Merge pull request #272 from traceon/various-perf-optimizations

Various perf optimizations

.gitmodules

@@ -14,3 +14,7 @@
 	path = contrib/googletest
 	url = https://github.com/google/googletest.git
 	branch = master
+[submodule "contrib/folly"]
+	path = contrib/folly
+	url = https://github.com/facebook/folly.git
+	branch = master

CMakeLists.txt

@@ -45,6 +45,7 @@ cmake_dependent_option (CH_ODBC_ENABLE_TESTING "Enable test targets" ON "BUILD_T
 option (CH_ODBC_PREFER_BUNDLED_THIRD_PARTIES "Prefer bundled over system variants of third party libraries" ON)
 cmake_dependent_option (CH_ODBC_PREFER_BUNDLED_POCO "Prefer bundled over system variants of Poco library" ON "CH_ODBC_PREFER_BUNDLED_THIRD_PARTIES" OFF)
 cmake_dependent_option (CH_ODBC_PREFER_BUNDLED_SSL "Prefer bundled over system variants of SSL library" ON "CH_ODBC_PREFER_BUNDLED_POCO" OFF)
+cmake_dependent_option (CH_ODBC_PREFER_BUNDLED_FOLLY "Prefer bundled over system variants of Folly library" ON "CH_ODBC_PREFER_BUNDLED_THIRD_PARTIES" OFF)
 cmake_dependent_option (CH_ODBC_PREFER_BUNDLED_GOOGLETEST "Prefer bundled over system variants of Google Test library" ON "CH_ODBC_PREFER_BUNDLED_THIRD_PARTIES" OFF)
 cmake_dependent_option (CH_ODBC_PREFER_BUNDLED_NANODBC "Prefer bundled over system variants of nanodbc library" ON "CH_ODBC_PREFER_BUNDLED_THIRD_PARTIES" OFF)
 option (CH_ODBC_RUNTIME_LINK_STATIC "Link with compiler and language runtime statically" OFF)
@@ -131,6 +132,11 @@ if (NOT CH_ODBC_PREFER_BUNDLED_POCO)
     find_package (Poco COMPONENTS Foundation Net NetSSL)
 endif ()
 
+if (NOT CH_ODBC_PREFER_BUNDLED_FOLLY)
+    message (WARNING "Folly: using system variant of the library currently not supported")
+#   find_package (Folly)
+endif ()
+
 if (CH_ODBC_ENABLE_TESTING)
     if (NOT CH_ODBC_PREFER_BUNDLED_GOOGLETEST)
         find_package (GTest)

contrib/CMakeLists.txt

@@ -90,6 +90,15 @@ if (MSVC)
     endif ()
 endif ()
 
+# if (CH_ODBC_PREFER_BUNDLED_FOLLY OR NOT Folly_FOUND)
+#     if (NOT Folly_FOUND AND NOT CH_ODBC_PREFER_BUNDLED_FOLLY)
+#         message (WARNING "Folly: unable to find system Folly, falling back to using the bundled variant of the library")
+#     endif ()
+#
+#    add_subdirectory (folly)
+    set (FOLLY_INLUDE_DIR "${CMAKE_CURRENT_SOURCE_DIR}/folly" CACHE INTERNAL "")
+# endif ()
+
 if (CH_ODBC_ENABLE_TESTING)
     if (CH_ODBC_PREFER_BUNDLED_GOOGLETEST OR NOT GTEST_FOUND)
         if (NOT GTEST_FOUND AND NOT CH_ODBC_PREFER_BUNDLED_GOOGLETEST)

driver/CMakeLists.txt

@@ -132,6 +132,7 @@ if (UNICODE)
 endif ()
 
 target_include_directories (${libname}-impl
+    PUBLIC ${FOLLY_INLUDE_DIR}      # TODO: switch to linking with folly lib target?
     PUBLIC ${PROJECT_BINARY_DIR}
     PUBLIC ${PROJECT_SOURCE_DIR}
 )

driver/format/ODBCDriver2.cpp

@@ -1,6 +1,6 @@
 #include "driver/format/ODBCDriver2.h"
 
