }
 
 /*

java/rocksjni/rocks_callback_object.cc

@@ -24,4 +24,5 @@ void Java_org_rocksdb_RocksCallbackObject_disposeInternal(
       //   2) Comparator -> BaseComparatorJniCallback + JniCallback -> ComparatorJniCallback
       // I think this is okay, as Comparator and JniCallback both have virtual destructors...
       delete reinterpret_cast<rocksdb::JniCallback*>(handle);
+// @lint-ignore TXT4 T25377293 Grandfathered in
 }

java/rocksjni/sst_file_writerjni.cc

@@ -20,24 +20,34 @@
 /*
  * Class:     org_rocksdb_SstFileWriter
  * Method:    newSstFileWriter
- * Signature: (JJJZ)J
+ * Signature: (JJJB)J
  */
-jlong Java_org_rocksdb_SstFileWriter_newSstFileWriter__JJJZ(JNIEnv *env,
-    jclass jcls, jlong jenvoptions,  jlong joptions, jlong jcomparator,
-    jboolean is_direct) {
+jlong Java_org_rocksdb_SstFileWriter_newSstFileWriter__JJJB(JNIEnv *env,
+    jclass jcls, jlong jenvoptions,  jlong joptions, jlong jcomparator_handle,
+    jbyte jcomparator_type) {
+  rocksdb::Comparator *comparator = nullptr;
+  switch(jcomparator_type) {
+      // JAVA_COMPARATOR
+      case 0x0:
+        comparator =
+            reinterpret_cast<rocksdb::ComparatorJniCallback*>(jcomparator_handle);
+        break;
+
+      // JAVA_DIRECT_COMPARATOR
+      case 0x1:
+        comparator =
+            reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jcomparator_handle);
+        break;
+
+      // JAVA_NATIVE_COMPARATOR_WRAPPER
+      case 0x2:
+        comparator =
+            reinterpret_cast<rocksdb::Comparator*>(jcomparator_handle);
+        break;
+  }
   auto *env_options =
       reinterpret_cast<const rocksdb::EnvOptions *>(jenvoptions);
   auto *options = reinterpret_cast<const rocksdb::Options *>(joptions);
-
-  rocksdb::Comparator *comparator = nullptr;
-  if(is_direct) {
-    comparator =
-        reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jcomparator);
-  } else {
-    comparator =
-        reinterpret_cast<rocksdb::ComparatorJniCallback*>(jcomparator);
-  }
-
   rocksdb::SstFileWriter *sst_file_writer =
       new rocksdb::SstFileWriter(*env_options, *options, comparator);
   return reinterpret_cast<jlong>(sst_file_writer);

java/rocksjni/statisticsjni.cc

@@ -30,4 +30,5 @@ namespace rocksdb {
 
     return true;
   }
+// @lint-ignore TXT4 T25377293 Grandfathered in
 };

java/rocksjni/statisticsjni.h

@@ -30,4 +30,5 @@ namespace rocksdb {
 
 }  // namespace rocksdb
 
+// @lint-ignore TXT4 T25377293 Grandfathered in
 #endif  // JAVA_ROCKSJNI_STATISTICSJNI_H_

java/rocksjni/transaction_db.cc

@@ -0,0 +1,431 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the "bridge" between Java and C++
+// for rocksdb::TransactionDB.
+
+#include <jni.h>
+#include <functional>
+#include <memory>
+#include <utility>
+
+
+#include "include/org_rocksdb_TransactionDB.h"
+
+#include "rocksdb/options.h"
+#include "rocksdb/utilities/transaction.h"
+#include "rocksdb/utilities/transaction_db.h"
+
+#include "rocksjni/portal.h"
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    open
+ * Signature: (JJLjava/lang/String;)J
+ */
+jlong Java_org_rocksdb_TransactionDB_open__JJLjava_lang_String_2(JNIEnv* env,
+    jclass jcls, jlong joptions_handle, jlong jtxn_db_options_handle,
+    jstring jdb_path) {
+  auto* options = reinterpret_cast<rocksdb::Options*>(joptions_handle);
+  auto* txn_db_options =
+      reinterpret_cast<rocksdb::TransactionDBOptions*>(jtxn_db_options_handle);
+  rocksdb::TransactionDB* tdb = nullptr;
+  const char* db_path = env->GetStringUTFChars(jdb_path, nullptr);
+  if (db_path == nullptr) {
+    // exception thrown: OutOfMemoryError
+    return 0;
+  }
+  rocksdb::Status s =
+      rocksdb::TransactionDB::Open(*options, *txn_db_options, db_path, &tdb);
+  env->ReleaseStringUTFChars(jdb_path, db_path);
+
+  if (s.ok()) {
+    return reinterpret_cast<jlong>(tdb);
+  } else {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+    return 0;
+  }
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    open
+ * Signature: (JJLjava/lang/String;[[B[J)[J
+ */
+jlongArray Java_org_rocksdb_TransactionDB_open__JJLjava_lang_String_2_3_3B_3J(
+    JNIEnv* env, jclass jcls, jlong jdb_options_handle,
+    jlong jtxn_db_options_handle, jstring jdb_path,
+    jobjectArray jcolumn_names,
+    jlongArray jcolumn_options_handles) {
+  const char* db_path = env->GetStringUTFChars(jdb_path, nullptr);
+  if (db_path == nullptr) {
+    // exception thrown: OutOfMemoryError
+    return nullptr;
+  }
+
+  const jsize len_cols = env->GetArrayLength(jcolumn_names);
+  if (env->EnsureLocalCapacity(len_cols) != 0) {
+    // out of memory
+    env->ReleaseStringUTFChars(jdb_path, db_path);
+    return nullptr;
+  }
+
+  jlong* jco = env->GetLongArrayElements(jcolumn_options_handles, nullptr);
+  if (jco == nullptr) {
+      // exception thrown: OutOfMemoryError
+      env->ReleaseStringUTFChars(jdb_path, db_path);
+      return nullptr;
+  }
+  std::vector<rocksdb::ColumnFamilyDescriptor> column_families;
+  for (int i = 0; i < len_cols; i++) {
+    const jobject jcn = env->GetObjectArrayElement(jcolumn_names, i);
+    if (env->ExceptionCheck()) {
+        // exception thrown: ArrayIndexOutOfBoundsException
+        env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+        env->ReleaseStringUTFChars(jdb_path, db_path);
+        return nullptr;
+    }
+    const jbyteArray jcn_ba = reinterpret_cast<jbyteArray>(jcn);
+    jbyte* jcf_name = env->GetByteArrayElements(jcn_ba, nullptr);
+    if (jcf_name == nullptr) {
+      // exception thrown: OutOfMemoryError
+      env->DeleteLocalRef(jcn);
+      env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+      env->ReleaseStringUTFChars(jdb_path, db_path);
+      return nullptr;
+    }
+
+    const int jcf_name_len = env->GetArrayLength(jcn_ba);
+    if (env->EnsureLocalCapacity(jcf_name_len) != 0) {
+        // out of memory
+        env->ReleaseByteArrayElements(jcn_ba, jcf_name, JNI_ABORT);
+        env->DeleteLocalRef(jcn);
+        env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+        env->ReleaseStringUTFChars(jdb_path, db_path);
+        return nullptr;
+    }
+    const std::string cf_name(reinterpret_cast<char *>(jcf_name), jcf_name_len);
+    const rocksdb::ColumnFamilyOptions* cf_options =
+      reinterpret_cast<rocksdb::ColumnFamilyOptions*>(jco[i]);
+    column_families.push_back(
+      rocksdb::ColumnFamilyDescriptor(cf_name, *cf_options));
+
+    env->ReleaseByteArrayElements(jcn_ba, jcf_name, JNI_ABORT);
+    env->DeleteLocalRef(jcn);
+  }
+  env->ReleaseLongArrayElements(jcolumn_options_handles, jco, JNI_ABORT);
+
+  auto* db_options = reinterpret_cast<rocksdb::DBOptions*>(jdb_options_handle);
+  auto* txn_db_options =
+      reinterpret_cast<rocksdb::TransactionDBOptions*>(jtxn_db_options_handle);
+  std::vector<rocksdb::ColumnFamilyHandle*> handles;
+  rocksdb::TransactionDB* tdb = nullptr;
+  const rocksdb::Status s = rocksdb::TransactionDB::Open(*db_options, *txn_db_options,
+      db_path, column_families, &handles, &tdb);
+
+  // check if open operation was successful
+  if (s.ok()) {
+    const jsize resultsLen = 1 + len_cols;  // db handle + column family handles
+    std::unique_ptr<jlong[]> results =
+        std::unique_ptr<jlong[]>(new jlong[resultsLen]);
+    results[0] = reinterpret_cast<jlong>(tdb);
+    for (int i = 1; i <= len_cols; i++) {
+      results[i] = reinterpret_cast<jlong>(handles[i - 1]);
+    }
+
+    jlongArray jresults = env->NewLongArray(resultsLen);
+    if (jresults == nullptr) {
+        // exception thrown: OutOfMemoryError
+        return nullptr;
+    }
+    env->SetLongArrayRegion(jresults, 0, resultsLen, results.get());
+    if (env->ExceptionCheck()) {
+        // exception thrown: ArrayIndexOutOfBoundsException
+        env->DeleteLocalRef(jresults);
+        return nullptr;
+    }
+    return jresults;
+  } else {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+    return nullptr;
+  }
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    beginTransaction
+ * Signature: (JJ)J
+ */
+jlong Java_org_rocksdb_TransactionDB_beginTransaction__JJ(JNIEnv* env,
+    jobject jobj, jlong jhandle, jlong jwrite_options_handle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  rocksdb::Transaction* txn = txn_db->BeginTransaction(*write_options);
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    beginTransaction
+ * Signature: (JJJ)J
+ */
+jlong Java_org_rocksdb_TransactionDB_beginTransaction__JJJ(JNIEnv* env,
+    jobject jobj, jlong jhandle, jlong jwrite_options_handle,
+    jlong jtxn_options_handle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  auto* txn_options =
+      reinterpret_cast<rocksdb::TransactionOptions*>(jtxn_options_handle);
+  rocksdb::Transaction* txn =
+      txn_db->BeginTransaction(*write_options, *txn_options);
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    beginTransaction_withOld
+ * Signature: (JJJ)J
+ */
+jlong Java_org_rocksdb_TransactionDB_beginTransaction_1withOld__JJJ(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_options_handle,
+    jlong jold_txn_handle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  auto* old_txn = reinterpret_cast<rocksdb::Transaction*>(jold_txn_handle);
+  rocksdb::TransactionOptions txn_options;
+  rocksdb::Transaction* txn =
+      txn_db->BeginTransaction(*write_options, txn_options, old_txn);
+
+  // RocksJava relies on the assumption that
+  // we do not allocate a new Transaction object
+  // when providing an old_txn
+  assert(txn == old_txn);
+
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    beginTransaction_withOld
+ * Signature: (JJJJ)J
+ */
+jlong Java_org_rocksdb_TransactionDB_beginTransaction_1withOld__JJJJ(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jwrite_options_handle,
+    jlong jtxn_options_handle, jlong jold_txn_handle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  auto* write_options =
+      reinterpret_cast<rocksdb::WriteOptions*>(jwrite_options_handle);
+  auto* txn_options =
+      reinterpret_cast<rocksdb::TransactionOptions*>(jtxn_options_handle);
+  auto* old_txn = reinterpret_cast<rocksdb::Transaction*>(jold_txn_handle);
+  rocksdb::Transaction* txn = txn_db->BeginTransaction(*write_options,
+      *txn_options, old_txn);
+
+  // RocksJava relies on the assumption that
+  // we do not allocate a new Transaction object
+  // when providing an old_txn
+  assert(txn == old_txn);
+
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    getTransactionByName
+ * Signature: (JLjava/lang/String;)J
+ */
+jlong Java_org_rocksdb_TransactionDB_getTransactionByName(JNIEnv* env,
+    jobject jobj, jlong jhandle, jstring jname) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  const char* name = env->GetStringUTFChars(jname, nullptr);
+  if (name == nullptr) {
+      // exception thrown: OutOfMemoryError
+      return 0;
+  }
+  rocksdb::Transaction* txn = txn_db->GetTransactionByName(name);
+  env->ReleaseStringUTFChars(jname, name);
+  return reinterpret_cast<jlong>(txn);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    getAllPreparedTransactions
+ * Signature: (J)[J
+ */
+jlongArray Java_org_rocksdb_TransactionDB_getAllPreparedTransactions(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  std::vector<rocksdb::Transaction*> txns;
+  txn_db->GetAllPreparedTransactions(&txns);
+
+  const size_t size = txns.size();
+  assert(size < UINT32_MAX);  // does it fit in a jint?
+
+  const jsize len = static_cast<jsize>(size);
+  jlong tmp[len];
+  for (jsize i = 0; i < len; ++i) {
+      tmp[i] = reinterpret_cast<jlong>(txns[i]);
+  }
+
+  jlongArray jtxns = env->NewLongArray(len);
+  if (jtxns == nullptr) {
+      // exception thrown: OutOfMemoryError
+      return nullptr;
+  }
+  env->SetLongArrayRegion(jtxns, 0, len, tmp);
+  if (env->ExceptionCheck()) {
+      // exception thrown: ArrayIndexOutOfBoundsException
+      env->DeleteLocalRef(jtxns);
+      return nullptr;
+  }
+
+  return jtxns;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    getLockStatusData
+ * Signature: (J)Ljava/util/Map;
+ */
+jobject Java_org_rocksdb_TransactionDB_getLockStatusData(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  const std::unordered_multimap<uint32_t, rocksdb::KeyLockInfo> lock_status_data =
+      txn_db->GetLockStatusData();
+  const jobject jlock_status_data = rocksdb::HashMapJni::construct(env,
+        static_cast<uint32_t>(lock_status_data.size()));
+  if (jlock_status_data == nullptr) {
+      // exception occurred
+      return nullptr;
+  }
+
+  const rocksdb::HashMapJni::FnMapKV<const int32_t, const rocksdb::KeyLockInfo> fn_map_kv =
+      [env, txn_db, &lock_status_data](const std::pair<const int32_t, const rocksdb::KeyLockInfo>& pair) {
+          const jobject jlong_column_family_id =
+              rocksdb::LongJni::valueOf(env, pair.first);
+          if (jlong_column_family_id == nullptr) {
+              // an error occurred
+              return std::unique_ptr<std::pair<jobject, jobject>>(nullptr);
+          }
+          const jobject jkey_lock_info =
+              rocksdb::KeyLockInfoJni::construct(env, pair.second);
+          if (jkey_lock_info == nullptr) {
+             // an error occurred
+             return std::unique_ptr<std::pair<jobject, jobject>>(nullptr);
+          }
+          return std::unique_ptr<std::pair<jobject, jobject>>(new std::pair<jobject, jobject>(jlong_column_family_id,
+              jkey_lock_info));
+      };
+
+  if(!rocksdb::HashMapJni::putAll(env, jlock_status_data,
+      lock_status_data.begin(), lock_status_data.end(), fn_map_kv)) {
+    // exception occcurred
+    return nullptr;
+  }
+
+  return jlock_status_data;
+}
+
+/*
+* Class:     org_rocksdb_TransactionDB
+* Method:    getDeadlockInfoBuffer
+* Signature: (J)[Lorg/rocksdb/TransactionDB/DeadlockPath;
+*/
+jobjectArray Java_org_rocksdb_TransactionDB_getDeadlockInfoBuffer(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  const std::vector<rocksdb::DeadlockPath> deadlock_info_buffer =
+      txn_db->GetDeadlockInfoBuffer();
+
+  const jsize deadlock_info_buffer_len =
+      static_cast<jsize>(deadlock_info_buffer.size());
+  jobjectArray jdeadlock_info_buffer =
+      env->NewObjectArray(deadlock_info_buffer_len,
+        rocksdb::DeadlockPathJni::getJClass(env), nullptr);
+  if (jdeadlock_info_buffer == nullptr) {
+      // exception thrown: OutOfMemoryError
+      return nullptr;
+  }
+  jsize jdeadlock_info_buffer_offset = 0;
+
+  auto buf_end = deadlock_info_buffer.end();
+  for (auto buf_it = deadlock_info_buffer.begin(); buf_it != buf_end; ++buf_it) {
+    const rocksdb::DeadlockPath deadlock_path = *buf_it;
+    const std::vector<rocksdb::DeadlockInfo> deadlock_infos
+        = deadlock_path.path;
+    const jsize deadlock_infos_len =
+        static_cast<jsize>(deadlock_info_buffer.size());
+    jobjectArray jdeadlock_infos = env->NewObjectArray(deadlock_infos_len,
+        rocksdb::DeadlockInfoJni::getJClass(env), nullptr);
+    if (jdeadlock_infos == nullptr) {
+        // exception thrown: OutOfMemoryError
+        env->DeleteLocalRef(jdeadlock_info_buffer);
+        return nullptr;
+    }
+    jsize jdeadlock_infos_offset = 0;
+
+    auto infos_end = deadlock_infos.end();
+    for (auto infos_it = deadlock_infos.begin(); infos_it != infos_end; ++infos_it) {
+        const rocksdb::DeadlockInfo deadlock_info = *infos_it;
+        const jobject jdeadlock_info = rocksdb::TransactionDBJni::newDeadlockInfo(
+            env, jobj, deadlock_info.m_txn_id, deadlock_info.m_cf_id,
+            deadlock_info.m_waiting_key, deadlock_info.m_exclusive);
+        if (jdeadlock_info == nullptr) {
+            // exception occcurred
+            env->DeleteLocalRef(jdeadlock_info_buffer);
+            return nullptr;
+        }
+        env->SetObjectArrayElement(jdeadlock_infos, jdeadlock_infos_offset++, jdeadlock_info);
+        if (env->ExceptionCheck()) {
+            // exception thrown: ArrayIndexOutOfBoundsException or ArrayStoreException
+            env->DeleteLocalRef(jdeadlock_info);
+            env->DeleteLocalRef(jdeadlock_info_buffer);
+            return nullptr;
+        }
+    }
+
+    const jobject jdeadlock_path =
+        rocksdb::DeadlockPathJni::construct(env, jdeadlock_infos,
+            deadlock_path.limit_exceeded);
+    if(jdeadlock_path == nullptr) {
+        // exception occcurred
+        env->DeleteLocalRef(jdeadlock_info_buffer);
+        return nullptr;
+    }
+    env->SetObjectArrayElement(jdeadlock_info_buffer, jdeadlock_info_buffer_offset++, jdeadlock_path);
+    if (env->ExceptionCheck()) {
+        // exception thrown: ArrayIndexOutOfBoundsException or ArrayStoreException
+        env->DeleteLocalRef(jdeadlock_path);
+        env->DeleteLocalRef(jdeadlock_info_buffer);
+        return nullptr;
+    }
+  }
+
+  return jdeadlock_info_buffer;
+}
+
+/*
+* Class:     org_rocksdb_TransactionDB
+* Method:    setDeadlockInfoBufferSize
+* Signature: (JI)V
+*/
+void Java_org_rocksdb_TransactionDB_setDeadlockInfoBufferSize(
+    JNIEnv* env, jobject jobj, jlong jhandle, jint jdeadlock_info_buffer_size) {
+  auto* txn_db = reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+  txn_db->SetDeadlockInfoBufferSize(jdeadlock_info_buffer_size);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDB
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_TransactionDB_disposeInternal(JNIEnv* env, jobject jobj,
+    jlong jhandle) {
+  delete reinterpret_cast<rocksdb::TransactionDB*>(jhandle);
+}

