Fundamentals of import system with {box} package

R already has library() to load packages you installed in R, then attaches the entire namespace onto the global namespace / environment, so the namespace from that package you load will be utilized globally. What’s the point of using {box} when there’s built-in library(), you may probably ask? Well, you’ll learn enough that {box} provides an actual module system in R which offers best practices and ergonomics, something that R misses. R’s module system is NOT modular — the actual modular programming that other programming languages do, and with {box}, it blows the import system from R out of the water.

{box}‘s ergonomics and most convenient features is its approach to compose and reuse / import R packages and scripts / folders’. Consequently, the module system in R is not consistent enough. But {box} does more than importing which will be discussed on the later chapters.

Discussion at Chapter 2

Within Chapter 2, the fundamentals of {box} package, the import system and semantics of box::use(), are discussed.

1 Loading packages

Unlike the traditional way to load R packages through library(), {box} is more explicit, more declarative, and more granular to import R packages, and it is more consistent to work with path directories.

# Base R 
library(dplyr)

# {box} package
box::use(dplyr, dplyr[...])

The rest will be discussed under Array of Imports with Monikers.

2 Loading scripts / folders

With {box} modules, scripts and folders can be attached and reused, as long as the namespace is defined. They are contained in an environment data structure called module. It’s good since the import system of {box} using box::use() supersedes source() and then it has 2 functions — box::file() and box::name() — that supersede setwd(). For more details, the rest will be discussed under Array of Imports with Monikers.

3 Reloading modules

If you talk about local modules, they are mostly applied for modules from scripts and folders. When the scripts and/or folders are imported as modules, the initial imports are cached. This means that the changes you make to your module’s source code won’t automatically take effect until you explicitly purge the cache through box::purge_cache() (used after unloading the modules) or reload it. The rest will be discussed under Chapter 2.1: Array of Imports with Monikers

---

Some restrictions

While {box} provides different approach to organize and import R code in an ergonomic way, there are some restrictions to be aware of. Specifically, {box} does not functionally do the following:

• Attaching meta-packages such as {tidyverse} and {tidymodels}.
• Using library() / require() inside a module when writing a module.

Base Packages

The author explicitly separates {base} package from other base R packages to be imported. Which means, when you construct a module, you don’t have to load {base} package, e.g. box::use(base[mean, sum]). The rest of base R packages, like {stats} and {utils}, are required to be called when constructing a module, unless it is just an executable script.