-ODBCDriver2ResultSet::ODBCDriver2ResultSet(std::istream & stream, std::unique_ptr<ResultMutator> && mutator)
+ODBCDriver2ResultSet::ODBCDriver2ResultSet(AmortizedIStreamReader & stream, std::unique_ptr<ResultMutator> && mutator)
     : ResultSet(stream, std::move(mutator))
 {
     std::int32_t num_header_rows = 0;
@@ -57,7 +57,7 @@ ODBCDriver2ResultSet::ODBCDriver2ResultSet(std::istream & stream, std::unique_pt
 }
 
 bool ODBCDriver2ResultSet::readNextRow(Row & row) {
-    if (raw_stream.peek() == EOF)
+    if (stream.eof())
         return false;
 
     for (std::size_t i = 0; i < row.fields.size(); ++i) {
@@ -68,32 +68,31 @@ bool ODBCDriver2ResultSet::readNextRow(Row & row) {
 }
 
 void ODBCDriver2ResultSet::readSize(std::int32_t & dest) {
-    constexpr auto size = sizeof(dest);
-    raw_stream.read(reinterpret_cast<char *>(&dest), size);
-
-    if (raw_stream.gcount() != size)
-        throw std::runtime_error("Incomplete result received, expected size: " + std::to_string(size));
+    stream.read(reinterpret_cast<char *>(&dest), sizeof(std::int32_t));
 }
 
 void ODBCDriver2ResultSet::readValue(std::string & dest, bool * is_null) {
     std::int32_t size = 0;
     readSize(size);
 
     if (size >= 0) {
-        dest.resize(size); // TODO: switch to uninitializing resize().
+        resize_without_initialization(dest, size);
 
         if (is_null)
             *is_null = false;
 
         if (size > 0) {
-            raw_stream.read(dest.data(), size);
-
-            if (raw_stream.gcount() != size)
-                throw std::runtime_error("Incomplete result received, expected size: " + std::to_string(size));
+            try {
+                stream.read(dest.data(), size);
+            }
+            catch (...) {
+                dest.clear();
+                throw;
+            }
         }
     }
     else /*if (size == -1) */{
-        dest.resize(0);
+        dest.clear();
 
         if (is_null)
             *is_null = true;
@@ -235,10 +234,10 @@ void ODBCDriver2ResultSet::readValue(std::string & src, DataSourceType<DataSourc
     return value_manip::from_value<std::string>::template to_value<DataSourceType<DataSourceTypeId::UUID>>::convert(src, dest);
 }
 
-ODBCDriver2ResultReader::ODBCDriver2ResultReader(std::istream & stream, std::unique_ptr<ResultMutator> && mutator)
-    : ResultReader(stream, std::move(mutator))
+ODBCDriver2ResultReader::ODBCDriver2ResultReader(std::istream & raw_stream, std::unique_ptr<ResultMutator> && mutator)
+    : ResultReader(raw_stream, std::move(mutator))
 {
-    if (stream.peek() == EOF)
+    if (stream.eof())
         return;
 
     result_set = std::make_unique<ODBCDriver2ResultSet>(stream, releaseMutator());

driver/format/ODBCDriver2.h

@@ -8,7 +8,7 @@ class ODBCDriver2ResultSet
     : public ResultSet
 {
 public:
-    explicit ODBCDriver2ResultSet(std::istream & stream, std::unique_ptr<ResultMutator> && mutator);
+    explicit ODBCDriver2ResultSet(AmortizedIStreamReader & stream, std::unique_ptr<ResultMutator> && mutator);
     virtual ~ODBCDriver2ResultSet() override = default;
 
 protected:
@@ -61,7 +61,7 @@ class ODBCDriver2ResultReader
     : public ResultReader
 {
 public:
-    explicit ODBCDriver2ResultReader(std::istream & stream, std::unique_ptr<ResultMutator> && mutator);
+    explicit ODBCDriver2ResultReader(std::istream & raw_stream, std::unique_ptr<ResultMutator> && mutator);
     virtual ~ODBCDriver2ResultReader() override = default;
 
     virtual bool advanceToNextResultSet() override;