java/rocksjni/transaction_db_options.cc

@@ -0,0 +1,147 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the "bridge" between Java and C++
+// for rocksdb::TransactionDBOptions.
+
+#include <jni.h>
+
+#include "include/org_rocksdb_TransactionDBOptions.h"
+
+#include "rocksdb/utilities/transaction_db.h"
+
+#include "rocksjni/portal.h"
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    newTransactionDBOptions
+ * Signature: ()J
+ */
+jlong Java_org_rocksdb_TransactionDBOptions_newTransactionDBOptions(
+    JNIEnv* env, jclass jcls) {
+  rocksdb::TransactionDBOptions* opts = new rocksdb::TransactionDBOptions();
+  return reinterpret_cast<jlong>(opts);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    getMaxNumLocks
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionDBOptions_getMaxNumLocks(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  return opts->max_num_locks;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    setMaxNumLocks
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_TransactionDBOptions_setMaxNumLocks(JNIEnv* env,
+    jobject jobj, jlong jhandle, jlong jmax_num_locks) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  opts->max_num_locks = jmax_num_locks;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    getNumStripes
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionDBOptions_getNumStripes(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  return opts->num_stripes;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    setNumStripes
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_TransactionDBOptions_setNumStripes(JNIEnv* env,
+    jobject jobj, jlong jhandle, jlong jnum_stripes) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  opts->num_stripes = jnum_stripes;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    getTransactionLockTimeout
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionDBOptions_getTransactionLockTimeout(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  return opts->transaction_lock_timeout;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    setTransactionLockTimeout
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_TransactionDBOptions_setTransactionLockTimeout(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jtransaction_lock_timeout) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  opts->transaction_lock_timeout = jtransaction_lock_timeout;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    getDefaultLockTimeout
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionDBOptions_getDefaultLockTimeout(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  return opts->default_lock_timeout;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    setDefaultLockTimeout
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_TransactionDBOptions_setDefaultLockTimeout(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jdefault_lock_timeout) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  opts->default_lock_timeout = jdefault_lock_timeout;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    getWritePolicy
+ * Signature: (J)B
+ */
+jbyte Java_org_rocksdb_TransactionDBOptions_getWritePolicy(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  return rocksdb::TxnDBWritePolicyJni::toJavaTxnDBWritePolicy(opts->write_policy);
+}
+
+/*
+* Class:     org_rocksdb_TransactionDBOptions
+* Method:    setWritePolicy
+* Signature: (JB)V
+*/
+void Java_org_rocksdb_TransactionDBOptions_setWritePolicy(
+    JNIEnv* env, jobject jobj, jlong jhandle, jbyte jwrite_policy) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+  opts->write_policy =
+      rocksdb::TxnDBWritePolicyJni::toCppTxnDBWritePolicy(jwrite_policy);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionDBOptions
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_TransactionDBOptions_disposeInternal(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  delete reinterpret_cast<rocksdb::TransactionDBOptions*>(jhandle);
+}

java/rocksjni/transaction_notifier.cc

@@ -0,0 +1,42 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the "bridge" between Java and C++
+// for rocksdb::TransactionNotifier.
+
+#include <jni.h>
+
+#include "include/org_rocksdb_AbstractTransactionNotifier.h"
+#include "rocksjni/transaction_notifier_jnicallback.h"
+
+/*
+ * Class:     org_rocksdb_AbstractTransactionNotifier
+ * Method:    createNewTransactionNotifier
+ * Signature: ()J
+ */
+jlong Java_org_rocksdb_AbstractTransactionNotifier_createNewTransactionNotifier(
+    JNIEnv* env, jobject jobj) {
+  auto* transaction_notifier =
+      new rocksdb::TransactionNotifierJniCallback(env, jobj);
+  auto* sptr_transaction_notifier =
+      new std::shared_ptr<rocksdb::TransactionNotifierJniCallback>(
+          transaction_notifier);
+  return reinterpret_cast<jlong>(sptr_transaction_notifier);
+}
+
+/*
+ * Class:     org_rocksdb_AbstractTransactionNotifier
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_AbstractTransactionNotifier_disposeInternal(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  // TODO(AR) refactor to use JniCallback::JniCallback
+  // when https://github.com/facebook/rocksdb/pull/1241/ is merged
+  std::shared_ptr<rocksdb::TransactionNotifierJniCallback>* handle =
+      reinterpret_cast<std::shared_ptr<
+          rocksdb::TransactionNotifierJniCallback>*>(jhandle);
+  delete handle;
+}

java/rocksjni/transaction_notifier_jnicallback.cc

@@ -0,0 +1,39 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the callback "bridge" between Java and C++ for
+// rocksdb::TransactionNotifier.
+
+#include "rocksjni/transaction_notifier_jnicallback.h"
+#include "rocksjni/portal.h"
+
+namespace rocksdb {
+
+TransactionNotifierJniCallback::TransactionNotifierJniCallback(JNIEnv* env,
+    jobject jtransaction_notifier) : JniCallback(env, jtransaction_notifier) {
+  // we cache the method id for the JNI callback
+  m_jsnapshot_created_methodID =
+      AbstractTransactionNotifierJni::getSnapshotCreatedMethodId(env);
+}
+
+void TransactionNotifierJniCallback::SnapshotCreated(
+    const Snapshot* newSnapshot) {
+  jboolean attached_thread = JNI_FALSE;
+  JNIEnv* env = getJniEnv(&attached_thread);
+  assert(env != nullptr);
+
+  env->CallVoidMethod(m_jcallback_obj,
+      m_jsnapshot_created_methodID, reinterpret_cast<jlong>(newSnapshot));
+
+  if(env->ExceptionCheck()) {
+    // exception thrown from CallVoidMethod
+    env->ExceptionDescribe();  // print out exception to stderr
+    releaseJniEnv(attached_thread);
+    return;
+  }
+
+  releaseJniEnv(attached_thread);
+}
+}  // namespace rocksdb

java/rocksjni/transaction_notifier_jnicallback.h

@@ -0,0 +1,42 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the callback "bridge" between Java and C++ for
+// rocksdb::TransactionNotifier.
+
+#ifndef JAVA_ROCKSJNI_TRANSACTION_NOTIFIER_JNICALLBACK_H_
+#define JAVA_ROCKSJNI_TRANSACTION_NOTIFIER_JNICALLBACK_H_
+
+#include <jni.h>
+
+#include "rocksdb/utilities/transaction.h"
+#include "rocksjni/jnicallback.h"
+
+namespace rocksdb {
+
+/**
+ * This class acts as a bridge between C++
+ * and Java. The methods in this class will be
+ * called back from the RocksDB TransactionDB or OptimisticTransactionDB (C++),
+ * we then callback to the appropriate Java method
+ * this enables TransactionNotifier to be implemented in Java.
+ *
+ * Unlike RocksJava's Comparator JNI Callback, we do not attempt
+ * to reduce Java object allocations by caching the Snapshot object
+ * presented to the callback. This could be revisited in future
+ * if performance is lacking.
+ */
+class TransactionNotifierJniCallback: public JniCallback,
+    public TransactionNotifier {
+ public:
+  TransactionNotifierJniCallback(JNIEnv* env, jobject jtransaction_notifier);
+  virtual void SnapshotCreated(const Snapshot* newSnapshot);
+
+ private:
+  jmethodID m_jsnapshot_created_methodID;
+};
+}  // namespace rocksdb
+
+#endif  // JAVA_ROCKSJNI_TRANSACTION_NOTIFIER_JNICALLBACK_H_

java/rocksjni/transaction_options.cc

@@ -0,0 +1,166 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// This file implements the "bridge" between Java and C++
+// for rocksdb::TransactionOptions.
+
+#include <jni.h>
+
+#include "include/org_rocksdb_TransactionOptions.h"
+
+#include "rocksdb/utilities/transaction_db.h"
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    newTransactionOptions
+ * Signature: ()J
+ */
+jlong Java_org_rocksdb_TransactionOptions_newTransactionOptions(JNIEnv* env,
+    jclass jcls) {
+  auto* opts = new rocksdb::TransactionOptions();
+  return reinterpret_cast<jlong>(opts);
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    isSetSnapshot
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_TransactionOptions_isSetSnapshot(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  return opts->set_snapshot;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    setSetSnapshot
+ * Signature: (JZ)V
+ */
+void Java_org_rocksdb_TransactionOptions_setSetSnapshot(JNIEnv* env,
+    jobject jobj, jlong jhandle, jboolean jset_snapshot) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  opts->set_snapshot = jset_snapshot;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    isDeadlockDetect
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_TransactionOptions_isDeadlockDetect(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  return opts->deadlock_detect;
+}
+
+/*
+* Class:     org_rocksdb_TransactionOptions
+* Method:    setDeadlockDetect
+* Signature: (JZ)V
+*/
+void Java_org_rocksdb_TransactionOptions_setDeadlockDetect(
+    JNIEnv* env, jobject jobj, jlong jhandle, jboolean jdeadlock_detect) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  opts->deadlock_detect = jdeadlock_detect;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    getLockTimeout
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionOptions_getLockTimeout(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  return opts->lock_timeout;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    setLockTimeout
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_TransactionOptions_setLockTimeout(JNIEnv* env,
+    jobject jobj, jlong jhandle, jlong jlock_timeout) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  opts->lock_timeout = jlock_timeout;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    getExpiration
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionOptions_getExpiration(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  return opts->expiration;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    setExpiration
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_TransactionOptions_setExpiration(JNIEnv* env,
+    jobject jobj, jlong jhandle, jlong jexpiration) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  opts->expiration = jexpiration;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    getDeadlockDetectDepth
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_TransactionOptions_getDeadlockDetectDepth(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  return opts->deadlock_detect_depth;
+}
+
+/*
+* Class:     org_rocksdb_TransactionOptions
+* Method:    setDeadlockDetectDepth
+* Signature: (JJ)V
+*/
+void Java_org_rocksdb_TransactionOptions_setDeadlockDetectDepth(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jdeadlock_detect_depth) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  opts->deadlock_detect_depth = jdeadlock_detect_depth;
+}
+
+/*
+* Class:     org_rocksdb_TransactionOptions
+* Method:    getMaxWriteBatchSize
+* Signature: (J)J
+*/
+jlong Java_org_rocksdb_TransactionOptions_getMaxWriteBatchSize(
+    JNIEnv* env, jobject jobj, jlong jhandle) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  return opts->max_write_batch_size;
+}
+
+/*
+* Class:     org_rocksdb_TransactionOptions
+* Method:    setMaxWriteBatchSize
+* Signature: (JJ)V
+*/
+void Java_org_rocksdb_TransactionOptions_setMaxWriteBatchSize(
+    JNIEnv* env, jobject jobj, jlong jhandle, jlong jmax_write_batch_size) {
+  auto* opts = reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+  opts->max_write_batch_size = jmax_write_batch_size;
+}
+
+/*
+ * Class:     org_rocksdb_TransactionOptions
+ * Method:    disposeInternal
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_TransactionOptions_disposeInternal(JNIEnv* env,
+    jobject jobj, jlong jhandle) {
+  delete reinterpret_cast<rocksdb::TransactionOptions*>(jhandle);
+}

java/rocksjni/write_batch.cc

@@ -27,12 +27,34 @@
  * Method:    newWriteBatch
  * Signature: (I)J
  */
-jlong Java_org_rocksdb_WriteBatch_newWriteBatch(
+jlong Java_org_rocksdb_WriteBatch_newWriteBatch__I(
     JNIEnv* env, jclass jcls, jint jreserved_bytes) {
   auto* wb = new rocksdb::WriteBatch(static_cast<size_t>(jreserved_bytes));
   return reinterpret_cast<jlong>(wb);
 }
 
