[Unreleased]

Added

• NIL

Changed

• NIL

Deprecated

• NIL

Removed

• NIL

Fixed

• NIL

Security

• NIL

[0.4.0] - 2026-02-15

Added

• WebAssembly/Emscripten support with automatic software fallback for SSE4.2 intrinsics

Changed

• CMake Simplification: Drastically simplified CMake configuration for header-only library
  • Removed multi-config export loop (header-only libraries don't need configuration-specific targets)
  • Removed unnecessary NFX_HASHING_LIB_DIR variable from package config
  • Cleaned up packaging to source-only generation (binary packages unnecessary for header-only library)
  • Removed binary packaging options: PACKAGE_ARCHIVE, PACKAGE_DEB, PACKAGE_RPM, PACKAGE_WIX

Removed

• NFX_HASHING_ENABLE_SIMD CMake option (header-only library - SIMD controlled by user's compile flags)
  • All SIMD compile options from INTERFACE target (users manage their own optimization flags)
• Binary Packaging: Removed all binary package generation (TGZ/ZIP archives, DEB, RPM, WIX MSI)
  • Users should use FetchContent, git submodules, or GitHub source archives

[0.3.1] - 2026-02-14

Fixed

• Critical: Removed strict compiler warnings from library INTERFACE target

[0.3.0] - 2026-02-14

Added

• Strict Compiler Warnings: Enabled comprehensive warning flags for all compilers
  • GCC/Clang: -Wall -Wextra -Werror (treat warnings as errors)
  • MSVC: /W4 /WX (warning level 4, treat warnings as errors)
• Improved code quality enforcement at compile time

[0.2.0] - 2026-02-08

Added

• Automatic SIMD Optimization: Compiler flags for SSE4.2 hardware acceleration are now automatically configured in Release builds
• NFX_HASHING_ENABLE_SIMD option to control SIMD optimizations (default: ON)
• GCC/Clang: automatic -march=native (or -msse4.2 fallback)
• MSVC: automatic /arch:AVX for SSE4.2 support
• Debug builds intentionally disable SIMD to allow testing software fallback paths

[0.1.2] - 2025-11-27

Changed

• Consolidated packaging tool detection in CMake configuration

Fixed

• Removed incorrect runtime dependencies from DEB/RPM packages

[0.1.1] - 2025-11-15

Changed

• Performance Optimization: Added compile-time SSE4.2 fast path in crc32c() to eliminate runtime CPU detection overhead when compiled with -march=native, -msse4.2, or /arch:AVX flags
• Benchmark Accuracy: Fixed single-step hash benchmarks to use dependency chains, preventing compiler pre-computation and ensuring realistic performance measurements

[0.1.0] - 2025-11-13 - Initial Release

Added

• hash<T>: Template-based hash function for clean, STL-style hashing
• Hasher<HashType, Seed>: General-purpose STL-compatible hash functor
• CRC32-C (Castagnoli): Hardware-accelerated hash using SSE4.2 intrinsics with software fallback
  • Polynomial: 0x1EDC6F41 (Castagnoli polynomial)
  • Runtime CPU detection: checks SSE4.2 support at first call, caches result
  • Hardware: uses _mm_crc32_u8 (MSVC) or __builtin_ia32_crc32qi (GCC/Clang)
  • Software: pure C++ implementation matching SSE4.2 behavior
  • Single binary deployment: optimally runs on both old and new CPUs
  • Dual-stream 64-bit implementation: low 32 bits (normal bytes), high 32 bits (inverted bytes byte ^ 0xFF)
• FNV-1a: Fast non-cryptographic hash (32-bit and 64-bit variants)
  • XOR-multiply algorithm with prime constants
  • Fully constexpr for compile-time hashing
• Larson Hash: Simple multiplicative hash (37 * hash + ch)
  • Provided for benchmarking comparisons
  • Supports 32-bit and 64-bit variants
• Integer Hashing: Optimized multiplicative hashing for integral types
  • 32-bit: Knuth's multiplicative constant (0x45d9f3b)
  • 64-bit: Wang's avalanche mixing (0xbf58476d1ce4e5b9, 0x94d049bb133111eb)
• Hash Combining: Two combination strategies
  • FNV-1a style: XOR then multiply by prime
  • Boost + MurmurHash3 finalizer: golden ratio mixing with avalanche finalization
• Seed Mixing: Hash table probing utilities
  • 32-bit: Thomas Wang's bit-mixing algorithm
  • 64-bit: MurmurHash3 avalanche mixing
• Samples
  • Sample_Hash.cpp - Unified hash<T> API demonstration
  • Sample_Hasher.cpp - STL container integration with Hasher<> functor
  • Sample_LowLevelHashing.cpp - Direct use of low-level hash primitives
  • Sample_AdvancedHashing.cpp - Advanced techniques (custom seeds, transparent lookup)
• Build System
  • CMake 3.20+ configuration with modular structure
  • Header-only interface library target
  • Cross-platform support (Linux GCC/Clang, Windows MinGW/Clang/MSVC)
  • Google Test integration for unit testing
  • Google Benchmark integration for performance benchmarking
  • Doxygen documentation generation with GitHub Pages deployment
  • CPack packaging for distribution (DEB, RPM, TGZ, ZIP, WiX MSI)
  • Installation support with CMake package config files
• Documentation
  • README with feature overview and usage examples
  • Detailed API documentation with Doxygen comments and RFC references
  • Sample applications
  • Build and installation instructions
• Testing & Benchmarking
  • Unit test suite
  • Performance benchmarks for all operations
  • Cross-compiler performance validation
• CI/CD: GitHub Actions workflows for automated testing, documentation and release deployment