3

Is it possible to move variables that reside in the global environment into a separate environment to declutter the global namespace? I understand how to create variables in a separate environment (with(env, ...)) but is there an efficient way to move them after creation in the global environment. I suppose it would be possible to copy them into a separate environment and then remove them from the global environment, but wanted to know if there was a more efficient manner.

Kevin Ushey
  • 20,530
  • 5
  • 56
  • 88
kamiks
  • 184
  • 7

4 Answers4

2

Not sure if this is a good idea but you can attach them to the search path. Starting with a fresh vanilla R session try this.

a <- 1
b <- 2
attach(as.list(.GlobalEnv), name = "myenv")

rm(a, b)
ls("myenv")
ls()
a
b
G. Grothendieck
  • 254,981
  • 17
  • 203
  • 341
2

Maybe:

library(purrr)

a <- 111
b <- 'hello'

my_envir <- new.env()

names(.GlobalEnv) %>% 
    walk(~ assign(.x, get(.x), envir = my_envir))

eapply(my_envir, function(x) x)
#> $my_envir
#> <environment: 0x7fed59e56dc8>
#> 
#> $a
#> [1] 111
#> 
#> $b
#> [1] "hello"

Or

library(purrr)
a <- 111
b <- 'hello'
my_envir <- new.env()

eapply(.GlobalEnv, function(x) x) %>% 
    discard(is.environment) %>% 
    {walk2(., names(.), ~{
               assign(.y, .x, envir = my_envir)
               exec('rm', .y, envir = .GlobalEnv)}
)}


eapply(my_envir, function(x) x)
#> $a
#> [1] 111
#> 
#> $b
#> [1] "hello"

Created on 2021-12-31 by the reprex package (v2.0.1)

jpdugo17
  • 6,816
  • 2
  • 11
  • 23
  • 2
    thanks, this worked exactly as expected! `lapply()` also works in place of `purrr::walk()` with some slight modifications. – kamiks Dec 31 '21 at 22:56
2

Using rlang

library(rlang)
a <- 111
b <- "hello"
my_envir <- env(!!! as.list(.GlobalEnv))

-checking

> ls(my_envir)
[1] "a" "b"
> my_envir$a
[1] 111
akrun
  • 874,273
  • 37
  • 540
  • 662
1

You may use multiple lines in the with.

e1 <- new.env()
e2 <- new.env()

with(e1, {
  k <- l <- m <- 0L
  x <- 1
  fo <- y ~ x
  fun <- function(x) x^2
  })

The objects are created in e1,

ls(e1)
# [1] "fo"  "fun" "k"   "l"   "m"   "x"

e2 stays empty,

ls(e2)
# character(0)

and in .GlobalEnv only the environments exist so far.

ls(.GlobalEnv)
# [1] "e1" "e2"

To work with objects, also use with or $.

with(e1, fun(2))
# [1] 4

e1$fun(2)
# [1] 4
jay.sf
  • 60,139
  • 8
  • 53
  • 110
  • `with()` doesn't always work as expected. As an example, I created a new environment and set up a multi-line `with(env, {})` block in order to source multiple interrelated scripts into (some of which had their own environments); however, some of the sourced scripts still returned variables into the global environment despite being clearly within the {} block. – kamiks Dec 31 '21 at 21:47
  • @kamiks Instead of narrating, you could give a (not) working example and possibly the solution you found in an answer. – jay.sf Dec 31 '21 at 21:53
  • My intention isn't to be ambiguous, it's that my use case is too long to fit into a comment block: I literally had an R project, where one script sourced a bunch of other scripts and I tried to place all of the `source()` calls within a `with(env,{})` block only to find certain sourced variables appearing in the global environment instead of the specified environment. I have yet to find the reason or solution for why this is occurring. – kamiks Dec 31 '21 at 22:23