Changelog

CmdStanModel$sample writes samples to {model}-{timestamp}-{chain}-{rand}.csv by default. The submit_model() method for stan models (?stan_submit_model) now overrides that default to drop the timestamp and random characters so that the file name is consistent across runs. (#167)

For nmbayes models, the posterior::as_draws() method (?bbr_as_draws) and read_fit_model() rename variables for compatibility with ?posterior::draws_rvars. This logic mishandled THETA names with multiple digits (e.g., THETA10). (#163)
The submit_model() method for nmbayes (?nmbayes_submit_model) did not generate initial values synchronously when a caller passed FALSE for .wait, leading to a race between creating the initial values and the subsequent model runs reading those values. (#162)
The documentation of the submit_model() methods for nmbayes (?nmbayes_submit_model) and stan (?stan_submit_model) models inherited their .overwrite description from bbr::submit_model(), which inaccurately described their handling. (#161)
install_torsten() aborts if the caller specifies a version argument that doesn’t uniquely identify an available version. It incorrectly aborted when the specified version was valid but a sub-string of another available version (e.g., “v0.91.0” could not be selected because a “v0.91.0rc2” release exists). (#159)

check_nonmem_finished.bbi_nmbayes_model() now returns FALSE if the output directory doesn’t exist, following a bbr v1.11.0 change in behavior. (#150)
The nmbayes get_data_path() method has been updated to follow changes in bbr v1.10.0.
- The first argument has been renamed from .mod to .bbi_object. (#138)
- The .check_exists argument is now supported. (#152)
- A data path is now extracted from the control stream if the model has not yet finished. (#152)

Initial support has been added for NONMEM Bayes models via the bbi_nmbayes_model model type. Compatibility note: the 0.1.0 release included a bbi_nmbayes_model model type that was marked as experimental and is not compatible with the new design. (#84, #108, #133)
New copy_model_as_nmbayes() function for creating a bbi_nmbayes_model model from a parent bbi_nonmem_model object. (#84)
New chain_paths() helper for constructing paths to files within chain sub-models. (#104)
New nm_join_bayes() and nm_join_bayes_quick() functions provide an interface for joining model output summaries to input data. (#106, #116, #118, #120, #122)
Various tailored methods for bbr generics have been added. (#84, #123, #126, #127, #128).

New install_torsten() function for downloading and installing Torsten. (#95)
The bundled bern-gq.stan model has been updated to use syntax compatible with Stan 2.33. (#99)
The signature of the submit_model() method for Stan models did not match the positional arguments of the generic. .bbi_args (unused for Stan models) has been added as the second argument. (#132)

submit_model no longer sets up a .gitignore rule for posterior CSVs of Stan models. (#92)

This initial release adds support for working with Stan models. (#81)

bbr.bayes 0.3.0