Availability & Reproducibility (#16)

- no CI for publication, instead single master_script

CMakeLists.txt

@@ -25,7 +25,7 @@ if (CMAKE_SYSTEM_NAME STREQUAL "Linux" AND ALP_ENABLE_CLANG_TIDY)
     else ()
         set(CMAKE_CXX_CLANG_TIDY
                 ${CLANG_TIDY_EXE};
-                -header-filter=include/alp;
+                -header-filter=include/alp,data/include;
                 -warnings-as-errors=*;)
 
     endif ()

availability_reproducibility_initiative_report.md

@@ -18,7 +18,6 @@
     - figure 6 : todo
 
 5) Documentation on how to compile, deploy, run the code, and use the scripts:
-    - Follow the [Publication CI](.github/workflows/PUBLICATION.yaml)
     - In [BENCHMARKING.md](/BENCHMARKING.md) we detail how to replicate the experiments and benchmarks presented in
       our [publication](https://dl.acm.org/doi/pdf/10.1145/3626717).
 6) A link to [a single master script](publication/master_script/master_script.sh) that runs the experiments, collects

benchmarks/bench_compression_ratio/alp32.cpp

@@ -128,10 +128,10 @@ class alp32_test : public ::testing::Test {
  * This test will output and write a file with the estimated bits/value after compression with alp
  */
 TEST_F(alp32_test, test_alprd32_on_whole_datasets) {
-	std::ofstream ofile(alp_bench::PATHS.RESULT_DIR_PATH + "alp_rd32_compression_ratio.csv", std::ios::out);
+	std::ofstream ofile(alp_bench::get_paths().result_dir_path + "alp_rd32_compression_ratio.csv", std::ios::out);
 	ofile << "dataset,size,rowgroups_count,vectors_count\n";
 
-	for (auto& dataset : alp_bench::sp_datasets) {
+	for (auto& dataset : alp_bench::get_sp_datasets()) {
 		if (!dataset.suitable_for_cutting) { continue; }
 
 		std::cout << dataset.name << std::endl;

benchmarks/bench_compression_ratio/bench_alp_compression_ratio.cpp

@@ -73,7 +73,7 @@ void read_data(std::vector<double>& data, const std::string& csv_file_path, cons
 		file.seekg(0, std::ios::beg);
 
 		// Ensure the file size is a multiple of the size of a double
-		if (fileSize % sizeof(double) != 0) { throw std::runtime_error("File size is not a multiple of double size!"); }
+		// if (fileSize % sizeof(double) != 0) { throw std::runtime_error("File size is not a multiple of double size!"); }
 		// Calculate the number of doubles
 		std::size_t numDoubles = fileSize / sizeof(double);
 
@@ -327,14 +327,10 @@ class alp_test : public ::testing::Test {
  * This test will output and write a file with the estimated bits/value after compression with alp
  */
 TEST_F(alp_test, test_alp_on_whole_datasets) {
-	if (const auto v = std::getenv("ALP_DATASET_DIR_PATH"); v == nullptr) {
-		throw std::runtime_error("Environment variable ALP_DATASET_DIR_PATH is not set!");
-	}
-
-	std::ofstream ofile(alp_bench::PATHS.RESULT_DIR_PATH + "alp_compression_ratio.csv", std::ios::out);
+	std::ofstream ofile(alp_bench::get_paths().result_dir_path + "alp_compression_ratio.csv", std::ios::out);
 	ofile << "dataset,size,rowgroups_count,vectors_count\n";
 
-	for (auto& dataset : alp_bench::alp_dataset) {
+	for (auto& dataset : alp_bench::get_alp_dataset()) {
 		bench_alp_compression_ratio(dataset, ofile);
 	}
 }
@@ -344,19 +340,19 @@ TEST_F(alp_test, test_alp_on_whole_datasets) {
  * This test will output and write a file with the estimated bits/value after compression with alp
  */
 TEST_F(alp_test, test_alprd_on_whole_datasets) {
-	std::ofstream ofile(alp_bench::PATHS.RESULT_DIR_PATH + "alp_rd_compression_ratio.csv", std::ios::out);
+	std::ofstream ofile(alp_bench::get_paths().result_dir_path + "alp_rd_compression_ratio.csv", std::ios::out);
 	ofile << "dataset,size,rowgroups_count,vectors_count\n";
 
-	for (auto& dataset : alp_bench::alp_dataset) {
+	for (auto& dataset : alp_bench::get_alp_dataset()) {
 		bench_alp_rd_compression_ratio(dataset, ofile);
 	}
 }
 
 TEST_F(alp_test, test_alprd_on_evalimplsts) {
-	std::ofstream ofile(alp_bench::PATHS.RESULT_DIR_PATH + "evalimplsts.csv", std::ios::out);
+	std::ofstream ofile(alp_bench::get_paths().result_dir_path + "evalimplsts.csv", std::ios::out);
 	ofile << "dataset,size,rowgroups_count,vectors_count\n";
 
