Rust 1.98.0 Released with Algebraic Floating-Point Methods and Buffered Integer Formatting

Rust 1.98.0 Release

The Rust team announced the release of Rust 1.98.0. This version introduces several new features and improvements for the programming language, which focuses on building reliable and efficient software. Users can update their existing Rust installations via `rustup update stable`.

Algebraic Floating-Point Methods

Floating-point types `f32` and `f64` now include "algebraic" methods for addition, subtraction, multiplication, division, and remainder. These methods enable the compiler to apply optimizations based on the algebraic properties of real numbers, even though floating-point representations have limitations. This can lead to reordering of operations, potentially allowing for broader loop-vectorization.

These algebraic methods are non-deterministic, meaning the compiler can choose different optimization strategies, but they are guaranteed not to cause undefined behavior. This feature is similar to the `-ffast-math` option found in other languages.

Buffered Integer Formatting

All primitive integer types now have a `format_into` method. This method takes a `&mut NumBuffer<Self>` parameter, which is a buffer sized to hold the decimal format of any value of that type. The method returns a formatted `&str` with a lifetime borrowed from the buffer.

The `format_into` method bypasses much of the dynamic dispatch typically involved with buffered `write!` formatting, resulting in performance gains. Benchmarks indicate that `format_into` performs similarly to the `itoa` crate, positioning it as a potential standard replacement for that dependency.