|
|
|
|
|
|
by rurban
3344 days ago
|
|
|
I hope you do realize that a good GC is usually faster then refcounting, and comparing it to slow GC's does not prove anything. For fairness you need to compare Rust to D or Pony or SBCL, which are also close to C++ (even faster), plus added concurrency safety or memory safety, which can be circumvented in Rust. |
|
|
In idiomatic Rust code, you typically have very few reference increments/decrements, making it faster and more efficient than both, GC and traditional RC-based approaches.
The reasons for this are that objects are often allocated on the stack, passed by reference (a safe pointer), and directly integrated into a larger structure, requiring fewer heap allocations and very few – if any – refcounted objects. In Rust, unlike C or C++, you can do this safely and ergonomically because Rust enforces clear and well-defined ownership semantics.