float_next_after

Releases

1. ▸▾2.0.014 Feb 2026

   Now allows moving away from infinity. Per IEEE 754 and C standard library behavior, stepping away from infinity returns MAX/MIN. Previously core::f64:

   Breaking changes:

   • Now allows moving away from infinity. Per IEEE 754 and C standard library behavior, stepping away from infinity returns MAX/MIN. Previously core::f64::INFINITY.next_after(core::f64::NEG_INFINITY) would return core::f64::INFINITY, now it returns core::f64::MAX. Likewise, core::f64::NEG_INFINITY.next_after(core::f64::INFINITY) would return core::f64::NEG_INFINITY, now it returns core::f64::MIN.

   New Features:

   • Inlined the short_circuit_operands macro, which provides a performance improvement of about 20%.
   • Increased testing surface area for further edge cases.

   Open source →
2. ▸▾1.0.011 Dec 2022

   The NextAfter trait is now provided for the types f32 and f64 and no longer provided as a generic constrained to the type num_traits::Float.

   Breaking changes:

   • The NextAfter trait is now provided for the types f32 and f64 and no longer provided as a generic constrained to the type num_traits::Float.
   • The library is now #![no_std].

   New Features:

   • Removed dependence on the std library, the library is now #![no_std].
   • Now uses macros to create the trait for f32 and f64 directly without any use of the generic contraint using num_traits::Float.
   • CI/CD test running on git push.

   Structural changes:

   • Tests are written with a macro now so no code needs to be repeated to run tests for f32 and f64.
   • Code cleanup with:
     • 0 equality check integrated into short_circuit_operands.
     • Early return is used throughout.
     • Single check for infinite source number (it had been a separate check for positive and for negative infinity).

   Performance:

   • Removed one unneccesary creation of a stack variable.