Changelog

New bf_analyze() function for exploring encoding trade-offs. Reports Pareto-optimal exponent/significand configurations per bit count, with quality metrics including RMSE, max error, underflow, overflow, and changed values at a given decimal precision. Includes a formatted print() method with summary header, type-specific output, and ready-to-paste bf_map() usage hints.
Floating-point encode/decode rewritten from string-based to arithmetic computation, improving performance and correctness.
Expanded validation throughout bf_map() and bf_flag() with more informative error messages.

bf_map() now uses template from the registry instead of requiring explicit length/width parameters.
Vignettes consolidated into a single “Getting Started” guide covering encoding design, bf_analyze(), custom protocols, and community standards.
Substantially expanded test suite covering bf_analyze, bf_export, bf_protocol, helpers, and edge cases across all core functions.
Documentation updates across all exported and internal functions.

Fixed bf_decode() flag processing and sign bit handling.
Fixed bf_export() metadata extraction to use correct field names from registry flags.

introduction of bf_flag() function, where building the flag can be found, which is moved out from bf_map().
complete overhaul of how information is stored in the registry, now it is vastly more W3C PROV compliant than before.
new protocol category for encoding categories, which are different from cases because they can be set explicitly.
the registry now contains also the levels or quantiles of the encoded data.

inclusion of the project() helper function that specifies, similarly to person(), the project metadata.
remove automatic creation of a registry when calling bf_map() for the sake of more care when users set up a new registry. This should contain author and project metadata.
instead of storing objects in the environment bf_env, they are now made available in the .GlobalEnv for easy use.

rename various things to get a more coherent and intuitive workflow: bf_test() -> bf_map(), operator -> protocol, bf_internal -> bf_pcl
new function bf_standards() to interact with github to pull, push or list community standards base on the github personal access token.
formalise the bitfield operators into a protocol, i.e., a standardised list with a dedicated .validateProtocol() function.

merge all functions into bf_test().
remove bf_standards() and bf_stds.
merge the previous test-functions into a structure that is similar to bd_stds, i.e., a list with instructions and documentation for that operation/test (bf_internal), which will be called in bf_test().
revise determination of encoding with more user-friendly interface. Now, everything is specified in bf_test() with the ... argument; documentation explains everything in much more detail.

include the .rast() function for simple handling of gridded objects.
unit tests for all functions.
introduction of a small set of standard encodings in bf_stds and the function bf_standard() to create flags from those.

major revision of the template function body for the bitfield operators
more options for bitfield operators, such as the possibility to specify an NA-value and the position in the bitfield, and a custom description
more stringent provenance handling, now provenance is reported as a combination of the Action name and the parameters that are used (following PROV terminology and logic)
revision of the bf_decode() output reporting, now with a more user-friendly return value
full documentation and functionality of all functions (except auto determination of floating-point encoding for numeric values)
document also MD5 checksum for registries

revise various functions
improve text (readme/vignette)
change naming of the functions
correct terminology, the class is not the bitfield but instead an array or registry that contains information about the bitfield. Hence, I renamed it to registry
use attributes to capture a bitflag description and name

bitfield 1.0.0