This is a collection of Cross-Platform libraries written in C. It is a work in progress.
Provides 2 types: GCU_float64_t and GCU_float32_t which are generated during the build process to be correct for the system on which it is being compiled.
Provides type unions based on bit size for use in other parts of this library. Names are GCU_Type64_Union, GCU_Type32_Union, GCU_Type16_Union, and GCU_Type8_Union. Union contains all basic types that will fit into that bit size. Pointers, for example, only exist in the 64-bit union. The programmer is responsible for the memory management of the pointed-to data.
Provides functions gcu_malloc(), gcu_calloc(), gcu_realloc(), and gcu_free() which are used by all other parts of the library. Calling gcu_mem_start() and gcu_mem_stop() will cause all calls to the afore-mentioned memory functions to be logged to stderr, including the calling location and the memory locations involved, making memory errors easy to track down.
Provides several functions which will calculate a hash on a set of bytes using the Murmur3 algorithm.
Provides hash tables that hold 8, 16, 32, and 64-bit values.
The programmer must supply a hash value which will uniquely identify the object to be stored/retrieved, but the string.h library provides a good and fast helper algorithm, Murmur3, to make this easy.
The hash table will also have a mutex, but it is the programmer's responsibility to use it when appropriate.
The programmer may provide a cleanup function which will be called when the hash table is destroyed.
Provides a generalized vector structure that, similar to the hash tables, will hold 8, 16, 32, and 64-bit values.
The vector will also have a mutex, but it is the programmer's responsibility to use it when appropriate.
The programmer may provide a cleanup function which will be called when the vector is destroyed.
Provides a thread abstraction layer to better manage threads and information about the threads.
Provides a mutex abstraction for use as a low-level synchronization tool.
Provides a counting semaphore implementation with a user-configurable limit. A counting semaphore with a limit of 1 will be, in effect, a binary semaphore.
All librarys contain a corresponding test written in C++ (demonstrating that the library can be used in C++ as well as C) using the Google Test (gtest) framework.
All prototypes, typedefs, and defines are documented using Doxygen. Documentation should be available under the /docs folder.
make
make install
Still working on this, but will probably focus on the Mingw toolchain.