Updating embedded data, for example sysdata.rda

Question

My latest submission to CRAN got bounced back because I have assignments to the global environment which is now frowned upon.

I have an embedded data set (sysdata.rda) t

Answer 1

Why not have an foo.R file in /data which loads the data and updates it when the user calls data(foo)? This is one of the allowed options for /data, but note the following from Writing R Extensions

Note that R code should be “self-sufficient” and not make use of extra functionality provided by the package, so that the data file can also be used without having to load the package.

If you can live with that restriction then data(foo) could load the data, update it, and ensure that it is in a specific named object, which you then refer to in your functions.

Answer 2

The key is to do the assignment in an environment other than the global environment. There are two basic techniques, using local() and <<- or explicitly creating a new environment:

Working with an explicit environment is straightforward: create the environment and then assign into it like a list:

my_opts <- new.env(parent = emptyenv())
set_state <- function(value) my_opts$state <- value
get_state <- function() my_opts$state

Using local() is a little more complex, and requires some tricks with <<-

set_state <- NULL
get_state <- NULL

local({
  state <- NULL
  get_state <<- function() state
  set_state <<- function(value) state <<- value
})

For more info on how <<- works see https://github.com/hadley/devtools/wiki/environments, in the "Assignment: binding names to values" section.