driver/format/RowBinaryWithNamesAndTypes.cpp

@@ -2,7 +2,7 @@
 
 #include <ctime>
 
-RowBinaryWithNamesAndTypesResultSet::RowBinaryWithNamesAndTypesResultSet(std::istream & stream, std::unique_ptr<ResultMutator> && mutator)
+RowBinaryWithNamesAndTypesResultSet::RowBinaryWithNamesAndTypesResultSet(AmortizedIStreamReader & stream, std::unique_ptr<ResultMutator> && mutator)
     : ResultSet(stream, std::move(mutator))
 {
     std::uint64_t num_columns = 0;
@@ -35,7 +35,7 @@ RowBinaryWithNamesAndTypesResultSet::RowBinaryWithNamesAndTypesResultSet(std::is
 }
 
 bool RowBinaryWithNamesAndTypesResultSet::readNextRow(Row & row) {
-    if (raw_stream.peek() == EOF)
+    if (stream.eof())
         return false;
 
     for (std::size_t i = 0; i < row.fields.size(); ++i) {
@@ -53,10 +53,7 @@ void RowBinaryWithNamesAndTypesResultSet::readSize(std::uint64_t & res) {
     std::uint8_t shift = 0;
 
     while (true) {
-        auto byte = raw_stream.get();
-
-        if (raw_stream.fail() || byte == EOF)
-            throw std::runtime_error("Incomplete result received, expected: at least 1 more byte");
+        const int byte = stream.get();
 
         const std::uint64_t chunk = (byte & 0b01111111);
         const std::uint64_t segment = (chunk << shift);
@@ -80,11 +77,7 @@ void RowBinaryWithNamesAndTypesResultSet::readSize(std::uint64_t & res) {
 }
 
 void RowBinaryWithNamesAndTypesResultSet::readValue(bool & dest) {
-    auto byte = raw_stream.get();
-
-    if (raw_stream.fail() || byte == EOF)
-        throw std::runtime_error("Incomplete result received, expected size: 1");
-
+    const int byte = stream.get();
     dest = (byte != 0);
 }
 
@@ -95,11 +88,15 @@ void RowBinaryWithNamesAndTypesResultSet::readValue(std::string & res) {
 }
 
 void RowBinaryWithNamesAndTypesResultSet::readValue(std::string & dest, const std::uint64_t size) {
-    dest.resize(size); // TODO: switch to uninitializing resize().
-    raw_stream.read(dest.data(), dest.size());
+    resize_without_initialization(dest, size);
 
-    if (raw_stream.gcount() != dest.size())
-        throw std::runtime_error("Incomplete result received, expected size: " + std::to_string(size));
+    try {
+        stream.read(dest.data(), dest.size());
+    }
+    catch (...) {
+        dest.clear();
+        throw;
+    }
 }
 
 void RowBinaryWithNamesAndTypesResultSet::readValue(Field & dest, ColumnInfo & column_info) {
@@ -285,11 +282,7 @@ void RowBinaryWithNamesAndTypesResultSet::readValue(DataSourceType<DataSourceTyp
     char buf[16];
 
     static_assert(sizeof(dest.value) == lengthof(buf));
-
-    raw_stream.read(buf, lengthof(buf));
-
-    if (raw_stream.gcount() != lengthof(buf))
-        throw std::runtime_error("Incomplete result received, expected size: " + std::to_string(lengthof(buf)));
+    stream.read(buf, lengthof(buf));
 
     auto * ptr = buf;
 
@@ -300,10 +293,10 @@ void RowBinaryWithNamesAndTypesResultSet::readValue(DataSourceType<DataSourceTyp
     std::copy(ptr, ptr + lengthof(dest.value.Data4), std::make_reverse_iterator(dest.value.Data4 + lengthof(dest.value.Data4)));
 }
 
