3  Packages


Functional packages require three things:

  1. A DESCRIPTION file with the following fields:
Package
Version
License
Description
Title
Author
Maintainer
  1. An Rproj file with the following fields:
BuildType: Package
PackageUseDevtools: Yes
PackageInstallArgs: --no-multiarch --with-keep.source

  Access these settings via Tools > Project Options > Build Tools

  1. devtools installed
install.packages('devtools')
library(devtools)

This chapter covers the ‘necessary ingredients’ of R packages. We’ll start by exploring the differences between Shiny projects, R packages, and Shiny app-packages. Then, we’ll dive into the requirements for 1) creating new Shiny app-packages, and 2) converting Shiny app projects into Shiny app-packages.

I’ve created the shinypak R package In an effort to make each section accessible and easy to follow:

Install shinypak using pak (or remotes):

# install.packages('pak')
pak::pak('mjfrigaard/shinypak')

Review the chapters in each section:

library(shinypak)
list_apps(regex = '^03')
## # A tibble: 3 × 2
##   branch              last_updated       
##   <chr>               <dttm>             
## 1 03.1_description    2024-01-17 12:40:11
## 2 03.2_rproj          2024-01-17 12:41:14
## 3 03.3_create-package 2024-01-17 12:42:45

Launch the app:

launch(app = "03.3_create-package")

Download the app:

get_app(app = "03.3_create-package")

Before we start, we should establish some operational definitions of what is meant by the terms ‘project’, ‘package’, and ‘app-package’.

I’ll differentiate Shiny app projects, R packages, and Shiny app-packages using icons

There are multiple reasons for deciding to develop a Shiny app-package, but if you’re reading this book, you’ve likely found yourself in one of two circumstances:

  1. You want to develop a new app, but want it structured as an R package.

  2. You’ve already developed a Shiny app project, but now you need to convert it to a Shiny app-packages

The previous chapter covered a few practices to adopt during Shiny development that improve the extensibility of your Shiny app project and make it easier to convert into a Shiny app-package.1

R Packages & Posit Workbench

Posit Workbench (formerly RStudio) is a popular integrated development environment (IDE) that streamlines many R package development tasks. I’ve purposely connected Posit Workbench to the definitions above for R package and Shiny app-packages. Specifically, the package development tools provided in the Build pane and the devtools package.

However, developing R packages in Posit Workbench (or using .Rproj files) is not required. There are alternative package development tools and processes outside of Posit Workbench, and many developers employ these setups.

Package development outside Posit Workbench would look almost identical to development in the IDE:

  1. Create the package structure (R/ folder for .R scripts, man/ folder for documentation, data/ folder for datasets, etc.)

  2. Add DESCRIPTION and NAMESPACE files, etc.

  3. Ensure the package can be installed and loaded into an R session.

It’s also possible to use many of the development workflow functions we’ll cover here outside of the IDE (roxygen2::roxygenize(), devtools::check(), devtools::install(), etc.).

3.1 R packages

R packages can be easily shared, reused, and reproduced because they all have a familiar structure, and each folder and file plays an essential role in extending R’s capabilities.2

Below is a folder tree with some of the typical files and folders found in R packages:

<R package>/
    ├── DESCRIPTION
    ├── <R package>.Rproj
    ├── LICENSE 
    ├── LICENSE.md 
    ├── NAMESPACE 
    ├── NEWS.md
    ├── README.Rmd
    ├── README.md
    ├── renv.lock
    ├── R/
    ├── man/
    ├── tests/
    ├── data/ 
    ├── data-raw/ 
    ├── vignettes/ 
    ├── inst/ 
    └── renv/
  

This chapter will cover the minimum requirements for an R package, so you can handle both creating new Shiny app-packages and converting existing Shiny projects into Shiny app-packages.