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    newWriteBatch
+ * Signature: ([BI)J
+ */
+jlong Java_org_rocksdb_WriteBatch_newWriteBatch___3BI(
+    JNIEnv* env, jclass jcls, jbyteArray jserialized,
+    jint jserialized_length) {
+  jboolean has_exception = JNI_FALSE;
+  std::string serialized = rocksdb::JniUtil::byteString<std::string>(env,
+    jserialized, jserialized_length,
+    [](const char* str, const size_t len) { return std::string(str, len); },
+    &has_exception);
+  if(has_exception == JNI_TRUE) {
+    // exception occurred
+    return 0;
+  }
+
+  auto* wb = new rocksdb::WriteBatch(serialized);
+  return reinterpret_cast<jlong>(wb);
+}
+
 /*
  * Class:     org_rocksdb_WriteBatch
  * Method:    count0
@@ -90,6 +112,37 @@ void Java_org_rocksdb_WriteBatch_rollbackToSavePoint0(
   rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
 }
 
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    popSavePoint
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_WriteBatch_popSavePoint(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  auto s = wb->PopSavePoint();
+
+  if (s.ok()) {
+    return;
+  }
+  rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    setMaxBytes
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_WriteBatch_setMaxBytes(
+    JNIEnv* env, jobject jobj, jlong jwb_handle, jlong jmax_bytes) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  wb->SetMaxBytes(static_cast<size_t>(jmax_bytes));
+}
+
 /*
  * Class:     org_rocksdb_WriteBatch
  * Method:    put
@@ -102,10 +155,13 @@ void Java_org_rocksdb_WriteBatch_put__J_3BI_3BI(
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
   assert(wb != nullptr);
   auto put = [&wb] (rocksdb::Slice key, rocksdb::Slice value) {
-    wb->Put(key, value);
+    return wb->Put(key, value);
   };
-  rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(put, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -122,10 +178,13 @@ void Java_org_rocksdb_WriteBatch_put__J_3BI_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto put = [&wb, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) {
-    wb->Put(cf_handle, key, value);
+    return wb->Put(cf_handle, key, value);
   };
-  rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(put, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -140,10 +199,13 @@ void Java_org_rocksdb_WriteBatch_merge__J_3BI_3BI(
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
   assert(wb != nullptr);
   auto merge = [&wb] (rocksdb::Slice key, rocksdb::Slice value) {
-    wb->Merge(key, value);
+    return wb->Merge(key, value);
   };
-  rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(merge, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -160,34 +222,41 @@ void Java_org_rocksdb_WriteBatch_merge__J_3BI_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto merge = [&wb, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) {
-    wb->Merge(cf_handle, key, value);
+    return wb->Merge(cf_handle, key, value);
   };
-  rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(merge, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
  * Class:     org_rocksdb_WriteBatch
- * Method:    remove
+ * Method:    delete
  * Signature: (J[BI)V
  */
-void Java_org_rocksdb_WriteBatch_remove__J_3BI(
+void Java_org_rocksdb_WriteBatch_delete__J_3BI(
     JNIEnv* env, jobject jobj, jlong jwb_handle,
     jbyteArray jkey, jint jkey_len) {
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
   assert(wb != nullptr);
   auto remove = [&wb] (rocksdb::Slice key) {
-    wb->Delete(key);
+    return wb->Delete(key);
   };
-  rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(remove, env,
+      jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
  * Class:     org_rocksdb_WriteBatch
- * Method:    remove
+ * Method:    delete
  * Signature: (J[BIJ)V
  */
-void Java_org_rocksdb_WriteBatch_remove__J_3BIJ(
+void Java_org_rocksdb_WriteBatch_delete__J_3BIJ(
     JNIEnv* env, jobject jobj, jlong jwb_handle,
     jbyteArray jkey, jint jkey_len, jlong jcf_handle) {
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
@@ -195,9 +264,55 @@ void Java_org_rocksdb_WriteBatch_remove__J_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto remove = [&wb, &cf_handle] (rocksdb::Slice key) {
-    wb->Delete(cf_handle, key);
+    return wb->Delete(cf_handle, key);
   };
-  rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(remove, env,
+      jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    singleDelete
+ * Signature: (J[BI)V
+ */
+void Java_org_rocksdb_WriteBatch_singleDelete__J_3BI(
+    JNIEnv* env, jobject jobj, jlong jwb_handle, jbyteArray jkey,
+    jint jkey_len) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+  auto single_delete = [&wb] (rocksdb::Slice key) {
+    return wb->SingleDelete(key);
+  };
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(single_delete,
+      env, jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    singleDelete
+ * Signature: (J[BIJ)V
+ */
+void Java_org_rocksdb_WriteBatch_singleDelete__J_3BIJ(
+    JNIEnv* env, jobject jobj, jlong jwb_handle, jbyteArray jkey,
+    jint jkey_len, jlong jcf_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+  auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
+  assert(cf_handle != nullptr);
+  auto single_delete = [&wb, &cf_handle] (rocksdb::Slice key) {
+    return wb->SingleDelete(cf_handle, key);
+  };
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(single_delete,
+      env, jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -205,19 +320,20 @@ void Java_org_rocksdb_WriteBatch_remove__J_3BIJ(
  * Method:    deleteRange
  * Signature: (J[BI[BI)V
  */
-JNIEXPORT void JNICALL Java_org_rocksdb_WriteBatch_deleteRange__J_3BI_3BI(
-    JNIEnv*, jobject, jlong, jbyteArray, jint, jbyteArray, jint);
-
 void Java_org_rocksdb_WriteBatch_deleteRange__J_3BI_3BI(
     JNIEnv* env, jobject jobj, jlong jwb_handle, jbyteArray jbegin_key,
     jint jbegin_key_len, jbyteArray jend_key, jint jend_key_len) {
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
   assert(wb != nullptr);
   auto deleteRange = [&wb](rocksdb::Slice beginKey, rocksdb::Slice endKey) {
-    wb->DeleteRange(beginKey, endKey);
+    return wb->DeleteRange(beginKey, endKey);
   };
-  rocksdb::JniUtil::kv_op(deleteRange, env, jobj, jbegin_key, jbegin_key_len,
-                          jend_key, jend_key_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(
+      deleteRange, env, jobj, jbegin_key, jbegin_key_len, jend_key,
+      jend_key_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -234,11 +350,15 @@ void Java_org_rocksdb_WriteBatch_deleteRange__J_3BI_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto deleteRange = [&wb, &cf_handle](rocksdb::Slice beginKey,
-                                       rocksdb::Slice endKey) {
-    wb->DeleteRange(cf_handle, beginKey, endKey);
+      rocksdb::Slice endKey) {
+    return wb->DeleteRange(cf_handle, beginKey, endKey);
   };
-  rocksdb::JniUtil::kv_op(deleteRange, env, jobj, jbegin_key, jbegin_key_len,
-                          jend_key, jend_key_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(
+      deleteRange, env, jobj, jbegin_key, jbegin_key_len, jend_key,
+      jend_key_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -252,9 +372,13 @@ void Java_org_rocksdb_WriteBatch_putLogData(
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
   assert(wb != nullptr);
   auto putLogData = [&wb] (rocksdb::Slice blob) {
-    wb->PutLogData(blob);
+    return wb->PutLogData(blob);
   };
-  rocksdb::JniUtil::k_op(putLogData, env, jobj, jblob, jblob_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(putLogData,
+      env, jobj, jblob, jblob_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -263,7 +387,7 @@ void Java_org_rocksdb_WriteBatch_putLogData(
  * Signature: (JJ)V
  */
 void Java_org_rocksdb_WriteBatch_iterate(
-    JNIEnv* env , jobject jobj, jlong jwb_handle, jlong handlerHandle) {
+    JNIEnv* env, jobject jobj, jlong jwb_handle, jlong handlerHandle) {
   auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
   assert(wb != nullptr);
 
@@ -276,6 +400,178 @@ void Java_org_rocksdb_WriteBatch_iterate(
   rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
 }
 
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    data
+ * Signature: (J)[B
+ */
+jbyteArray Java_org_rocksdb_WriteBatch_data(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  auto data = wb->Data();
+  return rocksdb::JniUtil::copyBytes(env, data);
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    getDataSize
+ * Signature: (J)J
+ */
+jlong Java_org_rocksdb_WriteBatch_getDataSize(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  auto data_size = wb->GetDataSize();
+  return static_cast<jlong>(data_size);
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasPut
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_WriteBatch_hasPut(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasPut();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasDelete
+ * Signature: (J)Z
+ */
+jboolean Java_org_rocksdb_WriteBatch_hasDelete(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasDelete();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasSingleDelete
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasSingleDelete(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasSingleDelete();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasDeleteRange
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasDeleteRange(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasDeleteRange();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasMerge
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasMerge(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasMerge();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasBeginPrepare
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasBeginPrepare(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasBeginPrepare();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasEndPrepare
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasEndPrepare(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasEndPrepare();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasCommit
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasCommit(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasCommit();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    hasRollback
+ * Signature: (J)Z
+ */
+JNIEXPORT jboolean JNICALL Java_org_rocksdb_WriteBatch_hasRollback(
+    JNIEnv* env , jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  return wb->HasRollback();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    markWalTerminationPoint
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_WriteBatch_markWalTerminationPoint(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  wb->MarkWalTerminationPoint();
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatch
+ * Method:    getWalTerminationPoint
+ * Signature: (J)Lorg/rocksdb/WriteBatch/SavePoint;
+ */
+jobject Java_org_rocksdb_WriteBatch_getWalTerminationPoint(
+    JNIEnv* env, jobject jobj, jlong jwb_handle) {
+  auto* wb = reinterpret_cast<rocksdb::WriteBatch*>(jwb_handle);
+  assert(wb != nullptr);
+
+  auto save_point = wb->GetWalTerminationPoint();
+  return rocksdb::WriteBatchSavePointJni::construct(env, save_point);
+}
+
 /*
  * Class:     org_rocksdb_WriteBatch
  * Method:    disposeInternal

java/rocksjni/write_batch_test.cc

@@ -87,8 +87,32 @@ jbyteArray Java_org_rocksdb_WriteBatchTest_getContents(
         state.append(")");
         count++;
         break;
+      case rocksdb::kTypeSingleDeletion:
+        state.append("SingleDelete(");
+        state.append(ikey.user_key.ToString());
+        state.append(")");
+        count++;
+        break;
+      case rocksdb::kTypeRangeDeletion:
+        state.append("DeleteRange(");
+        state.append(ikey.user_key.ToString());
+        state.append(", ");
+        state.append(iter->value().ToString());
+        state.append(")");
+        count++;
+        break;
+      case rocksdb::kTypeLogData:
+        state.append("LogData(");
+        state.append(ikey.user_key.ToString());
+        state.append(")");
+        count++;
+        break;
       default:
         assert(false);
+        state.append("Err:Expected(");
+        state.append(std::to_string(ikey.type));
+        state.append(")");
+        count++;
         break;
     }
     state.append("@");
@@ -96,8 +120,12 @@ jbyteArray Java_org_rocksdb_WriteBatchTest_getContents(
   }
   if (!s.ok()) {
     state.append(s.ToString());
-  } else if (count != rocksdb::WriteBatchInternal::Count(b)) {
-    state.append("CountMismatch()");
+  } else if (rocksdb::WriteBatchInternal::Count(b) != count) {
+    state.append("Err:CountMismatch(expected=");
+    state.append(std::to_string(rocksdb::WriteBatchInternal::Count(b)));
+    state.append(", actual=");
+    state.append(std::to_string(count));
+    state.append(")");
   }
   delete mem->Unref();
 

