6 releases
0.1.5 | Mar 6, 2020 |
---|---|
0.1.4 | Apr 22, 2019 |
0.1.1 | Mar 30, 2019 |
0.1.0 | Oct 24, 2018 |
#906 in Algorithms
56 downloads per month
Used in 5 crates
12KB
213 lines
SWF Fixed
SWF fixed-point numbers for Rust.
A fixed point number represents a decimal values using evenly-distributed bit patterns (as opposed to floating point numbers where the density increases with the proximity to zero).
A fixed point number can be simply thought as an integer divided by a constant value.
It is described by its integer part and fractional part:
its mathematical value is integer_part + fractional_part / 2^fractional_bits
.
For example, the type Ufixed8p8
is an unsigned fixed point number with an
8 bit integer part and 8 bit fractional part. It can represent the 2^16 values
corresponding to u16 / 256
, the gap between each value (epsilon) is 1 / 256
.
This crate defines the fixed points numbers used by SWF files:
Name | Integer part | Fractional part | Min value | Max value | Epsilon |
---|---|---|---|---|---|
Sfixed8P8 | i8 |
u8 |
-128 | 128 - 1/256 | 1 / 256 |
Ufixed8P8 | u8 |
u8 |
0 | 256 - 1/256 | 1 / 256 |
Sfixed16P16 | i16 |
u16 |
-2^15 | 2^15 - 1/2^16 | 1 / 2^16 |
Ufixed16P16 | u16 |
u16 |
0 | 2^16 - 1/2^16 | 1 / 2^16 |
Usage
use swf_fixed::Sfixed8P8;
fn main() {
let a = Sfixed8P8::from_epsilons(256);
let b = Sfixed8P8::from_value(1f32);
assert_eq!(a, b);
let sum: Sfixed8P8 = (a + b);
let sum_value: f32 = sum.into();
assert_eq!(sum_value, 2.0f32);
}
Contributing
This library is a standard Cargo project. You can test your changes with
cargo test
.
Prefer non-master
branches when sending a PR so your changes can be rebased if
needed. All the commits must be made on top of master
(fast-forward merge).
CI must pass for changes to be accepted.
Dependencies
~105–335KB