3.1.1 Functional R packages

If you’ve done some research on R packages, you’ve probably encountered one (or both) of the following statements,

‘RStudio [Posit Workbench] and devtools consider any directory containing DESCRIPTION to be a package’ - R Packages, 2ed (Chapter 9, DESCRIPTION)

‘all a project needs to be a package is a directory of R/ files and a DESCRIPTION file.’ - Mastering Shiny (Chapter 20, Packages)

The moviesApp directory has:

moviesApp/
├── DESCRIPTION
├── moviesApp.Rproj
├── R/
   ├── mod_scatter_display.R
   ├── mod_var_input.R
   └── utils.R
├── README.md
├── app.R
├── movies.RData
└── www/
    └── shiny.png

4 directories, 9 files

1
DESCRIPTION file
2
.Rproj file
3
R/ folder with .R files

So, is moviesApp a functioning R package?

3.1.1.1 Can we load it with devtools?

Package development kicks off with the load_all() function from devtools, which is similar to calling library() (we’ll cover this function extensively in the devtools chapter).

However, when we attempt to load the code in the R/ folder with load_all(), we see the following error:

install.packages("devtools")
library(devtools)
devtools::load_all()
(a) Load All Error
Figure 3.1: devtools is looking for the Package field in our DESCRIPTION file

3.1.1.2 Can we use the Build pane?

If you’re using the RStudio IDE, then functioning R packages also have access to the Build pane, which allows us to quickly load, install, and test code.

When we open moviesApp, the Build pane is not displayed in the IDE:

(a) moviesApp IDE
Figure 3.2: The Build pane is normally between Connections and Git.

As we’ve just learned, the presence of the DESCRIPTION file and an R/ folder are insufficient to turn a Shiny app project into a functioning R package that can be installed and loaded into an R session and has access to the Build pane in the IDE.

A project needs a DESCRIPTION file with specific fields, a directory of R/ files, and a properly configured .Rproj file to be a functioning R package.

Let’s see how each of these requirements work together to convert the contents of moviesApp from a Shiny project into a Shiny app-package.

3.1.2 DESCRIPTION

Launch app with the shinypak package:

launch('03.1_description')

The official R documentation3 lists the following required fields for the DESCRIPTION in R packages:

The ‘Package’, ‘Version’, ‘License’, ‘Description’, ‘Title’, ‘Author’, and ‘Maintainer’ fields are mandatory.

Below is an example DESCRIPTION file for moviesApp with the mandatory fields:4

Package: moviesApp
Title: movies app
Version: 0.0.0.9000
Author: John Smith [aut, cre]
Maintainer: John Smith <John.Smith@email.io>
Description: A movie-review Shiny application.
License: GPL-3

Note that the Author and Maintainer fields require additional information beyond first and last name (i.e., John Smith). These can be created with the utils::person() function:

# Author
utils::person(
  given = "John", 
  family = "Smith", 
  role = c("aut", "cre"))
## [1] "John Smith [aut, cre]"
# Maintainer
utils::person(
  given = "John", 
  family = "Smith", 
  email = "John.Smith@email.io")
## [1] "John Smith <John.Smith@email.io>"

After adding the mandatory fields to the DESCRIPTION file, devtools::load_all() runs without the previous error, but the IDE still doesn’t display the Build pane:

(a) moviesApp IDE
Figure 3.3: Where is the Build pane?

3.2 R Projects

.Rproj files are plain text files with various settings for the IDE. We were able to run devtools::load_all() above without the presence of the Build pane because the IDE displays the Build pane after reading the fields in the .Rproj file.

The quickest way to access the fields in the .Rproj file is under Tools > Project Options….

3.2.1 R Project options

Tools > Project Options… provide access to the project-level options. For example, fields 2 - 4 are available under General, 5 - 8 affect the Code options, and the final two fields deal with Sweave.

(a) moviesApp.Rproj fields
Figure 3.4: Field settings from moviesApp.Rproj file

The default settings were created when we selected the Shiny app project from the New Project Wizard in the last chapter.