-	for (auto& dataset : alp_bench::evalimplsts) {
+	for (auto& dataset : alp_bench::get_evalimplsts()) {
 		bench_alp_rd_compression_ratio(dataset, ofile);
 	}
 }

benchmarks/bench_speed/bench_alp_cutter_decode.cpp

@@ -85,7 +85,7 @@ void benchmark_all(benchmark::Benchmark& benchmark) {
 	unffor_left_arr  = new (std::align_val_t {64}) uint16_t[VECTOR_SIZE];
 	glue_arr         = new (std::align_val_t {64}) double[VECTOR_SIZE];
 
-	for (auto& dataset : alp_bench::alp_dataset) {
+	for (auto& dataset : alp_bench::get_alp_dataset()) {
 		std::ifstream ifile(dataset.csv_file_path, std::ios::in);
 		if (!dataset.suitable_for_cutting) { continue; }
 		if (dataset.name.find("bw") != std::string::npos) { continue; }

benchmarks/bench_speed/bench_alp_cutter_encode.cpp

@@ -83,7 +83,7 @@ void benchmark_all(benchmark::Benchmark& benchmark) {
 	unffor_left_arr  = new (std::align_val_t {64}) uint16_t[VECTOR_SIZE];
 	glue_arr         = new (std::align_val_t {64}) double[VECTOR_SIZE];
 
-	for (auto& dataset : alp_bench::alp_dataset) {
+	for (auto& dataset : alp_bench::get_alp_dataset()) {
 		std::ifstream ifile(dataset.csv_file_path, std::ios::in);
 		if (!dataset.suitable_for_cutting) { continue; }
 		if (dataset.name.find("bw") != std::string::npos) { continue; }

benchmarks/bench_speed/bench_alp_encode.cpp

@@ -60,7 +60,7 @@ void benchmark_all(benchmark::Benchmark& benchmark) {
 	base_arr    = new (std::align_val_t {64}) int64_t[1024];
 	rg_smp_arr  = new (std::align_val_t {64}) double[1024];
 
-	for (auto& dataset : alp_bench::alp_dataset) {
+	for (auto& dataset : alp_bench::get_alp_dataset()) {
 		std::ifstream ifile(dataset.csv_file_path, std::ios::in);
 		if (dataset.suitable_for_cutting) { continue; }
 		if (dataset.name.find("bw") != std::string::npos) { continue; }

benchmarks/bench_speed/bench_alp_without_sampling.cpp

@@ -101,7 +101,7 @@ void benchmark_all(benchmark::Benchmark& benchmark) {
 	base_arr    = new (std::align_val_t {64}) int64_t[1024];
 	rg_smp_arr  = new (std::align_val_t {64}) double[1024];
 
-	for (auto& dataset : alp_bench::alp_dataset) {
+	for (auto& dataset : alp_bench::get_alp_dataset()) {
 		std::ifstream ifile(dataset.csv_file_path, std::ios::in);
 		if (dataset.suitable_for_cutting) { continue; }
 		if (dataset.name.find("bw") != std::string::npos) { continue; }

data/include/column.hpp

@@ -1,8 +1,6 @@
 #ifndef COLUMN_HPP
 #define COLUMN_HPP
 
-// NOLINTBEGIN
-
 #include <array>
 #include <cstdint>
 #include <string>
@@ -21,24 +19,24 @@ struct Column {
 };
 
 struct paths {
-	std::string GENERATED_COLUMNS_CSV_PATH = std::string {CMAKE_SOURCE_DIR} + "/data/generated/";
-	std::string ALP_DATASET_CSV_PATH       = std::string {CMAKE_SOURCE_DIR} + "/data/samples/";
-	std::string EDGE_DATASET_CSV_PATH      = std::string {CMAKE_SOURCE_DIR} + "/data/edge_case/";
-	std::string RESULT_DIR_PATH            = std::string {CMAKE_SOURCE_DIR} + "/publication/";
-	std::string EVALIMPLSTS_CSV_PATH         = std::string {CMAKE_SOURCE_DIR} + "/data/evalimplsts/";
-
-	std::string ALP_DATASET_BINARY_DIR_PATH = " ";
+	std::string generated_columns_csv_path  = std::string {CMAKE_SOURCE_DIR} + "/data/generated/";
+	std::string alp_dataset_csv_path        = std::string {CMAKE_SOURCE_DIR} + "/data/samples/";
+	std::string edge_dataset_csv_path       = std::string {CMAKE_SOURCE_DIR} + "/data/edge_case/";
+	std::string result_dir_path             = std::string {CMAKE_SOURCE_DIR} + "/publication/";
+	std::string evalimplsts_csv_path        = std::string {CMAKE_SOURCE_DIR} + "/data/evalimplsts/";
+	std::string alp_dataset_binary_dir_path = std::string {CMAKE_SOURCE_DIR} + "/data/full_data/";
 
 	explicit paths() {
-		auto v = std::getenv("ALP_DATASET_DIR_PATH");
-		if (v) { ALP_DATASET_BINARY_DIR_PATH = v; }
+		const auto v = std::getenv("ALP_DATASET_DIR_PATH");
+		if (v) { alp_dataset_binary_dir_path = v; }
 	}
 };
 
-inline paths PATHS;
+inline paths get_paths() {
+	static paths PATHS;
+	return PATHS;
+}
 
 } // namespace alp_bench
 
 #endif
-
-// NOLINTEND