Open Edward-Gillian opened 3 months ago
The rationale behind the assess_dependencies
is to capture the "true" size of the code base for a packages, while Suggests
contributes this, these are not required for normal package function and are typically needed for development. Since they are not required for intended use I do not think it is necessary to include them in the assessment as this would needlesslyo penalize packages for their Suggests and thus penalize packages with, likely, better SDLC practices..
However, packages such as knitr
and rmarkdown
that are in Suggests
are often needed by users to access vignettes.
Also, packages have dependencies listed in Suggests
that are used in the code. For example, dplyr
lists purrr
in Suggests
(https://github.com/tidyverse/dplyr/blob/be36acf9c86e5d4c3d97f97b8d3999b713123392/DESCRIPTION#L51) but the function bind-cols
uses map_lgl
in its code (https://github.com/tidyverse/dplyr/blob/be36acf9c86e5d4c3d97f97b8d3999b713123392/R/bind-cols.R#L35)
Description
add Suggests in get_package_dependencies
https://github.com/pharmaR/riskmetric/blob/master/R/assess_dependencies.R#L105
and parse_dcf_dependencies
https://github.com/pharmaR/riskmetric/blob/master/R/assess_dependencies.R#L123
Use Case
This will allow for capturing dependency packages that are used in tests, examples, and vignettes