NEWS

---

float 0.3-4 (2026-10-02)

Bug fixes

• Arithmetic and comparison operations with an empty float32 operand no longer crash R. This includes exponentiation and combinations of vectors and matrices. Empty results and zero-dimension matrices now follow base R's behavior.

Documentation and maintenance

• Added regression tests for empty vectors, zero-dimension matrices, and uneven recycling in arithmetic and comparison operations.
• Documented the complete rcond() S4 method signature and corrected the comparison help page for R-devel checks.
• Added the missing RcppArmadillo vignette citation and updated the Intel MKL link.
• Removed a macOS startup workaround for R versions older than the package's minimum supported version (R 3.6.0).
• Stefano Cacciatore is the package maintainer.