java/rocksjni/write_batch_with_index.cc

@@ -39,19 +39,31 @@ jlong Java_org_rocksdb_WriteBatchWithIndex_newWriteBatchWithIndex__Z(
 /*
  * Class:     org_rocksdb_WriteBatchWithIndex
  * Method:    newWriteBatchWithIndex
- * Signature: (JZIZ)J
+ * Signature: (JBIZ)J
  */
-jlong Java_org_rocksdb_WriteBatchWithIndex_newWriteBatchWithIndex__JZIZ(
+jlong Java_org_rocksdb_WriteBatchWithIndex_newWriteBatchWithIndex__JBIZ(
     JNIEnv* env, jclass jcls, jlong jfallback_index_comparator_handle,
-    jboolean is_direct, jint jreserved_bytes, jboolean joverwrite_key) {
-rocksdb::Comparator *fallback_comparator = nullptr;
-if(is_direct) {
-  fallback_comparator =
-      reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jfallback_index_comparator_handle);
-} else {
-  fallback_comparator =
-      reinterpret_cast<rocksdb::ComparatorJniCallback*>(jfallback_index_comparator_handle);
-}
+    jbyte jcomparator_type, jint jreserved_bytes, jboolean joverwrite_key) {
+  rocksdb::Comparator *fallback_comparator = nullptr;
+  switch(jcomparator_type) {
+      // JAVA_COMPARATOR
+      case 0x0:
+        fallback_comparator =
+            reinterpret_cast<rocksdb::ComparatorJniCallback*>(jfallback_index_comparator_handle);
+        break;
+
+      // JAVA_DIRECT_COMPARATOR
+      case 0x1:
+        fallback_comparator =
+            reinterpret_cast<rocksdb::DirectComparatorJniCallback*>(jfallback_index_comparator_handle);
+        break;
+
+      // JAVA_NATIVE_COMPARATOR_WRAPPER
+      case 0x2:
+        fallback_comparator =
+            reinterpret_cast<rocksdb::Comparator*>(jfallback_index_comparator_handle);
+        break;
+  }
   auto* wbwi =
       new rocksdb::WriteBatchWithIndex(
           fallback_comparator,
@@ -83,10 +95,13 @@ void Java_org_rocksdb_WriteBatchWithIndex_put__J_3BI_3BI(
   auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
   assert(wbwi != nullptr);
   auto put = [&wbwi] (rocksdb::Slice key, rocksdb::Slice value) {
-    wbwi->Put(key, value);
+    return wbwi->Put(key, value);
   };
-  rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(put, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -103,10 +118,13 @@ void Java_org_rocksdb_WriteBatchWithIndex_put__J_3BI_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto put = [&wbwi, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) {
-    wbwi->Put(cf_handle, key, value);
+    return wbwi->Put(cf_handle, key, value);
   };
-  rocksdb::JniUtil::kv_op(put, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(put, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -120,10 +138,13 @@ void Java_org_rocksdb_WriteBatchWithIndex_merge__J_3BI_3BI(
   auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
   assert(wbwi != nullptr);
   auto merge = [&wbwi] (rocksdb::Slice key, rocksdb::Slice value) {
-    wbwi->Merge(key, value);
+    return wbwi->Merge(key, value);
   };
-  rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(merge, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -140,34 +161,41 @@ void Java_org_rocksdb_WriteBatchWithIndex_merge__J_3BI_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto merge = [&wbwi, &cf_handle] (rocksdb::Slice key, rocksdb::Slice value) {
-    wbwi->Merge(cf_handle, key, value);
+    return wbwi->Merge(cf_handle, key, value);
   };
-  rocksdb::JniUtil::kv_op(merge, env, jobj, jkey, jkey_len, jentry_value,
-      jentry_value_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(merge, env,
+      jobj, jkey, jkey_len, jentry_value, jentry_value_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
  * Class:     org_rocksdb_WriteBatchWithIndex
- * Method:    remove
+ * Method:    delete
  * Signature: (J[BI)V
  */
-void Java_org_rocksdb_WriteBatchWithIndex_remove__J_3BI(
+void Java_org_rocksdb_WriteBatchWithIndex_delete__J_3BI(
     JNIEnv* env, jobject jobj, jlong jwbwi_handle, jbyteArray jkey,
     jint jkey_len) {
   auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
   assert(wbwi != nullptr);
   auto remove = [&wbwi] (rocksdb::Slice key) {
-    wbwi->Delete(key);
+    return wbwi->Delete(key);
   };
-  rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(remove, env,
+      jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
  * Class:     org_rocksdb_WriteBatchWithIndex
- * Method:    remove
+ * Method:    delete
  * Signature: (J[BIJ)V
  */
-void Java_org_rocksdb_WriteBatchWithIndex_remove__J_3BIJ(
+void Java_org_rocksdb_WriteBatchWithIndex_delete__J_3BIJ(
     JNIEnv* env, jobject jobj, jlong jwbwi_handle, jbyteArray jkey,
     jint jkey_len, jlong jcf_handle) {
   auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
@@ -175,9 +203,55 @@ void Java_org_rocksdb_WriteBatchWithIndex_remove__J_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto remove = [&wbwi, &cf_handle] (rocksdb::Slice key) {
-    wbwi->Delete(cf_handle, key);
+    return wbwi->Delete(cf_handle, key);
   };
-  rocksdb::JniUtil::k_op(remove, env, jobj, jkey, jkey_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(remove, env,
+      jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatchWithIndex
+ * Method:    singleDelete
+ * Signature: (J[BI)V
+ */
+void Java_org_rocksdb_WriteBatchWithIndex_singleDelete__J_3BI(
+    JNIEnv* env, jobject jobj, jlong jwbwi_handle, jbyteArray jkey,
+    jint jkey_len) {
+  auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
+  assert(wbwi != nullptr);
+  auto single_delete = [&wbwi] (rocksdb::Slice key) {
+    return wbwi->SingleDelete(key);
+  };
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(single_delete,
+      env, jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatchWithIndex
+ * Method:    singleDelete
+ * Signature: (J[BIJ)V
+ */
+void Java_org_rocksdb_WriteBatchWithIndex_singleDelete__J_3BIJ(
+  JNIEnv* env, jobject jobj, jlong jwbwi_handle, jbyteArray jkey,
+  jint jkey_len, jlong jcf_handle) {
+  auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
+  assert(wbwi != nullptr);
+  auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
+  assert(cf_handle != nullptr);
+  auto single_delete = [&wbwi, &cf_handle] (rocksdb::Slice key) {
+    return wbwi->SingleDelete(cf_handle, key);
+  };
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(single_delete,
+      env, jobj, jkey, jkey_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -191,10 +265,14 @@ void Java_org_rocksdb_WriteBatchWithIndex_deleteRange__J_3BI_3BI(
   auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
   assert(wbwi != nullptr);
   auto deleteRange = [&wbwi](rocksdb::Slice beginKey, rocksdb::Slice endKey) {
-    wbwi->DeleteRange(beginKey, endKey);
+    return wbwi->DeleteRange(beginKey, endKey);
   };
-  rocksdb::JniUtil::kv_op(deleteRange, env, jobj, jbegin_key, jbegin_key_len,
-                          jend_key, jend_key_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(
+      deleteRange, env, jobj, jbegin_key, jbegin_key_len, jend_key,
+      jend_key_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -211,11 +289,15 @@ void Java_org_rocksdb_WriteBatchWithIndex_deleteRange__J_3BI_3BIJ(
   auto* cf_handle = reinterpret_cast<rocksdb::ColumnFamilyHandle*>(jcf_handle);
   assert(cf_handle != nullptr);
   auto deleteRange = [&wbwi, &cf_handle](rocksdb::Slice beginKey,
-                                         rocksdb::Slice endKey) {
-    wbwi->DeleteRange(cf_handle, beginKey, endKey);
+      rocksdb::Slice endKey) {
+    return wbwi->DeleteRange(cf_handle, beginKey, endKey);
   };
-  rocksdb::JniUtil::kv_op(deleteRange, env, jobj, jbegin_key, jbegin_key_len,
-                          jend_key, jend_key_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::kv_op(
+      deleteRange, env, jobj, jbegin_key, jbegin_key_len, jend_key,
+      jend_key_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -229,9 +311,13 @@ void Java_org_rocksdb_WriteBatchWithIndex_putLogData(
   auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
   assert(wbwi != nullptr);
   auto putLogData = [&wbwi] (rocksdb::Slice blob) {
-    wbwi->PutLogData(blob);
+    return wbwi->PutLogData(blob);
   };
-  rocksdb::JniUtil::k_op(putLogData, env, jobj, jblob, jblob_len);
+  std::unique_ptr<rocksdb::Status> status = rocksdb::JniUtil::k_op(putLogData,
+      env, jobj, jblob, jblob_len);
+  if (status != nullptr && !status->ok()) {
+    rocksdb::RocksDBExceptionJni::ThrowNew(env, status);
+  }
 }
 
 /*
@@ -279,6 +365,54 @@ void Java_org_rocksdb_WriteBatchWithIndex_rollbackToSavePoint0(
   rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
 }
 
+/*
+ * Class:     org_rocksdb_WriteBatchWithIndex
+ * Method:    popSavePoint
+ * Signature: (J)V
+ */
+void Java_org_rocksdb_WriteBatchWithIndex_popSavePoint(
+    JNIEnv* env, jobject jobj, jlong jwbwi_handle) {
+  auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
+  assert(wbwi != nullptr);
+
+  auto s = wbwi->PopSavePoint();
+
+  if (s.ok()) {
+    return;
+  }
+
+  rocksdb::RocksDBExceptionJni::ThrowNew(env, s);
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatchWithIndex
+ * Method:    setMaxBytes
+ * Signature: (JJ)V
+ */
+void Java_org_rocksdb_WriteBatchWithIndex_setMaxBytes(
+    JNIEnv* env, jobject jobj, jlong jwbwi_handle, jlong jmax_bytes) {
+  auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
+  assert(wbwi != nullptr);
+
+  wbwi->SetMaxBytes(static_cast<size_t>(jmax_bytes));
+}
+
+/*
+ * Class:     org_rocksdb_WriteBatchWithIndex
+ * Method:    getWriteBatch
+ * Signature: (J)Lorg/rocksdb/WriteBatch;
+ */
+jobject Java_org_rocksdb_WriteBatchWithIndex_getWriteBatch(
+    JNIEnv* env, jobject jobj, jlong jwbwi_handle) {
+  auto* wbwi = reinterpret_cast<rocksdb::WriteBatchWithIndex*>(jwbwi_handle);
+  assert(wbwi != nullptr);
+
+  auto* wb = wbwi->GetWriteBatch();
+
+  // TODO(AR) is the `wb` object owned by us?
+  return rocksdb::WriteBatchJni::construct(env, wb);
+}
+
 /*
  * Class:     org_rocksdb_WriteBatchWithIndex
  * Method:    iterator0
@@ -551,33 +685,15 @@ jlongArray Java_org_rocksdb_WBWIRocksIterator_entry1(
 
   jlong results[3];
 
-  //set the type of the write entry
-  switch (we.type) {
-    case rocksdb::kPutRecord:
-      results[0] = 0x1;
-      break;
+  // set the type of the write entry
+  results[0] = rocksdb::WriteTypeJni::toJavaWriteType(we.type);
 
-    case rocksdb::kMergeRecord:
-      results[0] = 0x2;
-      break;
-
-    case rocksdb::kDeleteRecord:
-      results[0] = 0x4;
-      break;
-
-    case rocksdb::kLogDataRecord:
-      results[0] = 0x8;
-      break;
-
-    default:
-      results[0] = 0x0;
-  }
-
-  // key_slice and value_slice will be freed by org.rocksdb.DirectSlice#close
+  // NOTE: key_slice and value_slice will be freed by org.rocksdb.DirectSlice#close
 
   auto* key_slice = new rocksdb::Slice(we.key.data(), we.key.size());
   results[1] = reinterpret_cast<jlong>(key_slice);
   if (we.type == rocksdb::kDeleteRecord
+      || we.type == rocksdb::kSingleDeleteRecord
       || we.type == rocksdb::kLogDataRecord) {
     // set native handle of value slice to null if no value available
     results[2] = 0;

java/rocksjni/writebatchhandlerjnicallback.cc

@@ -14,24 +14,62 @@ WriteBatchHandlerJniCallback::WriteBatchHandlerJniCallback(
     JNIEnv* env, jobject jWriteBatchHandler)
     : JniCallback(env, jWriteBatchHandler), m_env(env) {
 
+  m_jPutCfMethodId = WriteBatchHandlerJni::getPutCfMethodId(env);
+  if(m_jPutCfMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
   m_jPutMethodId = WriteBatchHandlerJni::getPutMethodId(env);
   if(m_jPutMethodId == nullptr) {
     // exception thrown
     return;
   }
 
+  m_jMergeCfMethodId = WriteBatchHandlerJni::getMergeCfMethodId(env);
+  if(m_jMergeCfMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
   m_jMergeMethodId = WriteBatchHandlerJni::getMergeMethodId(env);
   if(m_jMergeMethodId == nullptr) {
     // exception thrown
     return;
   }
 
+  m_jDeleteCfMethodId = WriteBatchHandlerJni::getDeleteCfMethodId(env);
+  if(m_jDeleteCfMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
   m_jDeleteMethodId = WriteBatchHandlerJni::getDeleteMethodId(env);
   if(m_jDeleteMethodId == nullptr) {
     // exception thrown
     return;
   }
 
+  m_jSingleDeleteCfMethodId =
+      WriteBatchHandlerJni::getSingleDeleteCfMethodId(env);
+  if(m_jSingleDeleteCfMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
+  m_jSingleDeleteMethodId = WriteBatchHandlerJni::getSingleDeleteMethodId(env);
+  if(m_jSingleDeleteMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
+  m_jDeleteRangeCfMethodId =
+      WriteBatchHandlerJni::getDeleteRangeCfMethodId(env);
+  if (m_jDeleteRangeCfMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
   m_jDeleteRangeMethodId = WriteBatchHandlerJni::getDeleteRangeMethodId(env);
   if (m_jDeleteRangeMethodId == nullptr) {
     // exception thrown
@@ -44,6 +82,45 @@ WriteBatchHandlerJniCallback::WriteBatchHandlerJniCallback(
     return;
   }
 
+  m_jPutBlobIndexCfMethodId =
+      WriteBatchHandlerJni::getPutBlobIndexCfMethodId(env);
+  if(m_jPutBlobIndexCfMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
+  m_jMarkBeginPrepareMethodId =
+      WriteBatchHandlerJni::getMarkBeginPrepareMethodId(env);
+  if(m_jMarkBeginPrepareMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
+  m_jMarkEndPrepareMethodId =
+      WriteBatchHandlerJni::getMarkEndPrepareMethodId(env);
+  if(m_jMarkEndPrepareMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
+  m_jMarkNoopMethodId = WriteBatchHandlerJni::getMarkNoopMethodId(env);
+  if(m_jMarkNoopMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+    
+  m_jMarkRollbackMethodId = WriteBatchHandlerJni::getMarkRollbackMethodId(env);
+  if(m_jMarkRollbackMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
+  m_jMarkCommitMethodId = WriteBatchHandlerJni::getMarkCommitMethodId(env);
+  if(m_jMarkCommitMethodId == nullptr) {
+    // exception thrown
+    return;
+  }
+
   m_jContinueMethodId = WriteBatchHandlerJni::getContinueMethodId(env);
   if(m_jContinueMethodId == nullptr) {
     // exception thrown
@@ -51,196 +128,272 @@ WriteBatchHandlerJniCallback::WriteBatchHandlerJniCallback(
   }
 }
 
+rocksdb::Status WriteBatchHandlerJniCallback::PutCF(uint32_t column_family_id,
+    const Slice& key, const Slice& value) {
+  auto put = [this, column_family_id] (
+      jbyteArray j_key, jbyteArray j_value) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jPutCfMethodId,
+      static_cast<jint>(column_family_id),
+      j_key,
+      j_value);
+  };
+  auto status = WriteBatchHandlerJniCallback::kv_op(key, value, put);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
+  }
+}
+
 void WriteBatchHandlerJniCallback::Put(const Slice& key, const Slice& value) {
-  const jbyteArray j_key = sliceToJArray(key);
-  if(j_key == nullptr) {
-    // exception thrown
-    if(m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    return;
-  }
-
-  const jbyteArray j_value = sliceToJArray(value);
-  if(j_value == nullptr) {
-    // exception thrown
-    if(m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    if(j_key != nullptr) {
-      m_env->DeleteLocalRef(j_key);
-    }
-    return;
-  }
-
-  m_env->CallVoidMethod(
+  auto put = [this] (
+        jbyteArray j_key, jbyteArray j_value) {
+    m_env->CallVoidMethod(
       m_jcallback_obj,
       m_jPutMethodId,
       j_key,
       j_value);
-  if(m_env->ExceptionCheck()) {
-    // exception thrown
-    m_env->ExceptionDescribe();
-    if(j_value != nullptr) {
-      m_env->DeleteLocalRef(j_value);
-    }
-    if(j_key != nullptr) {
-      m_env->DeleteLocalRef(j_key);
-    }
-    return;
-  }
+  };
+  WriteBatchHandlerJniCallback::kv_op(key, value, put);
+}
 
-  if(j_value != nullptr) {
-    m_env->DeleteLocalRef(j_value);
-  }
-  if(j_key != nullptr) {
-    m_env->DeleteLocalRef(j_key);
+rocksdb::Status WriteBatchHandlerJniCallback::MergeCF(uint32_t column_family_id,
+    const Slice& key, const Slice& value) {
+  auto merge = [this, column_family_id] (
+        jbyteArray j_key, jbyteArray j_value) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jMergeCfMethodId,
+      static_cast<jint>(column_family_id),
+      j_key,
+      j_value);
+  };
+  auto status = WriteBatchHandlerJniCallback::kv_op(key, value, merge);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
   }
 }
 
 void WriteBatchHandlerJniCallback::Merge(const Slice& key, const Slice& value) {
-  const jbyteArray j_key = sliceToJArray(key);
-  if(j_key == nullptr) {
-    // exception thrown
-    if(m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    return;
-  }
-
-  const jbyteArray j_value = sliceToJArray(value);
-  if(j_value == nullptr) {
-    // exception thrown
-    if(m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    if(j_key != nullptr) {
-      m_env->DeleteLocalRef(j_key);
-    }
-    return;
-  }
-
-  m_env->CallVoidMethod(
+  auto merge = [this] (
+        jbyteArray j_key, jbyteArray j_value) {
+    m_env->CallVoidMethod(
       m_jcallback_obj,
       m_jMergeMethodId,
       j_key,
       j_value);
-  if(m_env->ExceptionCheck()) {
-    // exception thrown
-    m_env->ExceptionDescribe();
-    if(j_value != nullptr) {
-      m_env->DeleteLocalRef(j_value);
-    }
-    if(j_key != nullptr) {
-      m_env->DeleteLocalRef(j_key);
-    }
-    return;
-  }
+  };
+  WriteBatchHandlerJniCallback::kv_op(key, value, merge);
+}
 
