Crates.io | packed_bools |
lib.rs | packed_bools |
version | 0.1.0 |
source | src |
created_at | 2022-09-11 00:03:20.931311 |
updated_at | 2022-09-11 00:03:20.931311 |
description | a lightweight ergonomic bit manipulation library for packing booleans |
homepage | |
repository | https://github.com/cavemanloverboy/packed_bools |
max_upload_size | |
id | 662767 |
size | 10,622 |
Want to stick several bool
values in a struct? Using core
/std
, you pay the price of 1 byte of space for each value so that each individual bool
can be addressable. For memory/storage sensitive applications, we'd really like a more efficient alternative. Enter packed_bools
.
There are already several bit manipulation libraries on crates.io. A quick search will reveal half a dozen or so popular ones. What makes this one different? It is dev-friendly and ergonomic. When I speak of (or to) my cat, I don't refer to him by his spatial coordinates. I refer to him by his name, "Ron" (or "Ronald" if he just threw my earphones on the floor). Likewise, I don't want to have to refer to individual bits in some larger type as "bit 2" or "bit 5" of this or that type... and have to document it somewhere so that I don't forget which bit is which... and then also implement bitwise operations to carry out operations on said bits. I want to refer to the booleans by my chosen flag name -- I want to name my bits.
This library provides a derive macro PackedBooleans
with which you can annotate a (or several) core::primitive::u8
with #[pack_bools(..)]
, in which you can provide the names of up to 8 boolean flags (per u8
).
This library introduces no new types. It simply implements methods on your struct that manipulates and retrieves your bits by name.
use packed_bools::PackedBooleans;
#[derive(Default, PackedBooleans)]
pub struct FooStruct<'a, T=()> {
#[pack_bools(active, admin, discount, premium, og, frozen, kyc, tos_agree)]
pub booleans: u8,
pub an_option_with_generics: Option<&'a T>,
}
// Initialize your struct
let mut foo_struct = FooStruct::<()>::default();
// Set the values of each bit individually
foo_struct.set_active(true);
foo_struct.set_admin(false);
foo_struct.set_discount(true);
foo_struct.set_premium(false);
foo_struct.set_og(false);
foo_struct.set_frozen(true);
foo_struct.set_kyc(false);
foo_struct.set_tos_agree(true);
// Read the values of each bit individually
assert_eq!(foo_struct.get_active(), true);
assert_eq!(foo_struct.get_admin(), false);
assert_eq!(foo_struct.get_discount(), true);
assert_eq!(foo_struct.get_premium(), false);
assert_eq!(foo_struct.get_og(), false);
assert_eq!(foo_struct.get_frozen(), true);
assert_eq!(foo_struct.get_kyc(), false);
assert_eq!(foo_struct.get_tos_agree(), true);
// Do not do this!
// foo_struct.booleans += 5;
Yes, I know...a more sophisticated library would use a #[repr(transparent)]
wrapper around the u8
to ensure that the user does not manipulate the u8. Or perhaps extend use to u16
or u32
if a user needs >8 flags. Feel free to submit a PR if you want this in the library ;).