README.md

In machow/importr: Simple and sane modularity for R code

importr

Example

library(importr)

module <- import({
  library(MASS)
  rnorm <- function() mvrnorm(1, 0, 1)
})

# we can call rnorm
module$rnorm()
# attaching the library MASS in the import did not affect the global environment
# so we can't call it from here
exists("mvrnorm") == FALSE
# assigning mvrnorm does not screw up our import
mvrnorm = "no"
module$rnorm()

Why importr?

R has good support for modularity and isolation of code via packaging. However, the packaging system requires a myriad of files, and is extremely irritable. Many R projects need to use a set of functions across project scripts, but don't need to go to the work of packaging them. One solution is to use source to load functions you will need. However, the downside of this approach is that if you reassign a variable that is used by a function from the sourced script, it will break.

For example, suppose I have the file utils.R that contains..

f <- function() sum(1:3)

doing the following may result in unexpected results..

source('utils.R')
f()    # 6

sum <- function(x) "no"
f()    # "no"

In most cases, relying on one big global workspace for everything is dangerous. importr resolves this by ensuring two things for imported code:

1. global variables won't affect the code.
2. the code won't affect global variables (e.g. via library)

This approach was heavily inspired by Python and Javascript!

machow/importr documentation built on May 21, 2019, 10:51 a.m.