-  if(j_value != nullptr) {
-    m_env->DeleteLocalRef(j_value);
-  }
-  if(j_key != nullptr) {
-    m_env->DeleteLocalRef(j_key);
+rocksdb::Status WriteBatchHandlerJniCallback::DeleteCF(uint32_t column_family_id,
+    const Slice& key) {
+  auto remove = [this, column_family_id] (jbyteArray j_key) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jDeleteCfMethodId,
+      static_cast<jint>(column_family_id),
+      j_key);
+  };
+  auto status = WriteBatchHandlerJniCallback::k_op(key, remove);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
   }
 }
 
 void WriteBatchHandlerJniCallback::Delete(const Slice& key) {
-  const jbyteArray j_key = sliceToJArray(key);
-  if(j_key == nullptr) {
-    // exception thrown
-    if(m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    return;
-  }
-
-  m_env->CallVoidMethod(
+  auto remove = [this] (jbyteArray j_key) {
+    m_env->CallVoidMethod(
       m_jcallback_obj,
       m_jDeleteMethodId,
       j_key);
-  if(m_env->ExceptionCheck()) {
-    // exception thrown
-    m_env->ExceptionDescribe();
-    if(j_key != nullptr) {
-      m_env->DeleteLocalRef(j_key);
-    }
-    return;
-  }
+  };
+  WriteBatchHandlerJniCallback::k_op(key, remove);
+}
 
-  if(j_key != nullptr) {
-    m_env->DeleteLocalRef(j_key);
+rocksdb::Status WriteBatchHandlerJniCallback::SingleDeleteCF(uint32_t column_family_id,
+    const Slice& key) {
+  auto singleDelete = [this, column_family_id] (jbyteArray j_key) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jSingleDeleteCfMethodId,
+      static_cast<jint>(column_family_id),
+      j_key);
+  };
+  auto status = WriteBatchHandlerJniCallback::k_op(key, singleDelete);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
+  }
+}
+
+void WriteBatchHandlerJniCallback::SingleDelete(const Slice& key) {
+  auto singleDelete = [this] (jbyteArray j_key) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jSingleDeleteMethodId,
+      j_key);
+  };
+  WriteBatchHandlerJniCallback::k_op(key, singleDelete);
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::DeleteRangeCF(uint32_t column_family_id,
+    const Slice& beginKey, const Slice& endKey) {
+  auto deleteRange = [this, column_family_id] (
+        jbyteArray j_beginKey, jbyteArray j_endKey) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jDeleteRangeCfMethodId,
+      static_cast<jint>(column_family_id),
+      j_beginKey,
+      j_endKey);
+  };
+  auto status = WriteBatchHandlerJniCallback::kv_op(beginKey, endKey, deleteRange);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
   }
 }
 
 void WriteBatchHandlerJniCallback::DeleteRange(const Slice& beginKey,
-                                               const Slice& endKey) {
-  const jbyteArray j_beginKey = sliceToJArray(beginKey);
-  if (j_beginKey == nullptr) {
-    // exception thrown
-    if (m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    return;
-  }
-
-  const jbyteArray j_endKey = sliceToJArray(beginKey);
-  if (j_endKey == nullptr) {
-    // exception thrown
-    if (m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    return;
-  }
-
-  m_env->CallVoidMethod(m_jcallback_obj, m_jDeleteRangeMethodId,
-                        j_beginKey, j_endKey);
-  if (m_env->ExceptionCheck()) {
-    // exception thrown
-    m_env->ExceptionDescribe();
-    if (j_beginKey != nullptr) {
-      m_env->DeleteLocalRef(j_beginKey);
-    }
-    if (j_endKey != nullptr) {
-      m_env->DeleteLocalRef(j_endKey);
-    }
-    return;
-  }
-
-  if (j_beginKey != nullptr) {
-    m_env->DeleteLocalRef(j_beginKey);
-  }
-
-  if (j_endKey != nullptr) {
-    m_env->DeleteLocalRef(j_endKey);
-  }
+    const Slice& endKey) {
+  auto deleteRange = [this] (
+        jbyteArray j_beginKey, jbyteArray j_endKey) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jDeleteRangeMethodId,
+      j_beginKey,
+      j_endKey);
+  };
+  WriteBatchHandlerJniCallback::kv_op(beginKey, endKey, deleteRange);
 }
 
 void WriteBatchHandlerJniCallback::LogData(const Slice& blob) {
-  const jbyteArray j_blob = sliceToJArray(blob);
-  if(j_blob == nullptr) {
-    // exception thrown
-    if(m_env->ExceptionCheck()) {
-      m_env->ExceptionDescribe();
-    }
-    return;
-  }
-
-  m_env->CallVoidMethod(
+  auto logData = [this] (jbyteArray j_blob) {
+    m_env->CallVoidMethod(
       m_jcallback_obj,
       m_jLogDataMethodId,
       j_blob);
-  if(m_env->ExceptionCheck()) {
+  };
+  WriteBatchHandlerJniCallback::k_op(blob, logData);
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::PutBlobIndexCF(uint32_t column_family_id,
+    const Slice& key, const Slice& value) {
+  auto putBlobIndex = [this, column_family_id] (
+      jbyteArray j_key, jbyteArray j_value) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jPutBlobIndexCfMethodId,
+      static_cast<jint>(column_family_id),
+      j_key,
+      j_value);
+  };
+  auto status = WriteBatchHandlerJniCallback::kv_op(key, value, putBlobIndex);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
+  }
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::MarkBeginPrepare() {
+  m_env->CallVoidMethod(m_jcallback_obj, m_jMarkBeginPrepareMethodId);
+
+  // check for Exception, in-particular RocksDBException
+  if (m_env->ExceptionCheck()) {
     // exception thrown
-    m_env->ExceptionDescribe();
-    if(j_blob != nullptr) {
-      m_env->DeleteLocalRef(j_blob);
+    jthrowable exception = m_env->ExceptionOccurred();
+    std::unique_ptr<rocksdb::Status> status = rocksdb::RocksDBExceptionJni::toCppStatus(m_env, exception);
+    if (status == nullptr) {
+      // unkown status or exception occurred extracting status
+      m_env->ExceptionDescribe();
+      return rocksdb::Status::OK();  // TODO(AR) probably need a better error code here
+
+    } else {
+      m_env->ExceptionClear();  // clear the exception, as we have extracted the status
+      return rocksdb::Status(*status);
     }
-    return;
   }
 
-  if(j_blob != nullptr) {
-    m_env->DeleteLocalRef(j_blob);
+  return rocksdb::Status::OK();
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::MarkEndPrepare(const Slice& xid) {
+  auto markEndPrepare = [this] (
+      jbyteArray j_xid) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jMarkEndPrepareMethodId,
+      j_xid);
+  };
+  auto status = WriteBatchHandlerJniCallback::k_op(xid, markEndPrepare);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
+  }
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::MarkNoop(bool empty_batch) {
+  m_env->CallVoidMethod(m_jcallback_obj, m_jMarkNoopMethodId, static_cast<jboolean>(empty_batch));
+
+  // check for Exception, in-particular RocksDBException
+  if (m_env->ExceptionCheck()) {
+    // exception thrown
+    jthrowable exception = m_env->ExceptionOccurred();
+    std::unique_ptr<rocksdb::Status> status = rocksdb::RocksDBExceptionJni::toCppStatus(m_env, exception);
+    if (status == nullptr) {
+      // unkown status or exception occurred extracting status
+      m_env->ExceptionDescribe();
+      return rocksdb::Status::OK();  // TODO(AR) probably need a better error code here
+
+    } else {
+      m_env->ExceptionClear();  // clear the exception, as we have extracted the status
+      return rocksdb::Status(*status);
+    }
+  }
+
+  return rocksdb::Status::OK();
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::MarkRollback(const Slice& xid) {
+  auto markRollback = [this] (
+      jbyteArray j_xid) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jMarkRollbackMethodId,
+      j_xid);
+  };
+  auto status = WriteBatchHandlerJniCallback::k_op(xid, markRollback);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
+  }
+}
+
+rocksdb::Status WriteBatchHandlerJniCallback::MarkCommit(const Slice& xid) {
+  auto markCommit = [this] (
+      jbyteArray j_xid) {
+    m_env->CallVoidMethod(
+      m_jcallback_obj,
+      m_jMarkCommitMethodId,
+      j_xid);
+  };
+  auto status = WriteBatchHandlerJniCallback::k_op(xid, markCommit);
+  if(status == nullptr) {
+    return rocksdb::Status::OK();   // TODO(AR) what to do if there is an Exception but we don't know the rocksdb::Status?
+  } else {
+    return rocksdb::Status(*status);
   }
 }
 
@@ -256,39 +409,101 @@ bool WriteBatchHandlerJniCallback::Continue() {
   return static_cast<bool>(jContinue == JNI_TRUE);
 }
 
-/*
- * Creates a Java Byte Array from the data in a Slice
- *
- * When calling this function
- * you must remember to call env->DeleteLocalRef
- * on the result after you have finished with it
- *
- * @param s A Slice to convery to a Java byte array
- *
- * @return A reference to a Java byte array, or a nullptr if an
- *     exception occurs
- */
-jbyteArray WriteBatchHandlerJniCallback::sliceToJArray(const Slice& s) {
-
-  // TODO(AR) move to JniUtil
-
-  jbyteArray ja = m_env->NewByteArray(static_cast<jsize>(s.size()));
-  if(ja == nullptr) {
-    // exception thrown: OutOfMemoryError
-    return nullptr;
-  }
-
-  m_env->SetByteArrayRegion(
-      ja, 0, static_cast<jsize>(s.size()),
-      const_cast<jbyte*>(reinterpret_cast<const jbyte*>(s.data())));
-  if(m_env->ExceptionCheck()) {
-    if(ja != nullptr) {
-      m_env->DeleteLocalRef(ja);
+std::unique_ptr<rocksdb::Status> WriteBatchHandlerJniCallback::kv_op(const Slice& key, const Slice& value, std::function<void(jbyteArray, jbyteArray)> kvFn) {
+    const jbyteArray j_key = JniUtil::copyBytes(m_env, key);
+  if (j_key == nullptr) {
+    // exception thrown
+    if (m_env->ExceptionCheck()) {
+      m_env->ExceptionDescribe();
     }
-    // exception thrown: ArrayIndexOutOfBoundsException
     return nullptr;
   }
 
-  return ja;
+  const jbyteArray j_value = JniUtil::copyBytes(m_env, value);
+  if (j_value == nullptr) {
+    // exception thrown
+    if (m_env->ExceptionCheck()) {
+      m_env->ExceptionDescribe();
+    }
+    if (j_key != nullptr) {
+      m_env->DeleteLocalRef(j_key);
+    }
+    return nullptr;
+  }
+
+  kvFn(j_key, j_value);
+
+  // check for Exception, in-particular RocksDBException
+  if (m_env->ExceptionCheck()) {
+    if (j_value != nullptr) {
+      m_env->DeleteLocalRef(j_value);
+    }
+    if (j_key != nullptr) {
+      m_env->DeleteLocalRef(j_key);
+    }
+
+    // exception thrown
+    jthrowable exception = m_env->ExceptionOccurred();
+    std::unique_ptr<rocksdb::Status> status = rocksdb::RocksDBExceptionJni::toCppStatus(m_env, exception);
+    if (status == nullptr) {
+      // unkown status or exception occurred extracting status
+      m_env->ExceptionDescribe();
+      return nullptr;
+
+    } else {
+      m_env->ExceptionClear();  // clear the exception, as we have extracted the status
+      return status;
+    }
+  }
+
+  if (j_value != nullptr) {
+    m_env->DeleteLocalRef(j_value);
+  }
+  if (j_key != nullptr) {
+    m_env->DeleteLocalRef(j_key);
+  }
+
+  // all OK
+  return std::unique_ptr<rocksdb::Status>(new rocksdb::Status(rocksdb::Status::OK()));
+}
+
+std::unique_ptr<rocksdb::Status> WriteBatchHandlerJniCallback::k_op(const Slice& key, std::function<void(jbyteArray)> kFn) {
+    const jbyteArray j_key = JniUtil::copyBytes(m_env, key);
+  if (j_key == nullptr) {
+    // exception thrown
+    if (m_env->ExceptionCheck()) {
+      m_env->ExceptionDescribe();
+    }
+    return nullptr;
+  }
+
+  kFn(j_key);
+
+  // check for Exception, in-particular RocksDBException
+  if (m_env->ExceptionCheck()) {
+    if (j_key != nullptr) {
+      m_env->DeleteLocalRef(j_key);
+    }
+
+    // exception thrown
+    jthrowable exception = m_env->ExceptionOccurred();
+    std::unique_ptr<rocksdb::Status> status = rocksdb::RocksDBExceptionJni::toCppStatus(m_env, exception);
+    if (status == nullptr) {
+      // unkown status or exception occurred extracting status
+      m_env->ExceptionDescribe();
+      return nullptr;
+
+    } else {
+      m_env->ExceptionClear();  // clear the exception, as we have extracted the status
+      return status;
+    }
+  }
+
+  if (j_key != nullptr) {
+    m_env->DeleteLocalRef(j_key);
+  }
+
+  // all OK
+  return std::unique_ptr<rocksdb::Status>(new rocksdb::Status(rocksdb::Status::OK()));
 }
 }  // namespace rocksdb

java/rocksjni/writebatchhandlerjnicallback.h

@@ -9,7 +9,9 @@
 #ifndef JAVA_ROCKSJNI_WRITEBATCHHANDLERJNICALLBACK_H_
 #define JAVA_ROCKSJNI_WRITEBATCHHANDLERJNICALLBACK_H_
 
+#include <functional>
 #include <jni.h>
+#include <memory>
 #include "rocksjni/jnicallback.h"
 #include "rocksdb/write_batch.h"
 
