Active bindings in R are much like properties in other languages: They look like a variable, but querying or setting the value triggers a function call. They can be created in R via makeActiveBinding(), but with this API the function used to compute or change the value of a binding cannot take additional arguments. The bindr package faciliates the creation of active bindings that are linked to a function that receives the binding name, and an arbitrary number of additional arguments.


You can install bindr from GitHub with:

# install.packages("devtools")

Getting started

For illustration, the append_random() function is used. This function appends a separator (a dash by default) and a random letter to its input, and talks about it, too.

append_random <- function(x, sep = "-") {
  message("Evaluating append_random(sep = ", deparse(sep), ")")
  paste(x, sample(letters, 1), sep = sep)

#> Evaluating append_random(sep = "-")
#> [1] "a-k"
append_random("X", sep = "+")
#> Evaluating append_random(sep = "+")
#> [1] "X+u"

In this example, we create an environment that contains bindings for all lowercase letters, which are evaluated with append_random(). As a result, a dash and a random letter are appended to the name of the binding:

env <- create_env(letters, append_random)
#>  [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q"
#> [18] "r" "s" "t" "u" "v" "w" "x" "y" "z"
#> Evaluating append_random(sep = "-")
#> [1] "a-p"
#> Evaluating append_random(sep = "-")
#> [1] "a-j"
#> Evaluating append_random(sep = "-")
#> [1] "a-b"
#> Evaluating append_random(sep = "-")
#> [1] "c-b"

Bindings can also be added to existing environments:

populate_env(env, LETTERS, append_random, "+")
#> Evaluating append_random(sep = "-")
#> [1] "a-z"
#> Evaluating append_random(sep = "+")
#> [1] "Z+j"

Further properties

Both named and unnamed arguments are supported:

create_env("binding", paste, "value", sep = "-")$binding
#> [1] "binding-value"

A parent environment can be specified for creation:

env2 <- create_env("a", identity, .enclos = env)
#> a
get("b", env2)
#> Evaluating append_random(sep = "-")
#> [1] "b-m"

The bindings by default have access to the calling environment:

create_local_env <- function(names) {
  paste_with_dash <- function(...) paste(..., sep = "-")
  binder <- function(name, append) paste_with_dash(name, append)
  create_env(names, binder, append = "appending")

env3 <- create_local_env("a")
#> [1] "a-appending"

All bindings are read-only:

env3$a <- NA
#> Error: Binding is read-only.
env3$a <- NULL
#> Error: Binding is read-only.

Existing variables or bindings are not overwritten:

env4 <- as.environment(list(a = 5))
populate_env(env4, quote(b), identity)
#> [1] "a" "b"
populate_env(env4, letters, identity)
#> Error in populate_env(env4, letters, identity): Not creating bindings for existing variables: a, b

Active bindings and C++

Active bindings must be R functions. To interface with C++ code, one must bind against an exported Rcpp function, possibly with rng = false if performance matters. The bindrcpp package uses bindr to provide an easy-to-use C++ interface for parametrized active bindings, and is the recommended way to interface with C++ code. In the remainder of this section, an alternative using an exported C++ function is shown.

The following C++ module exports a function change_case(to_upper = FALSE), which is bound against in R code later.

#include <Rcpp.h>

#include <algorithm>
#include <string>

using namespace Rcpp;

// [[Rcpp::export(rng = FALSE)]]
SEXP change_case(Symbol name, bool to_upper = false) {
  std::string name_string = name.c_str();
  std::transform(name_string.begin(), name_string.end(),
                 name_string.begin(), to_upper ? ::toupper : ::tolower);
  return CharacterVector(name_string);

Binding from R:

env <- create_env("__ToLower__"), change_case)
populate_env(env,"__tOuPPER__"), change_case, TRUE)
#> [1] "__ToLower__" "__tOuPPER__"
#> [1] "__tolower__"
get("__tOuPPER__", env)
#> [1] "__TOUPPER__"