Use "with" with an R6 object

user_3635471 0
Viewed 103

I'm trying to use a construct such as "with" to allow easier (lazier ?) manipulation of an R6 object. Consider the following (minimal) example:

library(R6)
ToyClass<-R6Class(classname = "ToyClass",
                   public = list(
                     data=data.frame(x=c(1,2,3),y=c(4,5,6)),
                     color = "red",
                     symbol = 16,
                     
                     initialize=function(mult){
                       self$data$x <- self$data * mult
                     }
                   )
)

foo <- ToyClass$new(2)

I can, obviously, plot the data in my object with

plot(foo$data$x,foo$data$y,col=foo$color,pch=foo$symbol)

However, being lazy (and since I have a lot of such objects to plot, so copy/paste/edit is time-consuming and prone to errors) I'd like to do

with(foo,
     plot(data$x,data$y,col=color,pch=symbol)
)

Which does not work:

Error in plot(data$x, data$y, col = color, pch = symbol) : 
  could not find function "plot"

I understand why (foo behaves as an environment, in which plot() is not defined), but I cannot find a solution. How can I look for function names, I presume, in the global environment ?

r6 r
1 Answers

Why not just add a plot method:

ToyClass2<-R6Class(classname = "ToyClass2",
                  public = list(
                    data=data.frame(x=c(1,2,3),y=c(4,5,6)),
                    color = "red",
                    symbol = 16,
                    plot = function() {
                      plot(self$data$x,self$data$y,col=self$color,pch=self$symbol)
                    },
                    
                    initialize=function(mult){
                      self$data$x <- self$data * mult
                    }
                  )
)

foo2 <- ToyClass2$new(2)

plot(foo2)

If that's no good, another approach would be to set a parent environment as follows:

parent.env(foo) <- .GlobalEnv
with(foo, plot(data$x, data$y, col=color, pch=symbol))
parent.env(foo) <- emptyenv() # removes the parent environment

This can even be wrapped up in a function to get a more with-like feel:

with_global <- function(object, task_expr) {
  task_expr <- substitute(task_expr)
  parent.env(object) <- current_env()
  with(object, eval(task_expr))
}
with_global(foo, plot(data$x, data$y, col=color, pch=symbol))
user_56330630
Related