Rename LLSM to TreeLine (#81)

* Codemod: llsm -> tl

* Codemod: LLSM -> TL

* include/llsm/ -> include/tl/

* Leftover compilation changes

* Support legacy use of llsm/pg_llsm in the experiments

* TL -> TreeLine in comments when appropriate

* include/tl -> include/treeline

* Remove references to Learned LSM

* Use full name in CMake targets

* Use the full TreeLine where meaningful

* Bump manifest version, TL -> TreeLine in a few more places

* Newlines and other nits

* Shorten manifest signature to TL

* Update CI build options

.github/workflows/ci.yml

@@ -29,7 +29,7 @@ jobs:
       run: sudo apt-get install libjemalloc-dev
 
     - name: Configure Project using CMake
-      run: cmake -B ${{github.workspace}}/build -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DLLSM_BUILD_TESTS=ON -DLLSM_BUILD_BENCHMARKS=OFF
+      run: cmake -B ${{github.workspace}}/build -DCMAKE_BUILD_TYPE=${{env.BUILD_TYPE}} -DTL_BUILD_TESTS=ON -DTL_BUILD_BENCHMARKS=OFF
 
     - name: Compile Project 
       working-directory: ${{github.workspace}}/build

AUTHORS

@@ -1,4 +1,4 @@
-# The Learned LSM project uses code and concepts from LevelDB and RocksDB. We
+# The TreeLine project uses code and concepts from LevelDB and RocksDB. We
 # reproduce the LevelDB and RocksDB AUTHORS file below.
 
 Facebook Inc.

CMakeLists.txt

@@ -1,10 +1,10 @@
 cmake_minimum_required(VERSION 3.17)
-project(learnedlsm LANGUAGES CXX C)
+project(treeline LANGUAGES CXX C)
 set(CMAKE_CXX_STANDARD 17)
 
-option(LLSM_BUILD_TESTS "Set to build the Learned LSM test suite." OFF)
-option(LLSM_BUILD_BENCHMARKS "Set to build the Learned LSM benchmarks." OFF)
-option(LLSM_BUILD_SHARED "Set to build the Learned LSM database as a shared library." OFF)
+option(TL_BUILD_TESTS "Set to build the TreeLine test suite." OFF)
+option(TL_BUILD_BENCHMARKS "Set to build the TreeLine benchmarks." OFF)
+option(TL_BUILD_SHARED "Set to build TreeLine as a shared library." OFF)
 
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")  # Needed for ALEX
 
@@ -18,42 +18,42 @@ find_library(tbb tbb)
 # Load any third party dependencies
 add_subdirectory(third_party)
 
-# The Learned LSM embedded database library targets
-if (LLSM_BUILD_SHARED)
-  add_library(llsm SHARED)
-  add_library(pg_llsm SHARED)
+# The TreeLine embedded database library targets
+if (TL_BUILD_SHARED)
+  add_library(treeline SHARED)
+  add_library(pg_treeline SHARED)
 else()
-  add_library(llsm STATIC)
-  add_library(pg_llsm STATIC)
+  add_library(treeline STATIC)
+  add_library(pg_treeline STATIC)
 endif()
-target_link_libraries(llsm PRIVATE crc32c tbb masstree)
-target_link_libraries(pg_llsm PRIVATE crc32c tbb masstree)
-target_include_directories(llsm
+target_link_libraries(treeline PRIVATE crc32c tbb masstree)
+target_link_libraries(pg_treeline PRIVATE crc32c tbb masstree)
+target_include_directories(treeline
   PUBLIC include
   PRIVATE .)
-target_include_directories(pg_llsm
+target_include_directories(pg_treeline
   PUBLIC include
   PRIVATE .)
 
 # Add API headers to the target for IDE support
-set(llsm_inc include/llsm)
-set(llsm_api
-  ${llsm_inc}/db.h
-  ${llsm_inc}/options.h
-  ${llsm_inc}/record_batch.h
-  ${llsm_inc}/slice.h
-  ${llsm_inc}/statistics.h
-  ${llsm_inc}/status.h
+set(treeline_inc include/treeline)
+set(treeline_api
+  ${treeline_inc}/db.h
+  ${treeline_inc}/options.h
+  ${treeline_inc}/record_batch.h
+  ${treeline_inc}/slice.h
+  ${treeline_inc}/statistics.h
+  ${treeline_inc}/status.h
 )
-target_sources(llsm PUBLIC ${llsm_api})
-set(pg_llsm_api
-  ${llsm_inc}/pg_db.h
-  ${llsm_inc}/pg_options.h
-  ${llsm_inc}/pg_stats.h
-  ${llsm_inc}/slice.h
-  ${llsm_inc}/status.h
+target_sources(treeline PUBLIC ${treeline_api})
+set(pg_treeline_api
+  ${treeline_inc}/pg_db.h
+  ${treeline_inc}/pg_options.h
+  ${treeline_inc}/pg_stats.h
+  ${treeline_inc}/slice.h
+  ${treeline_inc}/status.h
 )
-target_sources(pg_llsm PUBLIC ${pg_llsm_api})
+target_sources(pg_treeline PUBLIC ${pg_treeline_api})
 