-RowBinaryWithNamesAndTypesResultReader::RowBinaryWithNamesAndTypesResultReader(std::istream & stream, std::unique_ptr<ResultMutator> && mutator)
-    : ResultReader(stream, std::move(mutator))
+RowBinaryWithNamesAndTypesResultReader::RowBinaryWithNamesAndTypesResultReader(std::istream & raw_stream, std::unique_ptr<ResultMutator> && mutator)
+    : ResultReader(raw_stream, std::move(mutator))
 {
-    if (stream.peek() == EOF)
+    if (stream.eof())
         return;
 
     result_set = std::make_unique<RowBinaryWithNamesAndTypesResultSet>(stream, releaseMutator());

driver/format/RowBinaryWithNamesAndTypes.h

@@ -8,7 +8,7 @@ class RowBinaryWithNamesAndTypesResultSet
     : public ResultSet
 {
 public:
-    explicit RowBinaryWithNamesAndTypesResultSet(std::istream & stream, std::unique_ptr<ResultMutator> && mutator);
+    explicit RowBinaryWithNamesAndTypesResultSet(AmortizedIStreamReader & stream, std::unique_ptr<ResultMutator> && mutator);
     virtual ~RowBinaryWithNamesAndTypesResultSet() override = default;
 
 protected:
@@ -23,11 +23,7 @@ class RowBinaryWithNamesAndTypesResultSet
 
     template <typename T>
     void readPOD(T & dest) {
-        constexpr auto size = sizeof(T);
-        raw_stream.read(reinterpret_cast<char *>(&dest), size);
-
-        if (raw_stream.gcount() != size)
-            throw std::runtime_error("Incomplete result received, expected size: " + std::to_string(size));
+        stream.read(reinterpret_cast<char *>(&dest), sizeof(T));
     }
 
     void readValue(Field & dest, ColumnInfo & column_info);
@@ -73,7 +69,7 @@ class RowBinaryWithNamesAndTypesResultReader
     : public ResultReader
 {
 public:
-    explicit RowBinaryWithNamesAndTypesResultReader(std::istream & stream, std::unique_ptr<ResultMutator> && mutator);
+    explicit RowBinaryWithNamesAndTypesResultReader(std::istream & raw_stream, std::unique_ptr<ResultMutator> && mutator);
     virtual ~RowBinaryWithNamesAndTypesResultReader() override = default;
 
     virtual bool advanceToNextResultSet() override;

driver/result_set.cpp

@@ -117,8 +117,8 @@ SQLRETURN Row::extractField(std::size_t column_idx, BindingInfo & binding_info)
     return fields[column_idx].extract(binding_info);
 }
 
-ResultSet::ResultSet(std::istream & stream, std::unique_ptr<ResultMutator> && mutator)
-    : raw_stream(stream)
+ResultSet::ResultSet(AmortizedIStreamReader & str, std::unique_ptr<ResultMutator> && mutator)
+    : stream(str)
     , result_mutator(std::move(mutator))
     , string_pool(1000000)
     , row_pool(1000000)
@@ -296,8 +296,8 @@ void ResultSet::retireRow(Row && row) {
     row_pool.put(std::move(row));
 }
 
-ResultReader::ResultReader(std::istream & stream, std::unique_ptr<ResultMutator> && mutator)
-    : raw_stream(stream)
+ResultReader::ResultReader(std::istream & raw_stream, std::unique_ptr<ResultMutator> && mutator)
+    : stream(raw_stream)
     , result_mutator(std::move(mutator))
 {
 }

driver/result_set.h

@@ -85,7 +85,7 @@ class ResultMutator {
 
 class ResultSet {
 public:
-    explicit ResultSet(std::istream & stream, std::unique_ptr<ResultMutator> && mutator);
+    explicit ResultSet(AmortizedIStreamReader & str, std::unique_ptr<ResultMutator> && mutator);
 
     virtual ~ResultSet();
 
@@ -111,7 +111,7 @@ class ResultSet {
     virtual bool readNextRow(Row & row) = 0;
 
 protected:
-    std::istream & raw_stream;
+    AmortizedIStreamReader & stream;
     std::unique_ptr<ResultMutator> result_mutator;
     std::vector<ColumnInfo> columns_info;
     std::deque<Row> row_set;
@@ -139,7 +139,7 @@ class ResultReader {
     virtual bool advanceToNextResultSet() = 0;
 
 protected:
-    std::istream & raw_stream;
+    AmortizedIStreamReader stream;
     std::unique_ptr<ResultMutator> result_mutator;
     std::unique_ptr<ResultSet> result_set;
 };