3.2.2 Build tools

I’ve placed the .Rproj file from moviesApp with the example .Rproj file from R Packages, (2ed) side-by-side so you can compare them below:5

(a) .Rproj files
Figure 3.5: Comparison of moviesApp.Rproj file and .Rproj file in R Packages, 2ed

I’ve circled the fields in the .Rproj file that illustrate it’s configured to work with an R package. Note that in moviesApp, the Project build tools are initially set to (None) under Build Tools:

(a) moviesApp.Rproj build tools
Figure 3.6: Build tool settings in moviesApp.Rproj file

Changing the Project build tools option to Package will set the default Build Tools options:6

(a) Default package build tools
Figure 3.7: Default build tool settings

The links between the Build Tools options and fields in moviesApp.Rproj are in the figure below:

(a) moviesApp.Rproj build tool fields
Figure 3.8: Default build tool settings in moviesApp.Rproj file
  • BuildType: Package tells the IDE moviesApp is an R package and triggers the Build pane.

  • PackageUseDevtools: Yes links the options in the Build pane to the devtools package.

  • The PackageInstallArgs are complicated, but I’ve included some information about them in the callout block below (and you can read more in the official R documentation)

  • The fourth option (PackageRoxygenize) is available under Generate documentation with Roxygen > Use roxygen to generate:

    • These options affect the documentation in an R package. To match the example from R Packages, 2ed above, make sure Rd files, Collate field, and NAMESPACE file are selected and click OK.
(a) PackageRoxygenize in .Rproj file
Figure 3.9: roxygen2 build settings

After clicking OK, the IDE will automatically reboot, and the additional fields will be added to the bottom of the moviesApp.Rproj:

Launch app with the shinypak package:

launch('03.2_rproj')
Version: 1.0

RestoreWorkspace: Default
SaveWorkspace: Default
AlwaysSaveHistory: Default

EnableCodeIndexing: Yes
UseSpacesForTab: Yes
NumSpacesForTab: 2
Encoding: UTF-8

RnwWeave: Sweave
LaTeX: XeLaTeX

BuildType: Package
PackageUseDevtools: Yes
PackageInstallArgs: --no-multiarch --with-keep.source
PackageRoxygenize: rd,collate,namespace
                                            
  • --no-multiarch: refers to the option for the package installer to only compile the package for the architecture of the current R session. By default, R tries to compile packages for 32-bit and 64-bit architectures if running in a 64-bit R session on Windows. This flag can help avoid problems if a package can only be compiled on one architecture. Read more here.

  • --with-keep.source: In R, when a function is created, its body can be stored in two ways: 1) as a parsed but unevaluated expression and 2) as a character string containing the function’s source code. By default, only the parsed expression is kept. If –with-keep.source is specified, R will also keep the source code as a character string, which can be helpful for debugging and tools that analyze or modify source code. Read more here.

3.2.3 The Build pane

When the new session starts, the new project-level options activate the Build pane in the IDE, and I can check the R package functionality by loading the code with Build > Load All

(a) Load the code in the R/ folder
Figure 3.10: Identical to running devtools::load_all()

I should see the following in the Console:

ℹ Loading moviesApp

There you have it–moviesApp is a functional app-package!

3.2.4 Functional app-packages

In a functional app-package :

  1. The DESCRIPTION file contains the seven mandatory fields7, making running the necessary devtools functions possible.
  1. The .Rproj file contains the three package configuration fields8, which makes the Build pane accessible and functional.

Think of the two items above as a two-part process: the DESCRIPTION requires specific fields, and the IDE requires .Rproj fields to trigger the Build pane.

(a) Shiny app-package (with DESCRIPTION and Build pane)
Figure 3.11: Fully functional Shiny app-package

3.3 Creating app-packages

The Posit documentation lists the following way(s) to create R packages:9

  1. Call usethis::create_package().

  2. In Posit Workbench, do File > New Project > New Directory > R Package. This ultimately calls usethis::create_package(), so really there’s just one way.

See the 03.3_create-package branch of moviesApp.

create_package() is a great option if you’re looking for a way to quickly create or convert your Shiny project into a Shiny app-package. In the following sections I’ll cover some suggestions for using create_package().10

3.3.1 New Shiny app-packages

If you haven’t written any code and want to create a new Shiny app-package, create_package() is the quickest way to get started.