 # Add our sources by traversing the repository
 add_subdirectory(alex)
@@ -63,20 +63,19 @@ add_subdirectory(db)
 add_subdirectory(masstree)
 add_subdirectory(masstree_wrapper)
 add_subdirectory(model)
+add_subdirectory(page_grouping)
 add_subdirectory(record_cache)
 add_subdirectory(tlx)
 add_subdirectory(util)
 add_subdirectory(wal)
 
-add_subdirectory(page_grouping)
-
 # Build the tests iff requested
-if(LLSM_BUILD_TESTS)
+if(TL_BUILD_TESTS)
   enable_testing()
   add_subdirectory(tests)
 endif()
 
 # Build the benchmarks iff requested
-if(LLSM_BUILD_BENCHMARKS)
+if(TL_BUILD_BENCHMARKS)
   add_subdirectory(bench)
 endif()

README.md

@@ -1,6 +1,6 @@
 ![](https://github.com/mitdbg/learnedlsm/actions/workflows/ci.yml/badge.svg)
 
-# Learned LSM
+# TreeLine
 An embedded key-value store for modern SSDs.
 
 ## Building from source

alex/CMakeLists.txt

@@ -1,4 +1,4 @@
-target_sources(llsm PRIVATE
+target_sources(treeline PRIVATE
   alex_base.h
   alex_fanout_tree.h 
   alex_map.h

art_olc/CMakeLists.txt

@@ -11,5 +11,5 @@ set(art_sources
   Tree.cpp
   Tree.h)
 
-target_sources(llsm PRIVATE ${art_sources})
-target_sources(pg_llsm PRIVATE ${art_sources})
+target_sources(treeline PRIVATE ${art_sources})
+target_sources(pg_treeline PRIVATE ${art_sources})

art_olc/Tree.cpp

@@ -347,7 +347,7 @@ bool Tree::lookupRange(const Key& start, const Key& end, Key& continueKey,
 
 bool Tree::lookupRange(const Key& start, TID result[], std::size_t resultSize,
                        std::size_t& resultsFound, ThreadInfo& threadEpocheInfo,
-                       std::vector<llsm::RecordCacheEntry>* cache_entries,
+                       std::vector<tl::RecordCacheEntry>* cache_entries,
                        Key* continueKey,
                        std::optional<uint64_t> index_locked_already) const {
   EpocheGuard epocheGuard(threadEpocheInfo);

art_olc/Tree.h

@@ -77,7 +77,7 @@ class Tree {
   bool lookupRange(
       const Key& start, TID result[], std::size_t resultLen,
       std::size_t& resultCount, ThreadInfo& threadEpocheInfo,
-      std::vector<llsm::RecordCacheEntry>* cache_entries = nullptr,
+      std::vector<tl::RecordCacheEntry>* cache_entries = nullptr,
       Key* continueKey = nullptr,
       std::optional<uint64_t> index_locked_already = std::nullopt) const;
 

azure-ci-job.yml

@@ -64,8 +64,8 @@ jobs:
           # takes a lot of resources to build.
           cmakeArgs: >
             -DCMAKE_BUILD_TYPE=${{ parameters.buildType }}
-            -DLLSM_BUILD_TESTS=ON
-            -DLLSM_BUILD_BENCHMARKS=OFF
+            -DTL_BUILD_TESTS=ON
+            -DTL_BUILD_BENCHMARKS=OFF
             ..
 
       - bash: make -j

bench/CMakeLists.txt

@@ -1,4 +1,4 @@
-# Common utilities used in the LLSM benchmarks.
+# Common utilities used in the TreeLine benchmarks.
 add_library(bench_common
   common/data.cc
   common/data.h
@@ -7,13 +7,13 @@ add_library(bench_common
   common/startup.cc
   common/startup.h
   common/timing.h)
-# Any targets linking to `bench_common` will also link to `llsm` and will be
-# able to include LLSM internal headers.
-target_link_libraries(bench_common PUBLIC llsm)
+# Any targets linking to `bench_common` will also link to `treeline` and will be
+# able to include TreeLine internal headers.
+target_link_libraries(bench_common PUBLIC treeline)
 target_include_directories(bench_common PUBLIC ${PROJECT_SOURCE_DIR})
 
 # All `bench_common` utilities plus configuration utilities used in benchmarks
-# comparing LLSM against RocksDB.
+# comparing TreeLine against RocksDB.
 #
 # We use this additional library target because not all benchmark executables
 # need to link to RocksDB and gflags (e.g., `microbench`).
@@ -22,25 +22,31 @@ add_library(bench_common_config
   common/config.h
   common/kvell_interface.h
   common/leanstore_interface.h
-  common/llsm_interface.h
-  common/pg_llsm_interface.h
+  common/treeline_interface.h
+  common/pg_treeline_interface.h
   common/rocksdb_interface.h)
-target_link_libraries(bench_common_config PUBLIC bench_common gflags rocksdb leanstore pg_llsm libcuckoo)
+target_link_libraries(bench_common_config PUBLIC
+  bench_common
+  gflags
+  rocksdb
+  leanstore
+  pg_treeline
+  libcuckoo)
 