@@ -25,22 +27,57 @@ class WriteBatchHandlerJniCallback : public JniCallback, public WriteBatch::Hand
  public:
     WriteBatchHandlerJniCallback(
       JNIEnv* env, jobject jWriteBackHandler);
+    Status PutCF(uint32_t column_family_id, const Slice& key,
+        const Slice& value);
     void Put(const Slice& key, const Slice& value);
+    Status MergeCF(uint32_t column_family_id, const Slice& key,
+        const Slice& value);
     void Merge(const Slice& key, const Slice& value);
+    Status DeleteCF(uint32_t column_family_id, const Slice& key);
     void Delete(const Slice& key);
+    Status SingleDeleteCF(uint32_t column_family_id, const Slice& key);
+    void SingleDelete(const Slice& key);
+    Status DeleteRangeCF(uint32_t column_family_id, const Slice& beginKey,
+        const Slice& endKey);
     void DeleteRange(const Slice& beginKey, const Slice& endKey);
     void LogData(const Slice& blob);
+    Status PutBlobIndexCF(uint32_t column_family_id, const Slice& key,
+        const Slice& value);
+    Status MarkBeginPrepare();
+    Status MarkEndPrepare(const Slice& xid);
+    Status MarkNoop(bool empty_batch);
+    Status MarkRollback(const Slice& xid);
+    Status MarkCommit(const Slice& xid);
     bool Continue();
 
  private:
     JNIEnv* m_env;
-    jbyteArray sliceToJArray(const Slice& s);
+    jmethodID m_jPutCfMethodId;
     jmethodID m_jPutMethodId;
+    jmethodID m_jMergeCfMethodId;
     jmethodID m_jMergeMethodId;
+    jmethodID m_jDeleteCfMethodId;
     jmethodID m_jDeleteMethodId;
+    jmethodID m_jSingleDeleteCfMethodId;
+    jmethodID m_jSingleDeleteMethodId;
+    jmethodID m_jDeleteRangeCfMethodId;
     jmethodID m_jDeleteRangeMethodId;
     jmethodID m_jLogDataMethodId;
+    jmethodID m_jPutBlobIndexCfMethodId;
+    jmethodID m_jMarkBeginPrepareMethodId;
+    jmethodID m_jMarkEndPrepareMethodId;
+    jmethodID m_jMarkNoopMethodId;
+    jmethodID m_jMarkRollbackMethodId;
+    jmethodID m_jMarkCommitMethodId;
     jmethodID m_jContinueMethodId;
+    /**
+     * @return A pointer to a rocksdb::Status or nullptr if an unexpected exception occurred
+     */
+    std::unique_ptr<rocksdb::Status> kv_op(const Slice& key, const Slice& value, std::function<void(jbyteArray, jbyteArray)> kvFn);
+    /**
+     * @return A pointer to a rocksdb::Status or nullptr if an unexpected exception occurred
+     */
+    std::unique_ptr<rocksdb::Status> k_op(const Slice& key, std::function<void(jbyteArray)> kFn);
 };
 }  // namespace rocksdb
 

java/samples/src/main/java/OptimisticTransactionSample.java

@@ -0,0 +1,184 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+import org.rocksdb.*;
+
+import static java.nio.charset.StandardCharsets.UTF_8;
+
+/**
+ * Demonstrates using Transactions on an OptimisticTransactionDB with
+ * varying isolation guarantees
+ */
+public class OptimisticTransactionSample {
+  private static final String dbPath = "/tmp/rocksdb_optimistic_transaction_example";
+
+  public static final void main(final String args[]) throws RocksDBException {
+
+    try(final Options options = new Options()
+        .setCreateIfMissing(true);
+        final OptimisticTransactionDB txnDb =
+            OptimisticTransactionDB.open(options, dbPath)) {
+
+      try (final WriteOptions writeOptions = new WriteOptions();
+           final ReadOptions readOptions = new ReadOptions()) {
+
+        ////////////////////////////////////////////////////////
+        //
+        // Simple OptimisticTransaction Example ("Read Committed")
+        //
+        ////////////////////////////////////////////////////////
+        readCommitted(txnDb, writeOptions, readOptions);
+
+
+        ////////////////////////////////////////////////////////
+        //
+        // "Repeatable Read" (Snapshot Isolation) Example
+        //   -- Using a single Snapshot
+        //
+        ////////////////////////////////////////////////////////
+        repeatableRead(txnDb, writeOptions, readOptions);
+
+
+        ////////////////////////////////////////////////////////
+        //
+        // "Read Committed" (Monotonic Atomic Views) Example
+        //   --Using multiple Snapshots
+        //
+        ////////////////////////////////////////////////////////
+        readCommitted_monotonicAtomicViews(txnDb, writeOptions, readOptions);
+      }
+    }
+  }
+
+  /**
+   * Demonstrates "Read Committed" isolation
+   */
+  private static void readCommitted(final OptimisticTransactionDB txnDb,
+      final WriteOptions writeOptions, final ReadOptions readOptions)
+      throws RocksDBException {
+    final byte key1[] = "abc".getBytes(UTF_8);
+    final byte value1[] = "def".getBytes(UTF_8);
+
+    final byte key2[] = "xyz".getBytes(UTF_8);
+    final byte value2[] = "zzz".getBytes(UTF_8);
+
+    // Start a transaction
+    try(final Transaction txn = txnDb.beginTransaction(writeOptions)) {
+      // Read a key in this transaction
+      byte[] value = txn.get(readOptions, key1);
+      assert(value == null);
+
+      // Write a key in this transaction
+      txn.put(key1, value1);
+
+      // Read a key OUTSIDE this transaction. Does not affect txn.
+      value = txnDb.get(readOptions, key1);
+      assert(value == null);
+
+      // Write a key OUTSIDE of this transaction.
+      // Does not affect txn since this is an unrelated key.
+      // If we wrote key 'abc' here, the transaction would fail to commit.
+      txnDb.put(writeOptions, key2, value2);
+
+      // Commit transaction
+      txn.commit();
+    }
+  }
+
+  /**
+   * Demonstrates "Repeatable Read" (Snapshot Isolation) isolation
+   */
+  private static void repeatableRead(final OptimisticTransactionDB txnDb,
+      final WriteOptions writeOptions, final ReadOptions readOptions)
+      throws RocksDBException {
+
+    final byte key1[] = "ghi".getBytes(UTF_8);
+    final byte value1[] = "jkl".getBytes(UTF_8);
+
+    // Set a snapshot at start of transaction by setting setSnapshot(true)
+    try(final OptimisticTransactionOptions txnOptions =
+            new OptimisticTransactionOptions().setSetSnapshot(true);
+        final Transaction txn =
+            txnDb.beginTransaction(writeOptions, txnOptions)) {
+
+      final Snapshot snapshot = txn.getSnapshot();
+
+      // Write a key OUTSIDE of transaction
+      txnDb.put(writeOptions, key1, value1);
+
+      // Read a key using the snapshot.
+      readOptions.setSnapshot(snapshot);
+      final byte[] value = txn.getForUpdate(readOptions, key1, true);
+      assert(value == value1);
+
+      try {
+        // Attempt to commit transaction
+        txn.commit();
+        throw new IllegalStateException();
+      } catch(final RocksDBException e) {
+        // Transaction could not commit since the write outside of the txn
+        // conflicted with the read!
+        assert(e.getStatus().getCode() == Status.Code.Busy);
+      }
+
+      txn.rollback();
+    } finally {
+      // Clear snapshot from read options since it is no longer valid
+      readOptions.setSnapshot(null);
+    }
+  }
+
+  /**
+   * Demonstrates "Read Committed" (Monotonic Atomic Views) isolation
+   *
+   * In this example, we set the snapshot multiple times.  This is probably
+   * only necessary if you have very strict isolation requirements to
+   * implement.
+   */
+  private static void readCommitted_monotonicAtomicViews(
+      final OptimisticTransactionDB txnDb, final WriteOptions writeOptions,
+      final ReadOptions readOptions) throws RocksDBException {
+
+    final byte keyX[] = "x".getBytes(UTF_8);
+    final byte valueX[] = "x".getBytes(UTF_8);
+
+    final byte keyY[] = "y".getBytes(UTF_8);
+    final byte valueY[] = "y".getBytes(UTF_8);
+
+    try (final OptimisticTransactionOptions txnOptions =
+             new OptimisticTransactionOptions().setSetSnapshot(true);
+         final Transaction txn =
+             txnDb.beginTransaction(writeOptions, txnOptions)) {
+
+      // Do some reads and writes to key "x"
+      Snapshot snapshot = txnDb.getSnapshot();
+      readOptions.setSnapshot(snapshot);
+      byte[] value = txn.get(readOptions, keyX);
+      txn.put(valueX, valueX);
+
+      // Do a write outside of the transaction to key "y"
+      txnDb.put(writeOptions, keyY, valueY);
+
+      // Set a new snapshot in the transaction
+      txn.setSnapshot();
+      snapshot = txnDb.getSnapshot();
+      readOptions.setSnapshot(snapshot);
+
+      // Do some reads and writes to key "y"
+      // Since the snapshot was advanced, the write done outside of the
+      // transaction does not conflict.
+      value = txn.getForUpdate(readOptions, keyY, true);
+      txn.put(keyY, valueY);
+
+      // Commit.  Since the snapshot was advanced, the write done outside of the
+      // transaction does not prevent this transaction from Committing.
+      txn.commit();
+
+    } finally {
+      // Clear snapshot from read options since it is no longer valid
+      readOptions.setSnapshot(null);
+    }
+  }
+}

java/samples/src/main/java/TransactionSample.java

@@ -0,0 +1,183 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+import org.rocksdb.*;
+
+import static java.nio.charset.StandardCharsets.UTF_8;
+
+/**
+ * Demonstrates using Transactions on a TransactionDB with
+ * varying isolation guarantees
+ */
+public class TransactionSample {
+  private static final String dbPath = "/tmp/rocksdb_transaction_example";
+
+  public static final void main(final String args[]) throws RocksDBException {
+
+    try(final Options options = new Options()
+        .setCreateIfMissing(true);
+        final TransactionDBOptions txnDbOptions = new TransactionDBOptions();
+        final TransactionDB txnDb =
+            TransactionDB.open(options, txnDbOptions, dbPath)) {
+
+      try (final WriteOptions writeOptions = new WriteOptions();
+           final ReadOptions readOptions = new ReadOptions()) {
+
+        ////////////////////////////////////////////////////////
+        //
+        // Simple Transaction Example ("Read Committed")
+        //
+        ////////////////////////////////////////////////////////
+        readCommitted(txnDb, writeOptions, readOptions);
+
+
+        ////////////////////////////////////////////////////////
+        //
+        // "Repeatable Read" (Snapshot Isolation) Example
+        //   -- Using a single Snapshot
+        //
+        ////////////////////////////////////////////////////////
+        repeatableRead(txnDb, writeOptions, readOptions);
+
+
+        ////////////////////////////////////////////////////////
+        //
+        // "Read Committed" (Monotonic Atomic Views) Example
+        //   --Using multiple Snapshots
+        //
+        ////////////////////////////////////////////////////////
+        readCommitted_monotonicAtomicViews(txnDb, writeOptions, readOptions);
+      }
+    }
+  }
+
+  /**
+   * Demonstrates "Read Committed" isolation
+   */
+  private static void readCommitted(final TransactionDB txnDb,
+      final WriteOptions writeOptions, final ReadOptions readOptions)
+      throws RocksDBException {
+    final byte key1[] = "abc".getBytes(UTF_8);
+    final byte value1[] = "def".getBytes(UTF_8);
+
+    final byte key2[] = "xyz".getBytes(UTF_8);
+    final byte value2[] = "zzz".getBytes(UTF_8);
+
+    // Start a transaction
+    try(final Transaction txn = txnDb.beginTransaction(writeOptions)) {
+      // Read a key in this transaction
+      byte[] value = txn.get(readOptions, key1);
+      assert(value == null);
+
+      // Write a key in this transaction
+      txn.put(key1, value1);
+
+      // Read a key OUTSIDE this transaction. Does not affect txn.
+      value = txnDb.get(readOptions, key1);
+      assert(value == null);
+
+      // Write a key OUTSIDE of this transaction.
+      // Does not affect txn since this is an unrelated key.
+      // If we wrote key 'abc' here, the transaction would fail to commit.
+      txnDb.put(writeOptions, key2, value2);
+
+      // Commit transaction
+      txn.commit();
+    }
+  }
+
+  /**
+   * Demonstrates "Repeatable Read" (Snapshot Isolation) isolation
+   */
+  private static void repeatableRead(final TransactionDB txnDb,
+      final WriteOptions writeOptions, final ReadOptions readOptions)
+      throws RocksDBException {
+
+    final byte key1[] = "ghi".getBytes(UTF_8);
+    final byte value1[] = "jkl".getBytes(UTF_8);
+
+    // Set a snapshot at start of transaction by setting setSnapshot(true)
+    try(final TransactionOptions txnOptions = new TransactionOptions()
+          .setSetSnapshot(true);
+        final Transaction txn =
+            txnDb.beginTransaction(writeOptions, txnOptions)) {
+
+      final Snapshot snapshot = txn.getSnapshot();
+
+      // Write a key OUTSIDE of transaction
+      txnDb.put(writeOptions, key1, value1);
+
+      // Attempt to read a key using the snapshot.  This will fail since
+      // the previous write outside this txn conflicts with this read.
+      readOptions.setSnapshot(snapshot);
+
+      try {
+        final byte[] value = txn.getForUpdate(readOptions, key1, true);
+        throw new IllegalStateException();
+      } catch(final RocksDBException e) {
+        assert(e.getStatus().getCode() == Status.Code.Busy);
+      }
+
+      txn.rollback();
+    } finally {
+      // Clear snapshot from read options since it is no longer valid
+      readOptions.setSnapshot(null);
+    }
+  }
+
+  /**
+   * Demonstrates "Read Committed" (Monotonic Atomic Views) isolation
+   *
+   * In this example, we set the snapshot multiple times.  This is probably
+   * only necessary if you have very strict isolation requirements to
+   * implement.
+   */
+  private static void readCommitted_monotonicAtomicViews(
+      final TransactionDB txnDb, final WriteOptions writeOptions,
+      final ReadOptions readOptions) throws RocksDBException {
+
+    final byte keyX[] = "x".getBytes(UTF_8);
+    final byte valueX[] = "x".getBytes(UTF_8);
+
+    final byte keyY[] = "y".getBytes(UTF_8);
+    final byte valueY[] = "y".getBytes(UTF_8);
+
+    try (final TransactionOptions txnOptions = new TransactionOptions()
+        .setSetSnapshot(true);
+         final Transaction txn =
+             txnDb.beginTransaction(writeOptions, txnOptions)) {
+
+      // Do some reads and writes to key "x"
+      Snapshot snapshot = txnDb.getSnapshot();
+      readOptions.setSnapshot(snapshot);
+      byte[] value = txn.get(readOptions, keyX);
+      txn.put(valueX, valueX);
+
+      // Do a write outside of the transaction to key "y"
+      txnDb.put(writeOptions, keyY, valueY);
+
+      // Set a new snapshot in the transaction
+      txn.setSnapshot();
+      txn.setSavePoint();
+      snapshot = txnDb.getSnapshot();
+      readOptions.setSnapshot(snapshot);
+
+      // Do some reads and writes to key "y"
+      // Since the snapshot was advanced, the write done outside of the
+      // transaction does not conflict.
+      value = txn.getForUpdate(readOptions, keyY, true);
+      txn.put(keyY, valueY);
+
+      // Decide we want to revert the last write from this transaction.
+      txn.rollbackToSavePoint();
+
+      // Commit.
+      txn.commit();
+    } finally {
+      // Clear snapshot from read options since it is no longer valid
+      readOptions.setSnapshot(null);
+    }
+  }
+}

java/src/main/java/org/rocksdb/AbstractComparator.java

@@ -17,10 +17,23 @@ package org.rocksdb;
 public abstract class AbstractComparator<T extends AbstractSlice<?>>
     extends RocksCallbackObject {
 
+  protected AbstractComparator() {
+    super();
+  }
+
   protected AbstractComparator(final ComparatorOptions copt) {
     super(copt.nativeHandle_);
   }
 
