kmdata -- Data structures for C programmers, by C programmers.

MISSION STATEMENT
kmdata provides a simple, generic, and open set of abstractions for data
structures in the GNU C99 language. kmdata structures are highly efficient,
performant, and portable without sacrificing generality. If you hate the STL
and the general bloat of C++ but find yourself reinventing the wheel over and
over again in C, kmdata has your back.

Breaking that down further:
Efficiency - kmdata chooses asymptotically optimal algorithms and structures.
Performance - All algorithms and structures are tested and optimized for the 
    x86/Linux platform.
Portability - Written to compile and perform out of the box on IA32 and amd64
    systems. I haven't attempted a port to another architecture yet, but I'm
    sure ARM is not out of the question.

DOCUMENTATION
There is some. Not too much. Users shouldn't call functions starting with an 
underscore. Other than that, most structures follow a similar pattern for usage.

dict_t dict;
dict_init(&dict);
...
dict_clear(&dict, DICTHT_FREE_KEYS | DICTHT_FREE_VALUES);

The above code declares and initializes a dict, does some stuff with it, and
clears it, freeing keys and values as it goes. Most clear functions have flags
for specifying whether keys or values (or both) should be freed as the data
structure is being cleared out. Users should always call the init function
before starting to use a structure, and the clear function when finished.
The exception is the tuple type, which is so simple it practically documents
itself.