fhex

fhex is a Rust crate that provides the ToHex trait for converting floating-point numbers to their hexadecimal representation. The trait is implemented for f32 and f64 types.

The hexadecimal representation follows the format specified for the IEEE 754 standard, ±0xh.hhhp±d, described below.

To parse hexadecimal floating point numbers, see the hexf crate.

Usage

Add fhex to your Cargo.toml dependencies:

[dependencies]
fhex = "1.0.0"

Then, import the ToHex trait in your code:

use fhex::ToHex;

You can now call the to_hex method on any f32 or f64 value:

let num: f32 = 1.23;
let hex = num.to_hex();
println!("{}", hex);

This will print the hexadecimal representation of 1.23.

0x1.3ae148p+0

Format

Floating point numbers are represented in the format: ±0xh.hhhp±d, where:

• ± is the sign of the number, although + is omitted for positive numbers.
• 0x is the prefix for hexadecimal numbers.
• h.hhh is the significand (also known as the mantissa), represented as a hexadecimal number.
• p indicates that the following number is the exponent.
• ±d is the exponent, represented as a decimal number.

Special cases:

• ±0x0p+0 is used to represent zero.
• ±inf is used to represent infinity.
• nan is used to represent NaN (a "quiet NaN").
• nan:0x followed by a hexadecimal number is used to represent a NaN with a payload (a "signalling NaN").

Known differences with other implementations

• Go's %x format specifier for floating point numbers uses the format ±0xh.hhhp±d, but it insists on printing the exponent with at least two digits, zero padded. Infinity is represented as ±Inf and NaN as NaN. There is no special treatment of signalling NaNs.
• C++'s std::hexfloat format specifier for floating point numbers uses the format ±0xh.hhhp±d. There is no special treatment of signalling NaNs, they are just nan.

License

This project is licensed under the Apache License 2.0 - see the LICENSE file for details.