NEWS

rTRNG 4.23.1-5 (2026-02-24)

Specify C++ 17 as the required C++ standard, to avoid the R CMD check WARNING on R-devel <ciso646> is not a standard header since C++20 (#39). This is meant as an interim solution waiting for an upstream patch of the TRNG C++ library (https://github.com/rabauke/trng4/issues/34).
Fix broken link to R/Finance 2017 conference applied example (#40).
Maintenance of Continuous Integration GitHub Actions workflow (#37, #38).

Fix missing package anchors in the documentation (#33).
Fix obsolete URLs in the README (#34).
Maintenance of Continuous Integration GitHub Actions workflow (#29, #30, #31, #32).

Underlying TRNG C++ library upgraded to version 4.23.1 (#20). This is a maintenance release, mainly including: Enhanced numerical accuracy of several special mathematical functions; Re-implementation of the discard method of the lagged Fibonacci generators with logarithmic asymptotic complexity. The release also fixes the uninitialized-memory problems reported as valgrind issues in the CRAN package checks for rTRNG 4.20-1 (#16).
Unit tests for "invalid argument" errors are now robust to systems where the error class does not propagate correctly to R, such as R 3.6.3 on macOS (#15). This addresses the errors running tests in the CRAN package checks on macOS.
Continuous integration is now based on GitHub Actions and covers R-release, R-oldrel and R-devel on Ubuntu, macOS and Windows (#14), as well as running checks with valgrind (#20).

Underlying TRNG C++ library upgraded to version 4.22 (#6). This mainly includes: Fixing numerical convergence problems in the inverse of the incomplete Beta function; Removing support for C++ language standards older than C++11.
Documentation and examples updated to include enforcing the C++11 standard required by the updated TRNG C++ library.
New function check_rTRNG_linking() allows checking whether C++ code using the TRNG library can be built and linked against rTRNG (#10).
Package vignettes now use check_rTRNG_linking() to build C++ code using TRNG only on systems supporting it. This addresses the vignette re-building issues in the CRAN package checks on macOS (and Solaris), and is broadly documented in the package, including README and vignettes (#10).

Embed TRNG 4.20 C++ library.
Expose parallel random number generators and distributions from the library to R.
Make TRNG accessible to other R projects' C++ code, via package dependency or Rcpp::sourceCpp.
Vignettes and examples covering basic and advanced usage with both R and C++.