Attach a short human-readable description to a fit function. It is copied
into the recipe that apply_fit and apply_fits
record in metadata(se)$provenance, so a stored object says not only
which function produced a fit but also what that function does.
Arguments
- fn
the fit function to describe.
- model
string; name of the model, e.g.
"3-parameter log-logistic".- description
string; what the function computes, in a sentence or two.
- reference
string; citation of the method, ideally with a DOI.
- env
optional function without arguments returning a named list of strings that describe the external runtime the fit depends on, e.g. the version of a Python package called through reticulate. It is called each time a fit is recorded.