NEWS

TH 1.0.0 (2026-08-06)

Replaced substring-based NOTE acceptance with a strict line whitelist: only the optional maintainer line and the exact New submission marker are admissible.
Added internal clean/composite NOTE regression cases so an incoming NOTE containing any additional finding blocks release.
Strengthened the two-pass roxygen2 gate to compare the complete man//NAMESPACE file set before comparing hashes.
Refreshed the static audit, validation record, and SHA-256 manifest.

Stabilized the extreme upper-tail Z regression test by comparing the p-value with the tail probability evaluated at the statistic actually returned by the function.
Corrected the paired t-test educational hypotheses to use the population mean of the differences, mu[D], rather than the sample mean bar(D).
Added a two-pass roxygen2 idempotence gate and explicit synchronization checks for the chi-square help pages.
Strengthened review of the single admissible New submission NOTE by printing and saving its complete text.
Corrected the static-validation manifest count and refreshed the SHA-256 manifest.

Builds the final source archive with manual = TRUE so the reference manual is included in the CRAN release workflow.
Allows only the single informational New submission NOTE while blocking every error, warning, and unexpected NOTE.
Replaced chi-square htest alternatives with descriptions of the scientific hypotheses.
Replaced tests of internal helpers with tests through exported functions.
Added a programmatic package-name conflict check against current and archived CRAN packages and current Bioconductor repositories.
Consolidated the release history under a single version heading.

Added a strict rcmdcheck release gate that blocks on any error, warning, or unexpected note.
Regenerates man/ from canonical roxygen2 comments before every release build.
Regenerates README.md from README.Rmd and rejects stale YAML front matter.
Added PDF-device smoke tests for the plotting branches of all ten exported functions.
Made spelling and URL checks release-blocking and added a reviewed technical word list.
Preserved the statistical and interface corrections from the previous candidate.

Corrected recursive and external-object argument defaults.
Added validation for significance levels, alternatives, samples, variances, sample sizes, paired lengths, frequencies, and degrees of freedom.
Corrected the directional F test so the variance ratio remains s2X / s2Y.
Restricted goodness-of-fit, independence, and homogeneity chi-square procedures to the standard upper-tail omnibus test.
Corrected the FO/FE pathway in tq2a().
Retained fractional Welch-Satterthwaite degrees of freedom in tt2d().
Replaced statements that treated non-rejection as proof of equality or independence.
Restored graphical parameters after every function call.
Added structured invisible htest-compatible return objects.
Added executable help examples, vignette examples, and automated testthat regression tests.
Removed duplicated installed static HTML and manual test scripts from inst/.

Made raw-data and summary-statistics input modes mutually exclusive.
Added plot = TRUE to every exported function; tests and most vignette examples use plot = FALSE.
Replaced subtraction-based upper-tail probabilities with direct tail calculations.
Corrected both lower-panel F-distribution polygons in tf2().
Displayed both lower and upper critical values for two-sided F tests.
Preserved and aligned named categories in tq2a().
Required r = 0 in independence and homogeneity tests while retaining the argument for backward compatibility.
Re-established roxygen comments as the documentation source of truth.
Removed the obsolete installed reference to inst/htmlhelp/.
Expanded automated tests for summary modes, mixed-input errors, alternatives, extreme tails, category alignment, plotting, and F critical values.