Release v1.4.0

Changes

• BREAKING CHANGE: Added last_published_index and header magic in shared memory header
  • Fixed shared-memory attach race by adding an init-state handshake with readiness wait and legacy fallback
  • Added last-published tracking so read_last() only returns published data; added a regression test
  • Changed read_last() return signature to std::pair<T*, uint32_t>
• Updated read_last() documentation for legacy shared-memory segments
• Relaxed memory ordering in the shared atomic cursor path for lower overhead
• Fixed release workflow changelog extraction to match version headers with trailing dates

Release v1.3.0

Changes

• BREAKING CHANGE: Replaced platform-specific shared memory implementation with slick-shm library
  • Removed all Windows-specific code (CreateFileMapping, MapViewOfFile, etc.)
  • Removed all POSIX-specific code (shm_open, mmap, etc.)
  • Unified shared memory implementation using slick::shm::shared_memory RAII wrapper
  • Code reduction: ~92 lines removed (14% reduction in queue.h)
  • Improved race condition handling: replaced sleep_for(1ms) with atomic creator flag
  • Added CREATOR_FLAG_OFFSET constant for atomic creator detection at offset 48 in header
  • Added comprehensive shared memory layout documentation
  • Simplified destructor using RAII automatic cleanup
  • All 22 tests pass (100% success rate)
  • No changes to public API - fully backward compatible at API level
  • Binary compatibility break: shared memory segments from previous versions are incompatible
• Fixed CMake configuration to properly handle slick-shm dependency in export targets
  • Removed SLICK_SHM_INSTALL OFF to allow slick-shm installation when fetched
  • Added find_dependency(slick-shm REQUIRED) in slick-queueConfig.cmake.in
  • Removed obsolete platform-specific precompiled headers (windows.h, sys/mman.h, etc.)

Release v1.2.3

Changes

• Initialize reserved counters for local and shared-memory queues
• Validate shared-memory size is power-of-two and element size matches when attaching
• Add lossy semantics documentation and debug loss detection counter
• Add lossy overwrite tests for in-process and shared-memory queues
• Align hot atomics to cache lines and add light spin backoff under contention
• Add reserve(1) fast path using fetch_add
• Add SLICK_QUEUE_ENABLE_CPU_RELAX to control pause/yield backoff
• Guard against zero-length reservations
• Add limits header

Release v1.2.2

Changes

• Rename repository from slick_queue to slick-queue (hyphenated naming follows recommended convention)
• Refactor repository name in CMake configuration files
• Update documentation and build references to use new repository name

Release v1.2.1

Changes

• Fixed a race condition when multi-process trying to create the same shared memory segment
• Fixed CTest integration by moving enable_testing() to root CMakeLists.txt and removing EXCLUDE_FROM_ALL
• Added release workflow

v1.2.0

• BREAKING CHANGE: Refactored reserved_info from struct to packed uint64_t for guaranteed lock-free atomics
  • Uses 48-bit index (supports 2^48 = 281 trillion iterations) and 16-bit size
  • std::atomic<reserved_info> is now always lock-free on all platforms
  • Reserve size limit of 65,535 (2^16 - 1) when using read_last() - see documentation
• Added HEADER_SIZE constant to replace magic number 64 in shared memory layout
• Added element size validation in shared memory to prevent type mismatches between processes
• Updated README with vcpkg installation instructions
• Updated README with lock-free implementation details and constraints documentation

v1.1.2

• Add alias target to provide the same slick::slick_queue name during build time

v1.1.1

• Updated CMakeLists.txt and added cmake config for vcpkg integration.
• Changed version number to three digits.

v1.1.0.2

• Changed include folder from include/slick_queue to include/slick
• Changed test framework from catch2 to googletest
• Rename slick_queue.h to queue.h

v1.1.0.1

• Fixed Linux implementation to read size from shared memory when opening existing segments
• Added size validation when opening existing shared memory with size mismatch