CODEWITHSUNDEEP
rr-package
Avraham
Avraham

Reputation: 1719

How to cleanly define a set of variables when an R package loads and clear them on unload?

I would like to have a set of colors become defined when a package loads and cleared when package detaches.

What I've come up with that seems to work is shown in the following toy example which relies on deep assignment (which I know to be evil)

.onLoad <- function(libname, pkgname) {

}

.registerColors <- function(){
  C.1 <<- c("#FF0000FF", "#80FF00FF", "#00FFFFFF", "#8000FFFF")
  C.2 <<- c("#00AAFFFF", "#0000FFFF", "#AA00FFFF", "#FF00AAFF")
}

.onUnload <- function(libpath){
}
.onAttach <- function(libname, pkgname) {
  .registerColors()
  packageStartupMessage("Welcome to XYZ")
}

.onDetach <- function(libname, pkgname) {
  rm(C.1, C.2, pos = 1)
  packageStartupMessage("Buh-bye")
}

In this case, plot(seq(1:4, col = C.1) works. Is there a better or more elegant or less potentially destructive way to implement this?

Upvotes: 6

Views: 140

Answers (2)

Avraham
Avraham

Reputation: 1719

What seems to work the way I want it to is:

.registerColors <- function(){
  assign(x = 'C.1', value = c("#FF0000FF", "#80FF00FF", "#00FFFFFF", "#8000FFFF"), pos = 2)
  assign(x = 'C.2', value = c("#00AAFFFF", "#0000FFFF", "#AA00FFFF", "#FF00AAFF"), pos = 2)
}

.onAttach <- function(libname, pkgname) {
  .registerColors()
}

Now the colors are accessible by name, they do not appear in the global environment search list (saving us from evil), and thus they don't need to be cleared manually on detach, as the package environment in which they live is automatically cleared on detach.

The benefit of this is that the users can now just access the palettes directly as variables without the need for a function call.

Upvotes: 0

Hong Ooi
Hong Ooi

Reputation: 57696

You don't really need to go to all that trouble. Just define a function that checks if the package is loaded, and returns the appropriate colours.

chooseCols <- function()
{
    if("this_package" %in% search())
        C.1
    else # use default colours
}

plot(1:4, col=chooseCols())

Upvotes: 5

Related Questions