list_benchmark_releases() lists the
available releases and the release argument of the download
and retrieve functions selects one. See the new “Releases” article.safe_rbind() bug fixed below; all other files are
identical, and 2026.1 remains available.method_setting was mislabelled default have
corrected copies (the originals are kept).method = "RMA"). Downloads are verified against checksums
and cached; files shared by several releases are downloaded only once,
and prune_benchmark_archives() frees disk space.benchmark_conditions(),
benchmark_catalog(),
list_benchmark_releases(),
list_benchmark_resources(),
verify_benchmark_resources(),
download_dgm_metadata() and
prune_benchmark_archives().retrieve_dgm_dataset(),
retrieve_dgm_results() and
retrieve_dgm_measures() read your own unpublished files
with source = "local".compute_measures(), compare_measures() and
their single-measure versions gain results_source,
replacement_source and release: your method’s
results are read locally by default, and
replacement_source = "release" takes the results of the
replacement methods from the published benchmark..github/RELEASING.md in the
GitHub repository).NA in conditions where
the replaced method had no valid runs, because the internal
safe_rbind() dropped the replacement rows.compare_single_measure(),
compare_measures()) kept only the last condition of every
method pair; they now give one row per method pair and condition.
Recompute existing pairwise files with
overwrite = TRUE.compute_single_measure() creates the DGM’s
measures folder when it does not exist.. or / are no longer merged.fit_limit argument to run_method()
that aborts a fit exceeding the given number of minutes and returns a
standard failure result with
note = "time limit exceeded with <fit_limit> minutes".
The fit runs in a reused background R process so that methods sitting in
compiled sampling code (RoBMA, RTMA, MMPH) can be stopped as well.measure() function to list available performance
measures (renamed from measures()).measure_mcse() function to list available
performance measure MCSE functions.measure() and
measure_mcse() to retrieve specific functions (e.g.,
measure("bias"), measure_mcse("bias")).method() and dgm() to list
available methods and DGMs when called without arguments.method() and dgm() to return the
corresponding function when called with a single argument (e.g.,
method("RMA")).measure(), measure_mcse(),
method(), and dgm() now dynamically retrieve
available options using methods().Initial CRAN submission.