First, install devtools:

install.packages("devtools")
library(devtools)

devtools automatically loads usethis

Loading required package: usethis

Assuming your current working directory is where you want your new Shiny app-package, call usethis::create_package() with the path argument set to getwd()

usethis::create_package(path = getwd())

This call launches a series of actions:

  • First, the active project is set to whatever was given to the path argument.

    ✔ Setting active project to 'path/to/newApp'
  • The R/ folder and DESCRIPTION/ file are created:

    ✔ Creating 'R/'
    ✔ Writing 'DESCRIPTION'
  • The NAMESPACE and .Rproj files are created:

    ✔ Writing 'NAMESPACE'
    ✔ Writing 'newApp.Rproj'
  • The .Rproj is added to the .Rbuildignore file, the .Rproj.user folder is added to the .gitignore and .Rbuildignore files.

    ✔ Adding '^newApp\\.Rproj$' to '.Rbuildignore'
    ✔ Adding '.Rproj.user' to '.gitignore'
    ✔ Adding '^\\.Rproj\\.user$' to '.Rbuildignore'
  • A new session is opened from the new .Rproj file:

    ✔ Opening 'path/to/newApp/' in new RStudio session

When the new session opens, newApp has the following contents:

newApp/
  ├── .Rbuildignore
  ├── .Rproj.user/
  ├── .gitignore
  ├── DESCRIPTION
  ├── NAMESPACE
  ├── R/
  └── newApp.Rproj

You’re now free to develop newApp. Store and document any .R files in the R/ folder, edit the DESCRIPTION file with details about the application, read through R Packages and Mastering Shiny, and add the Shiny code to complete your Shiny app-package.

3.3.2 Converting Shiny projects

If you already have a Shiny app project that needs to be converted into a Shiny app-package, you can also use create_package(path = getwd()) in your root folder, but I recommend using the following arguments:

3.3.2.1 DESCRIPTION arguments

  • fields: these are arguments passed to usethis::use_description(). If the fields argument is empty, a boilerplate DESCRIPTION file is created (similar to this one).

    • These boilerplate fields in the DESCRIPTION work, but I’ve found some of the fields are unnecessary (i.e., I’ve never needed Authors@R: or ORCID) and inevitably require revision, so I’d prefer to handle this during the creation process (and remove the risk of forgetting to change it later).

    • All fields should be passed in a list() as field = 'value' pairs.

    usethis::use_description(
        list(Package = 'moviesApp',
             Version = '0.0.0.9000',
             Title = 'movies app',
             Description = 'A movie-review Shiny application.',
             "Authors@R" = NULL,
             Author = utils::person(
                given = "John", 
                family = "Smith", 
                role = c("aut", "cre")),
              Maintainer = utils::person(
                given = "John", 
                family = "Smith",
                email = "John.Smith@email.io"),
              License = "GPL-3"))
    • A few fields require specially formatted values (see the utils::person() examples below).

      utils::person("John", "Smith", 
                    email = "John.Smith@email.io", 
                    role = c("aut", "cre"))
      [1] "John Smith <John.Smith@email.io> [aut, cre]"
  • Two additional arguments from use_description() are passed to create_package():

    • check_name: verifies your Shiny app-package name is valid for CRAN, so we can set this to FALSE (unless you’re planning on submitting to CRAN)

    • roxygen2: is TRUE by default and adds the fields required to use roxygen2 (which I won’t cover here because we’ll cover documentation in-depth in a future chapter).

When converting your existing Shiny app project into a Shiny app-package with usethis::create_package(), don’t use '.' in the path argument:

usethis::create_package('.')

This will return the following warning about creating nested projects, and ask if you want to proceed anyway:

New project 'moviesApp' is nested inside an existing project
'./', which is rarely a good idea. If this is unexpected, 
the here package has a function,
`here::dr_here()` that reveals why './' is regarded as a
project.
  
Do you want to create anyway?

1: Yes
2: No way
3: Not now

We can avoid this warning altogether by passing getwd() to the path argument, so I recommend cancelling the project creation:

Selection: 2
Error: Cancelling project creation.