+  /**
+   * Get the type of this comparator.
+   *
+   * Used for determining the correct C++ cast in native code.
+   *
+   * @return The type of the comparator.
+   */
+  abstract ComparatorType getComparatorType();
+
   /**
    * The name of the comparator.  Used to check for comparator
    * mismatches (i.e., a DB created with one comparator is

java/src/main/java/org/rocksdb/AbstractTransactionNotifier.java

@@ -0,0 +1,54 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+package org.rocksdb;
+
+/**
+ * Provides notification to the caller of SetSnapshotOnNextOperation when
+ * the actual snapshot gets created
+ */
+public abstract class AbstractTransactionNotifier
+    extends RocksCallbackObject {
+
+  protected AbstractTransactionNotifier() {
+    super();
+  }
+
+  /**
+   * Implement this method to receive notification when a snapshot is
+   * requested via {@link Transaction#setSnapshotOnNextOperation()}.
+   *
+   * @param newSnapshot the snapshot that has been created.
+   */
+  public abstract void snapshotCreated(final Snapshot newSnapshot);
+
+  /**
+   * This is intentionally private as it is the callback hook
+   * from JNI
+   */
+  private void snapshotCreated(final long snapshotHandle) {
+    snapshotCreated(new Snapshot(snapshotHandle));
+  }
+
+  @Override
+  protected long initializeNative(final long... nativeParameterHandles) {
+    return createNewTransactionNotifier();
+  }
+
+  private native long createNewTransactionNotifier();
+
+  /**
+   * Deletes underlying C++ TransactionNotifier pointer.
+   *
+   * Note that this function should be called only after all
+   * Transactions referencing the comparator are closed.
+   * Otherwise an undefined behavior will occur.
+   */
+  @Override
+  protected void disposeInternal() {
+    disposeInternal(nativeHandle_);
+  }
+  protected final native void disposeInternal(final long handle);
+}

java/src/main/java/org/rocksdb/AbstractWriteBatch.java

@@ -18,52 +18,80 @@ public abstract class AbstractWriteBatch extends RocksObject
   }
 
   @Override
-  public void put(byte[] key, byte[] value) {
+  public void put(byte[] key, byte[] value) throws RocksDBException {
     put(nativeHandle_, key, key.length, value, value.length);
   }
 
   @Override
   public void put(ColumnFamilyHandle columnFamilyHandle, byte[] key,
-      byte[] value) {
+      byte[] value) throws RocksDBException {
     put(nativeHandle_, key, key.length, value, value.length,
         columnFamilyHandle.nativeHandle_);
   }
 
   @Override
-  public void merge(byte[] key, byte[] value) {
+  public void merge(byte[] key, byte[] value) throws RocksDBException {
     merge(nativeHandle_, key, key.length, value, value.length);
   }
 
   @Override
   public void merge(ColumnFamilyHandle columnFamilyHandle, byte[] key,
-      byte[] value) {
+      byte[] value) throws RocksDBException {
     merge(nativeHandle_, key, key.length, value, value.length,
         columnFamilyHandle.nativeHandle_);
   }
 
   @Override
-  public void remove(byte[] key) {
-    remove(nativeHandle_, key, key.length);
+  @Deprecated
+  public void remove(byte[] key) throws RocksDBException {
+    delete(nativeHandle_, key, key.length);
   }
 
   @Override
-  public void remove(ColumnFamilyHandle columnFamilyHandle, byte[] key) {
-    remove(nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_);
+  @Deprecated
+  public void remove(ColumnFamilyHandle columnFamilyHandle, byte[] key)
+      throws RocksDBException {
+    delete(nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_);
   }
 
   @Override
-  public void deleteRange(byte[] beginKey, byte[] endKey) {
+  public void delete(byte[] key) throws RocksDBException {
+    delete(nativeHandle_, key, key.length);
+  }
+
+  @Override
+  public void delete(ColumnFamilyHandle columnFamilyHandle, byte[] key)
+      throws RocksDBException {
+    delete(nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_);
+  }
+
+
+  @Override
+  public void singleDelete(byte[] key) throws RocksDBException {
+    singleDelete(nativeHandle_, key, key.length);
+  }
+
+  @Override
+  public void singleDelete(ColumnFamilyHandle columnFamilyHandle, byte[] key)
+      throws RocksDBException {
+    singleDelete(nativeHandle_, key, key.length, columnFamilyHandle.nativeHandle_);
+  }
+
+  @Override
+  public void deleteRange(byte[] beginKey, byte[] endKey)
+      throws RocksDBException {
     deleteRange(nativeHandle_, beginKey, beginKey.length, endKey, endKey.length);
   }
 
   @Override
-  public void deleteRange(ColumnFamilyHandle columnFamilyHandle, byte[] beginKey, byte[] endKey) {
+  public void deleteRange(ColumnFamilyHandle columnFamilyHandle,
+      byte[] beginKey, byte[] endKey) throws RocksDBException {
     deleteRange(nativeHandle_, beginKey, beginKey.length, endKey, endKey.length,
         columnFamilyHandle.nativeHandle_);
   }
 
   @Override
-  public void putLogData(byte[] blob) {
+  public void putLogData(byte[] blob) throws RocksDBException {
     putLogData(nativeHandle_, blob, blob.length);
   }
 
@@ -82,38 +110,67 @@ public abstract class AbstractWriteBatch extends RocksObject
     rollbackToSavePoint0(nativeHandle_);
   }
 
+  @Override
+  public void popSavePoint() throws RocksDBException {
+    popSavePoint(nativeHandle_);
+  }
+
+  @Override
+  public void setMaxBytes(final long maxBytes) {
+    setMaxBytes(nativeHandle_, maxBytes);
+  }
+
+  @Override
+  public WriteBatch getWriteBatch() {
+    return getWriteBatch(nativeHandle_);
+  }
+
   abstract int count0(final long handle);
 
   abstract void put(final long handle, final byte[] key, final int keyLen,
-      final byte[] value, final int valueLen);
+      final byte[] value, final int valueLen) throws RocksDBException;
 
   abstract void put(final long handle, final byte[] key, final int keyLen,
-      final byte[] value, final int valueLen, final long cfHandle);
+      final byte[] value, final int valueLen, final long cfHandle)
+      throws RocksDBException;
 
   abstract void merge(final long handle, final byte[] key, final int keyLen,
-      final byte[] value, final int valueLen);
+      final byte[] value, final int valueLen) throws RocksDBException;
 
   abstract void merge(final long handle, final byte[] key, final int keyLen,
-      final byte[] value, final int valueLen, final long cfHandle);
+      final byte[] value, final int valueLen, final long cfHandle)
+      throws RocksDBException;
 
-  abstract void remove(final long handle, final byte[] key,
-      final int keyLen);
+  abstract void delete(final long handle, final byte[] key,
+      final int keyLen) throws RocksDBException;
 
-  abstract void remove(final long handle, final byte[] key,
-      final int keyLen, final long cfHandle);
+  abstract void delete(final long handle, final byte[] key,
+      final int keyLen, final long cfHandle) throws RocksDBException;
+
+  abstract void singleDelete(final long handle, final byte[] key,
+                       final int keyLen) throws RocksDBException;
+
+  abstract void singleDelete(final long handle, final byte[] key,
+                       final int keyLen, final long cfHandle) throws RocksDBException;
 
   abstract void deleteRange(final long handle, final byte[] beginKey, final int beginKeyLen,
-      final byte[] endKey, final int endKeyLen);
+      final byte[] endKey, final int endKeyLen) throws RocksDBException;
 
   abstract void deleteRange(final long handle, final byte[] beginKey, final int beginKeyLen,
-      final byte[] endKey, final int endKeyLen, final long cfHandle);
+      final byte[] endKey, final int endKeyLen, final long cfHandle) throws RocksDBException;
 
   abstract void putLogData(final long handle, final byte[] blob,
-      final int blobLen);
+      final int blobLen) throws RocksDBException;
 
   abstract void clear0(final long handle);
 
   abstract void setSavePoint0(final long handle);
 
   abstract void rollbackToSavePoint0(final long handle);
+
+  abstract void popSavePoint(final long handle) throws RocksDBException;
+
+  abstract void setMaxBytes(final long handle, long maxBytes);
+
+  abstract WriteBatch getWriteBatch(final long handle);
 }

java/src/main/java/org/rocksdb/ColumnFamilyDescriptor.java

@@ -5,6 +5,8 @@
 
 package org.rocksdb;
 
+import java.util.Arrays;
+
 /**
  * <p>Describes a column family with a
  * name and respective Options.</p>
@@ -32,7 +34,7 @@ public class ColumnFamilyDescriptor {
    * @since 3.10.0
    */
   public ColumnFamilyDescriptor(final byte[] columnFamilyName,
-      final ColumnFamilyOptions columnFamilyOptions) {
+                                final ColumnFamilyOptions columnFamilyOptions) {
     columnFamilyName_ = columnFamilyName;
     columnFamilyOptions_ = columnFamilyOptions;
   }
@@ -43,19 +45,65 @@ public class ColumnFamilyDescriptor {
    * @return column family name.
    * @since 3.10.0
    */
-  public byte[] columnFamilyName() {
+  public byte[] getName() {
     return columnFamilyName_;
   }
 
+  /**
+   * Retrieve name of column family.
+   *
+   * @return column family name.
+   * @since 3.10.0
+   *
+   * @deprecated Use {@link #getName()} instead.
+   */
+  @Deprecated
+  public byte[] columnFamilyName() {
+    return getName();
+  }
+
   /**
    * Retrieve assigned options instance.
    *
    * @return Options instance assigned to this instance.
    */
-  public ColumnFamilyOptions columnFamilyOptions() {
+  public ColumnFamilyOptions getOptions() {
     return columnFamilyOptions_;
   }
 
+  /**
+   * Retrieve assigned options instance.
+   *
+   * @return Options instance assigned to this instance.
+   *
+   * @deprecated Use {@link #getOptions()} instead.
+   */
+  @Deprecated
+  public ColumnFamilyOptions columnFamilyOptions() {
+    return getOptions();
+  }
+
+  @Override
+  public boolean equals(final Object o) {
+    if (this == o) {
+      return true;
+    }
+    if (o == null || getClass() != o.getClass()) {
+      return false;
+    }
+
+    final ColumnFamilyDescriptor that = (ColumnFamilyDescriptor) o;
+    return Arrays.equals(columnFamilyName_, that.columnFamilyName_)
+            && columnFamilyOptions_.nativeHandle_ == that.columnFamilyOptions_.nativeHandle_;
+  }
+
+  @Override
+  public int hashCode() {
+    int result = (int) (columnFamilyOptions_.nativeHandle_ ^ (columnFamilyOptions_.nativeHandle_ >>> 32));
+    result = 31 * result + Arrays.hashCode(columnFamilyName_);
+    return result;
+  }
+
   private final byte[] columnFamilyName_;
   private final ColumnFamilyOptions columnFamilyOptions_;
 }

java/src/main/java/org/rocksdb/ColumnFamilyHandle.java

@@ -5,6 +5,9 @@
 
 package org.rocksdb;
 
+import java.util.Arrays;
+import java.util.Objects;
+
 /**
  * ColumnFamilyHandle class to hold handles to underlying rocksdb
  * ColumnFamily Pointers.
@@ -21,6 +24,63 @@ public class ColumnFamilyHandle extends RocksObject {
     this.rocksDB_ = rocksDB;
   }
 
+  /**
+   * Gets the name of the Column Family.
+   *
+   * @return The name of the Column Family.
+   */
+  public byte[] getName() {
+    return getName(nativeHandle_);
+  }
+
+  /**
+   * Gets the ID of the Column Family.
+   *
+   * @return the ID of the Column Family.
+   */
+  public int getID() {
+    return getID(nativeHandle_);
+  }
+
+  /**
+   * Gets the up-to-date descriptor of the column family
+   * associated with this handle. Since it fills "*desc" with the up-to-date
+   * information, this call might internally lock and release DB mutex to
+   * access the up-to-date CF options. In addition, all the pointer-typed
+   * options cannot be referenced any longer than the original options exist.
+   *
+   * Note that this function is not supported in RocksDBLite.
+   *
+   * @return the up-to-date descriptor.
+   *
+   * @throws RocksDBException if an error occurs whilst retrieving the
+   *     descriptor.
+   */
+  public ColumnFamilyDescriptor getDescriptor() throws RocksDBException {
+    assert(isOwningHandle());
+    return getDescriptor(nativeHandle_);
+  }
+
+  @Override
+  public boolean equals(final Object o) {
+    if (this == o) {
+      return true;
+    }
+    if (o == null || getClass() != o.getClass()) {
+      return false;
+    }
+
+    final ColumnFamilyHandle that = (ColumnFamilyHandle) o;
+    return rocksDB_.nativeHandle_ == that.rocksDB_.nativeHandle_ &&
+        getID() == that.getID() &&
+        Arrays.equals(getName(), that.getName());
+  }
+
+  @Override
+  public int hashCode() {
+    return Objects.hash(getName(), getID(), rocksDB_.nativeHandle_);
+  }
+
   /**
    * <p>Deletes underlying C++ iterator pointer.</p>
    *
@@ -36,6 +96,9 @@ public class ColumnFamilyHandle extends RocksObject {
     }
   }
 
+  private native byte[] getName(final long handle);
+  private native int getID(final long handle);
+  private native ColumnFamilyDescriptor getDescriptor(final long handle) throws RocksDBException;
   @Override protected final native void disposeInternal(final long handle);
 
   private final RocksDB rocksDB_;

java/src/main/java/org/rocksdb/ColumnFamilyOptions.java

@@ -33,6 +33,18 @@ public class ColumnFamilyOptions extends RocksObject
     super(newColumnFamilyOptions());
   }
 
+  /**
+   * <p>Constructor to be used by
+   * {@link #getColumnFamilyOptionsFromProps(java.util.Properties)},
+   * {@link ColumnFamilyDescriptor#columnFamilyOptions()}
+   * and also called via JNI.</p>
+   *
+   * @param handle native handle to ColumnFamilyOptions instance.
+   */
+  ColumnFamilyOptions(final long handle) {
+    super(handle);
+  }
+
   /**
    * <p>Method to get a options instance by using pre-configured
    * property values. If one or many values are undefined in
@@ -131,7 +143,7 @@ public class ColumnFamilyOptions extends RocksObject
       final AbstractComparator<? extends AbstractSlice<?>> comparator) {
     assert (isOwningHandle());
     setComparatorHandle(nativeHandle_, comparator.nativeHandle_,
-            comparator instanceof DirectComparator);
+            comparator.getComparatorType().getValue());
     comparator_ = comparator;
     return this;
   }
@@ -768,17 +780,6 @@ public class ColumnFamilyOptions extends RocksObject
     return forceConsistencyChecks(nativeHandle_);
   }
 
-  /**
-   * <p>Constructor to be used by
-   * {@link #getColumnFamilyOptionsFromProps(java.util.Properties)}</p>
-   * and also called via JNI.
-   *
-   * @param handle native handle to ColumnFamilyOptions instance.
-   */
-  public ColumnFamilyOptions(final long handle) {
-    super(handle);
-  }
-
   private static native long getColumnFamilyOptionsFromProps(
       String optString);
 
@@ -794,7 +795,7 @@ public class ColumnFamilyOptions extends RocksObject
       long memtableMemoryBudget);
   private native void setComparatorHandle(long handle, int builtinComparator);
   private native void setComparatorHandle(long optHandle,
-      long comparatorHandle, boolean isDirect);
+      long comparatorHandle, byte comparatorType);
   private native void setMergeOperatorName(long handle, String name);
   private native void setMergeOperator(long handle, long mergeOperatorHandle);
   private native void setCompactionFilterHandle(long handle,

java/src/main/java/org/rocksdb/Comparator.java

@@ -25,5 +25,10 @@ public abstract class Comparator extends AbstractComparator<Slice> {
     return createNewComparator0(nativeParameterHandles[0]);
   }
 
