daniel1noble / BACE / 25336882118

91%

push

github

Coverage push: 80.6% -> 87.76% (1214 tests)

Five new test files target previously-uncovered orchestration paths:

- test-verbose_orchestration.R: bace()/bace_imp()/bace_final_imp()
  verbose=TRUE cat() lines + nitt_cat_mult + ovr_categorical=FALSE.
- test-bace_orchestration_extras.R: phylo_signal=TRUE short-circuit,
  plot=TRUE branches, sample_size pooling messages, .geweke_test
  internal helper unit tests, categorical convergence summary.
- test-convergence_extras.R: use_all_data=TRUE on factor/threshold
  branch, energy/wasserstein/all methods on categorical, plot helper
  error paths.
- test-simulate_extras.R: random slopes (rr_form), multinomial/
  threshold response variants, mixed predictor types, custom
  missingness.
- test-species_true_dualRE.R: bace_imp/bace_final_imp/bace with
  species=TRUE dual phylo+species RE on replicated data, plus
  phylo_signal_summary categorical/threshold dual-RE.

Source fix in bace.R: `if (plot)` calls now guard with
`inherits(converge, "bace_convergence")` and a non-null
summary_stats check. Previously, when runs<3, assess_convergence()
warns and returns a plain list; downstream `plot(converge)` then
dispatched to plot.default and crashed with `'x' is a list, but
does not have components 'x' and 'y'`. Regression test included.

R CMD check: 0E/0W/1N (timestamp note, environmental).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

5 of 6 new or added lines in 1 file covered. (83.33%)

3227 of 3677 relevant lines covered (87.76%)

83.53 hits per line