/nanovec-rs

Space-saving packed arrays of small integers.

Primary LanguageRustMIT LicenseMIT

nanovec: Arrays and friends, packed in an integer or two.

Crates.io docs.rs

Ever felt the need to store a few small integers, but a Vec (or even tinyvec) takes up more space than you'd like?

nanovec offers both fixed- and variable-length arrays of integers with limited range, all packed within one or two machine words that you can effortlessly lug around.

This crate:

  • is no_std.
  • is inspired by tinyvec, including the name.
  • has minimum dependencies.

Cheatsheet of types offered

[NanoArray] (trait) [NanoDeque] (adapter) [NanoStack] (adapter)
[NanoArrayBit] (impl) [NanoDequeBit] (alias) [NanoStackBit] (alias)
[NanoArrayRadix] (impl) [NanoDequeRadix] (alias) [NanoStackRadix] (alias)

Packed Arrays

Two space-saving strategies are offered: bit-packing and radix-packing. Both support the same set of operations defined as trait [NanoArray].

A wide unsigned integer (e.g. u64) can be treated as an array of narrower integers (e.g. 16 x 4-bit or 5 x 12-bit). Given the packed integer type (n bits) and the bit-width of each element (k bits), the capacity can be determined as floor(n / k). This is implemented as [NanoArrayBit].

Generalizing the bit-packing approach, a base-r integer can be treated as an array of integers in the range 0..r. A good example is a decimal (base-10) number --- 12345678 can be treated as an array of [8, 7, 6, 5, 4, 3, 2, 1] (least-significant digit first). This is implemented as [NanoArrayRadix].

Bit-packing is expected to perform better than radix-packing, as bit operations are cheaper than mul-div-mod. Therefore, bit-packing is the preferred approach, unless you need to squeeze in more elements when the element range is inconvenient for bit-packing (i.e. when r is only marginally larger than a power of two, but much smaller than the next power of two).

Adapters

Building upon the fixed-length packed arrays, the following variable-length data structures are offered:

  • [NanoDeque] implements a double-ended queue as [NanoArray] + length. This is the most versatile as it supports pushing/popping at both ends, and get/set at any index. The only drawback is the extra space and padding needed for storing the length.

  • [NanoStack] implements a zero-terminated stack backed by [NanoArray] alone, but its elements must be non-zero (think C-style strings with '\0' at the end). This supports pushing/popping at only one end, and get at any index.