Crate heapless

Expand description

static friendly data structures that don’t require dynamic memory allocation

The core principle behind heapless is that its data structures are backed by a static memory allocation. For example, you can think of heapless::Vec as an alternative version of std::Vec with fixed capacity and that can’t be re-allocated on the fly (e.g. via push).

All heapless data structures store their memory allocation inline and specify their capacity via their type parameter N. This means that you can instantiate a heapless data structure on the stack, in a static variable, or even in the heap.

use heapless::Vec; // fixed capacity `std::Vec`
use heapless::consts::U8; // type level integer used to specify capacity

// on the stack
let mut xs: Vec<u8, U8> = Vec::new(); // can hold up to 8 elements
xs.push(42).unwrap();
assert_eq!(xs.pop(), Some(42));

// in a `static` variable
// static mut XS: Vec<u8, U8> = Vec::new(); // requires feature `const-fn`

// work around
static mut XS: Option<Vec<u8, U8>> = None;
unsafe { XS = Some(Vec::new()) };
let xs = unsafe { XS.as_mut().unwrap() };

xs.push(42);
assert_eq!(xs.pop(), Some(42));

// in the heap (though kind of pointless because no reallocation)
let mut ys: Box<Vec<u8, U8>> = Box::new(Vec::new());
ys.push(42).unwrap();
assert_eq!(ys.pop(), Some(42));

Because they have fixed capacity heapless data structures don’t implicitly reallocate. This means that operations like heapless::Vec.push are truly constant time rather than amortized constant time with potentially unbounded (depends on the allocator) worst case execution time (which is bad / unacceptable for hard real time applications).

heapless data structures don’t use a memory allocator which means no risk of an uncatchable Out Of Memory (OOM) condition (which defaults to abort) while performing operations on them. It’s certainly possible to run out of capacity while growing heapless data structures, but the API lets you handle this possibility by returning a Result on operations that may exhaust the capacity of the data structure.

List of currently implemented data structures:

BinaryHeap – priority queue
IndexMap – hash table
IndexSet – hash set
LinearMap
Queue – single producer single consumer lockless queue
String
Vec

In order to target the Rust stable toolchain, there are some feature gates. The features need to be enabled in Cargo.toml in order to use them. Once the underlaying features in Rust are stable, these feature gates might be activated by default.

Example of Cargo.toml:

...
[dependencies]
heapless = { version = "0.4.0", features = ["const-fn"] }
...

Currently the following features are availbale and not active by default:

"const-fn" – Enable the nightly const_fn feature and make most new methods const. This way they can be used to initialize static memory at compile time.