Re:Rpmalloc – A lock-free memory allocator
by maniccoder on Monday March 20, 2017 @05:38PM
(#13917908)
Thank you for the clarifications, I will definitely improve the documentation w.r.t trouble cases, fragmentation and similar allocators.
The goal is of course not to replace the standard allocator, but rather to offer an alternative for those cases where the standard allocator performs poorly.
I've never had a desire to deviate from my underlying library's libc allocator, but the principle reason for that (aside from potential for mismatched malloc/free calls due to integration errors), is that I know this stuff is a dark art to nail correctly, and I'm opening myself up to all kinds of potential unknowns by discarding the finely tweaked (although potentially not the most efficient) libc allocator that's been getting patched on to handle adversity for the past 20-30 years.
I'd expect documentation of this sort to appear in the readme, even if just to make mention of another allocator this one is patterned on, where I could maybe go digging for further information around failure cases
This is just a conservative engineer talking, I've never actually written my own general purpose heap allocator, although I understand the design of a few, but perhaps not sufficiently to understand the motivation behind the design (e.g. the kinds of failure cases the design is hoping to avoid). Just more interested in how it's likely to break my app :)