- local: macOS, R 4.x,
R CMD checkOK (with--as-cranrecommended on poptiplex before upload) - GitHub Actions: ubuntu / macos / windows, R release
0 errors | 0 warnings | 0 notes (verify on submit box)
None known.
Not run (no known reverse dependencies with tight coupling).
Bug fix (user-visible): find_discords_rra(..., seed = 0L) could segfault when
grammar rule intervals outnumbered the series length (visit buffer sized to length(ts)
instead of interval count).
Tests: tier-B RRA pins on bundled ecg0606 (w100/w120 exact span/distance; w150/p7/a4
rejects zero-gap boundary discords).
Docs / packaging: README CRAN path; Codecov URL on app.codecov.io; roxygen examples;
NEWS.md in tarball; exclude audit HTML; DESCRIPTION summary refresh.
No algorithm changes beyond the RRA buffer fix.
1.3.1 was a performance / memory release with identical numerical results on regression tests; 1.3.0 introduced intentional behavioral alignment changes (documented in NEWS).