3.3.2.2 IDE arguments

  • rstudio: adds the necessary Build Tools fields in the .Rproj file (leave as TRUE)

  • open: can be set to FALSE because we don’t need RStudio/Posit Workbench to open in a new session

usethis::create_package(
  path = getwd(),
  fields = list(Package = 'moviesApp',
         Version = '0.0.0.9000',
         Title = 'movies app',
         Description = 'A movie-review Shiny application.',
         "Authors@R" = NULL,
         Author = utils::person(
            given = "John", 
            family = "Smith", 
            email = "John.Smith@email.io", 
            role = c("aut", "cre")),
          Maintainer = utils::person(
            given = "John", 
            family = "Smith",
            email = "John.Smith@email.io"),
          License = "GPL-3"),
  roxygen = TRUE,
  check_name = FALSE, 
  rstudio = TRUE,
  open = FALSE)

After running usethis::create_package() with the arguments above, the IDE will present us with a few prompts to confirm:

Overwrite pre-existing file 'DESCRIPTION'?
Overwrite pre-existing file 'moviesApp.Rproj'?

The Shiny app-package structure is below:

├── DESCRIPTION
├── NAMESPACE
├── R
   ├── mod_scatter_display.R
   ├── mod_var_input.R
   └── utils.R
├── README.md
├── app.R
├── movies.RData
├── moviesApp.Rproj
└── www
    └── Shiny.png

3 directories, 10 files

The DESCRIPTION file (shown below) has a few additional fields (Encoding, Roxygen, and RoxygenNote) we didn’t include when we converted moviesApp above, but we will cover these in the upcoming chapters.

Package: moviesApp
Title: movies app
Version: 0.0.0.9000
Author: John Smith <John.Smith@email.io> [aut, cre]
Maintainer: John Smith <John.Smith@email.io>
Description: A movie-review Shiny application.
License: GPL-3
Encoding: UTF-8
Roxygen: list(markdown = TRUE)
RoxygenNote: 7.2.3

Launch app with the shinypak package:

launch('03.3_create-package')

Recap

This chapter has covered the mandatory fields in the DESCRIPTION file and the R package configuration fields in .Rproj. We also covered creating and converting Shiny app projects using the usethis::create_package() function.

Recap: Packages!
  • The DESCRIPTION file ultimately controls whether you have a functional Shiny app-package.

    • The mandatory fields are Package, Version, License, Description, Title, Author, and Maintainer.
  • usethis::create_package() can be used to create a new R package and to convert an existing Shiny project into a Shiny app-package.

  • The IDE reads Posit Workbench (.Rproj) files and determines R session settings at the project-level (i.e., working directory, workspace, history, code formatting, etc.)

    • Package development settings can be accessed via Project Options > Build Tools.

In the next chapter, I’ll cover how you can quickly Load, Document, and Install your package!

Please open an issue on GitHub


  1. Shiny app project features are covered in Section 2.3 and Section 2.4 of the previous chapter.↩︎

  2. Fortunately, Shiny app-packages don’t require all the files and folders displayed in the folder tree to gain the functionality and benefits of an R package.↩︎

  3. The mandatory fields are covered in Writing R Extensions, ‘The DESCRIPTION file’↩︎

  4. Always leave an empty final line in the DESCRIPTION file.↩︎

  5. .Rproj files are covered in the What makes an RStudio Project? section of R Packages (2 ed)↩︎

  6. The initial Build Tools settings (i.e., (None)) should help explain the absence of any R package development fields in the moviesApp.Rproj when we compared it to the version in R Packages, 2ed (i.e., BuildType, PackageUseDevtools, PackageInstallArgs, and PackageRoxygenize).↩︎

  7. The mandatory fields DESCRIPTION fields are: Package, Version, License, Description, Title, Author, and Maintainer↩︎

  8. The .Rproj package configuration fields are BuildType: Package, PackageUseDevtools: Yes, and PackageInstallArgs: --no-multiarch --with-keep.source.↩︎

  9. This information comes from the Writing R Packages documentation for Posit Workbench.↩︎

  10. usethis::create_package() is covered in the ‘Fundamental development workflows’ chapter of R Packages, 2ed↩︎