+  @Override
+  final ComparatorType getComparatorType() {
+    return ComparatorType.JAVA_COMPARATOR;
+  }
+
   private native long createNewComparator0(final long comparatorOptionsHandle);
 }

java/src/main/java/org/rocksdb/ComparatorType.java

@@ -0,0 +1,49 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+package org.rocksdb;
+
+enum ComparatorType {
+  JAVA_COMPARATOR((byte)0x0),
+  JAVA_DIRECT_COMPARATOR((byte)0x1),
+  JAVA_NATIVE_COMPARATOR_WRAPPER((byte)0x2);
+
+  private final byte value;
+
+  ComparatorType(final byte value) {
+    this.value = value;
+  }
+
+  /**
+   * <p>Returns the byte value of the enumerations value.</p>
+   *
+   * @return byte representation
+   */
+  byte getValue() {
+    return value;
+  }
+
+  /**
+   * <p>Get the ComparatorType enumeration value by
+   * passing the byte identifier to this method.</p>
+   *
+   * @param byteIdentifier of ComparatorType.
+   *
+   * @return ComparatorType instance.
+   *
+   * @throws IllegalArgumentException if the comparator type for the byteIdentifier
+   *     cannot be found
+   */
+  static ComparatorType getComparatorType(final byte byteIdentifier) {
+    for (final ComparatorType comparatorType : ComparatorType.values()) {
+      if (comparatorType.getValue() == byteIdentifier) {
+        return comparatorType;
+      }
+    }
+
+    throw new IllegalArgumentException(
+        "Illegal value provided for ComparatorType.");
+  }
+}

java/src/main/java/org/rocksdb/DBOptionsInterface.java

@@ -269,7 +269,10 @@ public interface DBOptionsInterface<T extends DBOptionsInterface> {
    * Statistics objects should not be shared between DB instances as
    * it does not use any locks to prevent concurrent updates.</p>
    *
+   * @param statistics The statistics to set
+   *
    * @return the instance of the current object.
+   *
    * @see RocksDB#open(org.rocksdb.Options, String)
    */
   T setStatistics(final Statistics statistics);
@@ -277,7 +280,9 @@ public interface DBOptionsInterface<T extends DBOptionsInterface> {
   /**
    * <p>Returns statistics object.</p>
    *
-   * @return the instance of the statistics object or null if there is no statistics object.
+   * @return the instance of the statistics object or null if there is no
+   * statistics object.
+   *
    * @see #setStatistics(Statistics)
    */
   Statistics statistics();

java/src/main/java/org/rocksdb/DirectComparator.java

@@ -25,6 +25,11 @@ public abstract class DirectComparator extends AbstractComparator<DirectSlice> {
     return createNewDirectComparator0(nativeParameterHandles[0]);
   }
 
+  @Override
+  final ComparatorType getComparatorType() {
+    return ComparatorType.JAVA_DIRECT_COMPARATOR;
+  }
+
   private native long createNewDirectComparator0(
       final long comparatorOptionsHandle);
 }

java/src/main/java/org/rocksdb/NativeComparatorWrapper.java

@@ -0,0 +1,57 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+package org.rocksdb;
+
+/**
+ * A simple abstraction to allow a Java class to wrap a custom comparator
+ * implemented in C++.
+ *
+ * The native comparator must directly extend rocksdb::Comparator.
+ */
+public abstract class NativeComparatorWrapper
+    extends AbstractComparator<Slice> {
+
+  @Override
+  final ComparatorType getComparatorType() {
+    return ComparatorType.JAVA_NATIVE_COMPARATOR_WRAPPER;
+  }
+
+  @Override
+  public final String name() {
+    throw new IllegalStateException("This should not be called. " +
+        "Implementation is in Native code");
+  }
+
+  @Override
+  public final int compare(final Slice s1, final Slice s2) {
+    throw new IllegalStateException("This should not be called. " +
+        "Implementation is in Native code");
+  }
+
+  @Override
+  public final String findShortestSeparator(final String start, final Slice limit) {
+    throw new IllegalStateException("This should not be called. " +
+        "Implementation is in Native code");
+  }
+
+  @Override
+  public final String findShortSuccessor(final String key) {
+    throw new IllegalStateException("This should not be called. " +
+        "Implementation is in Native code");
+  }
+
+  /**
+   * We override {@link RocksCallbackObject#disposeInternal()}
+   * as disposing of a native rocksd::Comparator extension requires
+   * a slightly different approach as it is not really a RocksCallbackObject
+   */
+  @Override
+  protected void disposeInternal() {
+    disposeInternal(nativeHandle_);
+  }
+
+  private native void disposeInternal(final long handle);
+}

java/src/main/java/org/rocksdb/OptimisticTransactionDB.java

@@ -0,0 +1,175 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+package org.rocksdb;
+
+import java.util.List;
+
+/**
+ * Database with Transaction support.
+ */
+public class OptimisticTransactionDB extends RocksDB
+    implements TransactionalDB<OptimisticTransactionOptions> {
+
+  /**
+   * Private constructor.
+   *
+   * @param nativeHandle The native handle of the C++ OptimisticTransactionDB
+   *     object
+   */
+  private OptimisticTransactionDB(final long nativeHandle) {
+    super(nativeHandle);
+  }
+
+  /**
+   * Open an OptimisticTransactionDB similar to
+   * {@link RocksDB#open(Options, String)}.
+   *
+   * @param options {@link org.rocksdb.Options} instance.
+   * @param path the path to the rocksdb.
+   *
+   * @return a {@link OptimisticTransactionDB} instance on success, null if the
+   * specified {@link OptimisticTransactionDB} can not be opened.
+   *
+   * @throws RocksDBException if an error occurs whilst opening the database.
+   */
+  public static OptimisticTransactionDB open(final Options options,
+      final String path) throws RocksDBException {
+    final OptimisticTransactionDB otdb = new OptimisticTransactionDB(open(
+        options.nativeHandle_, path));
+
+    // when non-default Options is used, keeping an Options reference
+    // in RocksDB can prevent Java to GC during the life-time of
+    // the currently-created RocksDB.
+    otdb.storeOptionsInstance(options);
+
+    return otdb;
+  }
+
+  /**
+   * Open an OptimisticTransactionDB similar to
+   * {@link RocksDB#open(DBOptions, String, List, List)}.
+   *
+   * @param dbOptions {@link org.rocksdb.DBOptions} instance.
+   * @param path the path to the rocksdb.
+   * @param columnFamilyDescriptors list of column family descriptors
+   * @param columnFamilyHandles will be filled with ColumnFamilyHandle instances
+   *
+   * @return a {@link OptimisticTransactionDB} instance on success, null if the
+   *     specified {@link OptimisticTransactionDB} can not be opened.
+   *
+   * @throws RocksDBException if an error occurs whilst opening the database.
+   */
+  public static OptimisticTransactionDB open(final DBOptions dbOptions,
+      final String path,
+      final List<ColumnFamilyDescriptor> columnFamilyDescriptors,
+      final List<ColumnFamilyHandle> columnFamilyHandles)
+      throws RocksDBException {
+
+    final byte[][] cfNames = new byte[columnFamilyDescriptors.size()][];
+    final long[] cfOptionHandles = new long[columnFamilyDescriptors.size()];
+    for (int i = 0; i < columnFamilyDescriptors.size(); i++) {
+      final ColumnFamilyDescriptor cfDescriptor = columnFamilyDescriptors
+          .get(i);
+      cfNames[i] = cfDescriptor.columnFamilyName();
+      cfOptionHandles[i] = cfDescriptor.columnFamilyOptions().nativeHandle_;
+    }
+
+    final long[] handles = open(dbOptions.nativeHandle_, path, cfNames,
+        cfOptionHandles);
+    final OptimisticTransactionDB otdb =
+        new OptimisticTransactionDB(handles[0]);
+
+    // when non-default Options is used, keeping an Options reference
+    // in RocksDB can prevent Java to GC during the life-time of
+    // the currently-created RocksDB.
+    otdb.storeOptionsInstance(dbOptions);
+
+    for (int i = 1; i < handles.length; i++) {
+      columnFamilyHandles.add(new ColumnFamilyHandle(otdb, handles[i]));
+    }
+
+    return otdb;
+  }
+
+  @Override
+  public Transaction beginTransaction(final WriteOptions writeOptions) {
+    return new Transaction(this, beginTransaction(nativeHandle_,
+        writeOptions.nativeHandle_));
+  }
+
+  @Override
+  public Transaction beginTransaction(final WriteOptions writeOptions,
+      final OptimisticTransactionOptions optimisticTransactionOptions) {
+    return new Transaction(this, beginTransaction(nativeHandle_,
+        writeOptions.nativeHandle_,
+        optimisticTransactionOptions.nativeHandle_));
+  }
+
+  // TODO(AR) consider having beingTransaction(... oldTransaction) set a
+  // reference count inside Transaction, so that we can always call
+  // Transaction#close but the object is only disposed when there are as many
+  // closes as beginTransaction. Makes the try-with-resources paradigm easier for
+  // java developers
+
+  @Override
+  public Transaction beginTransaction(final WriteOptions writeOptions,
+      final Transaction oldTransaction) {
+    final long jtxn_handle = beginTransaction_withOld(nativeHandle_,
+        writeOptions.nativeHandle_, oldTransaction.nativeHandle_);
+
+    // RocksJava relies on the assumption that
+    // we do not allocate a new Transaction object
+    // when providing an old_txn
+    assert(jtxn_handle == oldTransaction.nativeHandle_);
+
+    return oldTransaction;
+  }
+
+  @Override
+  public Transaction beginTransaction(final WriteOptions writeOptions,
+      final OptimisticTransactionOptions optimisticTransactionOptions,
+      final Transaction oldTransaction) {
+    final long jtxn_handle = beginTransaction_withOld(nativeHandle_,
+        writeOptions.nativeHandle_, optimisticTransactionOptions.nativeHandle_,
+        oldTransaction.nativeHandle_);
+
+    // RocksJava relies on the assumption that
+    // we do not allocate a new Transaction object
+    // when providing an old_txn
+    assert(jtxn_handle == oldTransaction.nativeHandle_);
+
+    return oldTransaction;
+  }
+
+  /**
+   * Get the underlying database that was opened.
+   *
+   * @return The underlying database that was opened.
+   */
+  public RocksDB getBaseDB() {
+    final RocksDB db = new RocksDB(getBaseDB(nativeHandle_));
+    db.disOwnNativeHandle();
+    return db;
+  }
+
+  protected static native long open(final long optionsHandle,
+      final String path) throws RocksDBException;
+  protected static native long[] open(final long handle, final String path,
+      final byte[][] columnFamilyNames, final long[] columnFamilyOptions);
+  private native long beginTransaction(final long handle,
+      final long writeOptionsHandle);
+  private native long beginTransaction(final long handle,
+      final long writeOptionsHandle,
+      final long optimisticTransactionOptionsHandle);
+  private native long beginTransaction_withOld(final long handle,
+      final long writeOptionsHandle, final long oldTransactionHandle);
+  private native long beginTransaction_withOld(final long handle,
+      final long writeOptionsHandle,
+      final long optimisticTransactionOptionsHandle,
+      final long oldTransactionHandle);
+  private native long getBaseDB(final long handle);
+  @Override protected final native void disposeInternal(final long handle);
+}

java/src/main/java/org/rocksdb/OptimisticTransactionOptions.java

@@ -0,0 +1,53 @@
+// Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+package org.rocksdb;
+
+public class OptimisticTransactionOptions extends RocksObject
+    implements TransactionalOptions {
+
+  public OptimisticTransactionOptions() {
+    super(newOptimisticTransactionOptions());
+  }
+
+  @Override
+  public boolean isSetSnapshot() {
+    assert(isOwningHandle());
+    return isSetSnapshot(nativeHandle_);
+  }
+
+  @Override
+  public OptimisticTransactionOptions setSetSnapshot(
+      final boolean setSnapshot) {
+    assert(isOwningHandle());
+    setSetSnapshot(nativeHandle_, setSnapshot);
+    return this;
+  }
+
+  /**
+   * Should be set if the DB has a non-default comparator.
+   * See comment in
+   * {@link WriteBatchWithIndex#WriteBatchWithIndex(AbstractComparator, int, boolean)}
+   * constructor.
+   *
+   * @param comparator The comparator to use for the transaction.
+   *
+   * @return this OptimisticTransactionOptions instance
+   */
+  public OptimisticTransactionOptions setComparator(
+      final AbstractComparator<? extends AbstractSlice<?>> comparator) {
+    assert(isOwningHandle());
+    setComparator(nativeHandle_, comparator.nativeHandle_);
+    return this;
+  }
+
+  private native static long newOptimisticTransactionOptions();
+  private native boolean isSetSnapshot(final long handle);
+  private native void setSetSnapshot(final long handle,
+      final boolean setSnapshot);
+  private native void setComparator(final long handle,
+      final long comparatorHandle);
+  @Override protected final native void disposeInternal(final long handle);
+}

java/src/main/java/org/rocksdb/Options.java

@@ -170,7 +170,7 @@ public class Options extends RocksObject
       final AbstractComparator<? extends AbstractSlice<?>> comparator) {
     assert(isOwningHandle());
     setComparatorHandle(nativeHandle_, comparator.nativeHandle_,
-            comparator instanceof DirectComparator);
+            comparator.getComparatorType().getValue());
     comparator_ = comparator;
     return this;
   }
@@ -1734,7 +1734,7 @@ public class Options extends RocksObject
       long memtableMemoryBudget);
   private native void setComparatorHandle(long handle, int builtinComparator);
   private native void setComparatorHandle(long optHandle,
-      long comparatorHandle, boolean isDirect);
+      long comparatorHandle, byte comparatorType);
   private native void setMergeOperatorName(
       long handle, String name);
   private native void setMergeOperator(

java/src/main/java/org/rocksdb/RocksDB.java

@@ -435,7 +435,7 @@ public class RocksDB extends RocksObject {
         path));
   }
 
-  private void storeOptionsInstance(DBOptionsInterface options) {
+  protected void storeOptionsInstance(DBOptionsInterface options) {
     options_ = options;
   }
 
@@ -1683,7 +1683,7 @@ public class RocksDB extends RocksObject {
    * @return The handle of the default column family
    */
   public ColumnFamilyHandle getDefaultColumnFamily() {
-    ColumnFamilyHandle cfHandle = new ColumnFamilyHandle(this,
+    final ColumnFamilyHandle cfHandle = new ColumnFamilyHandle(this,
         getDefaultColumnFamily(nativeHandle_));
     cfHandle.disOwnNativeHandle();
     return cfHandle;
@@ -2359,8 +2359,9 @@ public class RocksDB extends RocksObject {
       final long[] columnFamilyHandles, final long readOptHandle)
       throws RocksDBException;
   protected native long getSnapshot(long nativeHandle);
-  protected native void releaseSnapshot(long nativeHandle, long snapshotHandle);
-  @Override protected final native void disposeInternal(final long handle);
+  protected native void releaseSnapshot(
+      long nativeHandle, long snapshotHandle);
+  @Override protected native void disposeInternal(final long handle);
   private native long getDefaultColumnFamily(long handle);
   private native long createColumnFamily(final long handle,
       final byte[] columnFamilyName, final long columnFamilyOptions)