 # Synthetic Write: Measure the write throughput when writing synthetic data.
-# This benchmark compares the Learned LSM database against RocksDB.
+# This benchmark compares TreeLine against RocksDB.
 add_executable(synth_write synth_write.cc)
 target_link_libraries(synth_write bench_common_config)
 
-# YCSB: Run extracted YCSB workloads against LLSM and RocksDB.
+# YCSB: Run extracted YCSB workloads against TreeLine and RocksDB.
 add_executable(ycsb ycsb.cc)
 target_link_libraries(ycsb bench_common_config ycsbr)
 
 # Hash Table YCSB: Run extracted YCSB workloads against different hash table implementations.
 add_executable(hashtable_ycsb hashtable_ycsb.cc)
 target_link_libraries(hashtable_ycsb Threads::Threads gflags ycsbr)
 
-# Microbench: An executable containing all of LLSM's microbenchmarks.
+# Microbench: An executable containing all of TreeLine's microbenchmarks.
 add_executable(microbench
   memtable_benchmark.cc
   packed_map_benchmark.cc
@@ -71,7 +77,7 @@ target_link_libraries(wal bench_common gflags benchmark::benchmark)
 add_executable(bufmgr_benchmark buffer_manager_benchmark.cc)
 target_link_libraries(bufmgr_benchmark bench_common_config gflags ycsbr benchmark::benchmark)
 
-# Run Custom: An executable that runs YSCBR-generated workloads against LLSM
+# Run Custom: An executable that runs YSCBR-generated workloads against TreeLine
 # and RocksDB.
 add_executable(run_custom run_custom.cc)
 target_link_libraries(run_custom bench_common_config ycsbr-gen)

bench/README.md

@@ -1,4 +1,3 @@
-# Learned LSM Benchmarks
+# TreeLine Benchmarks
 
-This directory contains various benchmarks used to evaluate the Learned LSM
-database.
+This directory contains various benchmarks used to evaluate TreeLine.

bench/buffer_manager_benchmark.cc

@@ -4,7 +4,7 @@
 #include "common/config.h"
 #include "db/page.h"
 #include "gflags/gflags.h"
-#include "llsm/options.h"
+#include "treeline/options.h"
 #include "ycsbr/ycsbr.h"
 
 namespace {
@@ -39,18 +39,18 @@ void SimulateBM(benchmark::State& state, bool large_buffer) {
       ycsbr::BulkLoadTrace::LoadFromFile(FLAGS_load_path, loptions);
 
   // Create key hints.
-  llsm::KeyDistHints key_hints;
+  tl::KeyDistHints key_hints;
   key_hints.num_keys = load.size();
-  key_hints.page_fill_pct = FLAGS_llsm_page_fill_pct;
+  key_hints.page_fill_pct = FLAGS_tl_page_fill_pct;
   key_hints.record_size = record_size;
 
   // Initialize an alex instance.
-  alex::Alex<uint64_t, llsm::PhysicalPageId> model;
+  alex::Alex<uint64_t, tl::PhysicalPageId> model;
   size_t records_per_page = key_hints.records_per_page();
   size_t i = 0;
   for (const auto& req : load) {
     if (i % records_per_page == 0) {
-      llsm::PhysicalPageId page_id(0, i / records_per_page);
+      tl::PhysicalPageId page_id(0, i / records_per_page);
       model.insert(__builtin_bswap64(req.key), page_id);
     }
     ++i;
@@ -65,25 +65,25 @@ void SimulateBM(benchmark::State& state, bool large_buffer) {
       ycsbr::Trace::LoadFromFile(FLAGS_workload_path, woptions);
 
   // Pre-calculate page numbers appropriately
-  std::vector<llsm::PhysicalPageId> page_ids;
+  std::vector<tl::PhysicalPageId> page_ids;
   for (const auto& req : workload) {
     page_ids.push_back(
         *model.get_payload_last_no_greater_than(__builtin_bswap64(req.key)));
   }
 
   // Create buffer manager options.
-  llsm::BufMgrOptions bm_options;
+  tl::BufMgrOptions bm_options;
   bm_options.simulation_mode = true;
   bm_options.buffer_pool_size = large_buffer
-                                    ? model.size() * llsm::Page::kSize
+                                    ? model.size() * tl::Page::kSize
                                     : 64 * 1024 * 1024;
 
   // Bookkeeping
   std::string db_path = "test";
   double numIOs = 0;
 
   for (auto _ : state) {
-    llsm::BufferManager buf_mgr(bm_options, db_path);
+    tl::BufferManager buf_mgr(bm_options, db_path);
     for (const auto& page_id : page_ids) {
       auto& bf = buf_mgr.FixPage(page_id, /*exclusive = */ true);
       if (bf.IsNewlyFixed()) ++numIOs;