anidmap
This is a small library that will be used primarily in the Alux Operating System. It facilitates the allocation of unique identifiers and the association of these identifiers with objects.
In the context of operating systems, this could be used to allocate file descriptors, process IDs, outgoing TCP/UDP port numbers, and more.
Maps
anidmap includes a fixed-size hashmap structure HashMap for mapping objects to their identifiers. This is a subclass of the Map abstract class. In the future, I may create other methods of ID-object association, such as structures based on binary trees.
Identifiers
An IdAllocator subclass can be used to allocate unique identifiers. Currently, there are two subclasses of this:
- The PoolIdAllocator uses an expanding memory pool to track identifiers that have been freed.
- The StepIdAllocator continually increments a counter until it needs to wrap around. To wrap around the identifier counter, StepIdAllocator finds a chunk of unused identifiers in O(n^2) time (with a hash-map backing).
Identifier Maps
The IdMap class can be used to join the functionality of an IdAllocator with that of a Map. Since allocators may depend on a map to find unused IDs, you must use a specialized IdMap subclass for the type of IdAllocator you would like to use.
Currently, the StepIdMap and PoolIdMap classes serve as IdMap subclasses.
Dependencies
anidmap uses ansa for its locking mechanism.
The following standard headers are required:
- <cstdint> - integer types such as
int_fast32_t
- <cassert> -
assert
macro - <cstddef> -
NULL
definition
Configuration
You may notice that anidmap includes a file that it does not provide itself:
#include <anidmap/lock>
You must create this include file. It must define a class called anidmap::Lock
that is a subclass of ansa::Lock
. The anidmap::Lock
class will be used for all synchronization in the library. Usually, you will be able to achieve this using the GCC/Clang -I
directive. See the lock implementation based on stdc++ in test/stdcpp-lock.cpp.