Minimum Viable Product

[Elsklivet]

Trolloc hasn't gotten anywhere in a while. In order to do the interesting developments (making it troll people), it needs to first be an actual allocator. The MVP for trolloc is the ability to use it as a normal dynamic allocator. This issue will be updated with pertinent tasks and subtasks to achieve a simple allocator over time.

For now:

• [x] Allow allocation up to a maximum heap size
  • [x] First free
  • [ ] Expand heap by block+metadata size on reaching the end without hitting a free block
• [x] Allow allocated blocks to be freed
• [x] Allow allocated blocks to be reallocated

Eventually:

• [x] Ensure minimal fragmentation
  • [x] Block splitting (internal)
  • [x] Coalescence (external)
• [ ] Optimize searching algorithms
  • [ ] Choose one of
    • Best-fit
    • Next fit

[Elsklivet]

So, about heap expansion. I don't know if I feel like battling with dynamic memory allocation to reallocate the heap itself, nor fighting with sbrk.

Edit: A memory mapped file might actually be the move...

Edit edit: No, I think that would require a crate, and while I might accept begrudgingly allowing std in this project, I refuse to use any external crates.

[Elsklivet]

Closing this issue as cfcd7f3 and 0c70667 really covered what we plan on doing here for an MVP. We've moved on to GlobalAlloc in c1f29a1 and #3.