R’s options are a useful tool for managing global settings. options
aims to make them easy to configure and use, reduce boilerplate code and handle more involved options handling so you don’t have to.
While writing your package, we start by defining which options our package might use.
Define your options using the define_options()
shorthand. Interlace descriptions and default values to define multiple options at once.
::define_options(
options"This is an example of how a package author would document their internally
used options. This option could make the package default to executing
quietly.",
quiet = TRUE,
"Multiple options can be defined, providing default values if a global option
or environment variable isn't set.",
second_example = FALSE,
"Default values are lazily evaluated, so you are free to use package functions
without worrying about build-time evaluation order",
lazy_example = fn_not_defined_until_later()
)
When you want more control, you can use define_option()
to declare all aspects of how your option behaves.
::define_option(
optionsoption = "concrete_example",
default = TRUE,
desc = paste0(
"Or, if you prefer a more concrete constructor you can define each option ",
"explicitly."
),option_name = "mypackage_concrete", # define custom option names
envvar_name = "MYPACKAGE_CONCRETE", # and custom environment variable names
envvar_fn = envvar_is_true() # and use helpers to handle envvar parsing
)
As long as the options have been created as shown above, documenting your options is as easy as adding this small roxygen stub within your package.
#' @eval options::as_roxygen_docs()
NULL
Which produces a ?mypackage::options
help page. Moreover, any options
page will show up in an index when using ?options
to search for documentation, making it easier to discover which packages have options for you to use.
mypackage Options
Description:
Internally used, package-specific options. All options will
prioritize R options() values, and fall back to environment
variables if undefined. If neither the option nor the environment
variable is set, a default value is used.
Options:
quiet
This is an example of how a package author would document their
internally used options. This option could make the package default to
executing quietly.
default:
TRUE
option: mypackage.quiet
envvar: R_MYPACKAGE_QUIET (raw)
...
When your options are used as default values to parameters, you can use the option documentation to populate your function parameter docs.
This is made simple when all of your parameters share the same names as your options.
#' @eval options::as_params()
#' @name options_params
#'
NULL
#' Count to Three
#'
#' @inheritParams option_params
#'
function(quiet = opt("quiet")) {
count_to_three <-for (i in 1:3) if (!quiet) cat(i, "\n")
}
In situations where you have identically named parameters where you don’t want to inherit the option documentation, you can provide their names to as_params
to use just a subset of options. You can also reassign documentation for an option to a parameter of a different name.
#' Hello World!
#'
#' @eval options::as_params("silent" = "quiet")
#'
function(who, silent = opt("quiet")) {
hello <-cat(paste0("Hello, ", who, "!"), "\n")
}
When using define_option
you can set the option_name
and envvar_name
that will be used directly.
But it can be tedious and typo-prone to write these out for each and every option. Instead, you might consider providing a function that sets the default format for your option and environment variable names.
For this, you can use set_option_name_fn
and set_envvar_name_fn
, which each accept a function as an argument. This function accepts two arguments, a package name and internal option name, which it uses to produce and return the corresponding global option name or environment variable name.
::set_option_name_fn(function(package, name) {
optionstolower(paste0(package, ".", name))
})
::set_envvar_name_fn(function(package, name) {
optionsgsub("[^A-Z0-9]", "_", toupper(paste0(package, "_", name)))
})