SimCList 1.2

SimCList project is a high quality C library for handling lists.
SimCList is available for free, under restrictions imposed by the BSD license.
SimCList API is good because:
- it is simple, yet powerful
- it makes elegant and consistent use of information hiding
- it abstracts the actual data type to store
- it is fairly total
The library itself is very performant and makes a good compromise between performance in terms of time and space:
- insertion is O(n) [typically n/8]
- extraction and deletion are O(n) [typically n/8]
- iteration is O(1)
- sorting is always O(n logn), without worst case
Thread safety
The SimCList library is thread safe, meaning that many threads can run SimCList operators on different lists concurrently without hurt, or also read operations on the same list.
However, of course, no safety guarantee is made for performing concurrently write+write or read+write operations on the same list (eg: insertion, deletion, sorting, ...). For performance and portability reasons, protecting such operations from concurrency is left to the library adopter.
Performance
SimCList has been designed with ease to use and performance in mind. There is some example factors that have been taken into account, and against which the code has been optimized when implementing SimCList:
- overhead of function calls
- cache locality
- number of branches, possible CPU stalls or pipeline flushes
- dynamic memory allocation weight on the OS
- other mathematic/probabilistic optimizations for avoiding worst-case or improving average case behaviours in algorithms
Many parts of SimCLists code have been deeply improved with profiling analysis.
Enhancements:
- Transparent inclusion in C++ applications is natively supported.
- Sentinels are used to improve performance on list traversals.
- A problem was fixed with list_insert_at, which could output inconsistent lists when inserting into even-sized lists in position 0.