-
Notifications
You must be signed in to change notification settings - Fork 756
Package basics
An R package is the basic unit of reusable code. You need to master the art of making R packages if you want others to use your code. At their heart, packages are quite simple and only have two essential components:
-
a
DESCRIPTION
file that describes the package, what is does, who's allowed to use it (the license) and who to contact if you need help -
an
R
directory that contains your R code.
If you want to distribute R code to someone else, there's no excuse not to use a simple package: it's a standard structure, and you can easily build it out into a complete package by adding documentation, data and tests.
This document explains how to get started, with a description of package structure, tips for naming your package, and more details about the DESCRIPTION
file.
The most accurate resource for up-to-date details on package development is always the official writing R extensions guide. However, it's rather hard to read and follow if you're not already familiar with the basis of packages. It's also exhaustive, covering every possible package component, rather than focussing on the most common and useful components as this package does. Once you are familiar with the content here, you should find R extensions a little easier to read.
As mentioned above, there are only two elements that you must have:
-
the
DESCRIPTION
file, which provides metadata about the package, and is described in the following section. -
the
R/
directory where your R code lives (in.R
or.r
files).
If you don't want to create this by hand, you can use devtools::create
which initialises the directory structure and includes a few other files that most packages have.
Almost all R packages also have:
- the
man/
directory where your function documentation. In the style of package development described in this book, you'll never personally touch the files in this directory. Instead, they will be automatically generated from comments in your source code using theroxygen2
package
After the code and function documentation, the most important optional components of an R package help your users learn how to use your package. The following files and directories are described in more detail in package documentation.
-
the
NEWS
file describes the changes in each version of the package. Using the standard R format will allow you to take advantage of many automated tools for displaying changes between versions. -
the
README
file gives a general overview of your package, including why it's important. This text should be included in any package announcement, to help others understand why they might want to use your package. -
the
inst/CITATION
file describes how to cite your package. If you have published a peer reviewed article which you'd like people to cite when they use your software, this is the place to put it. -
the
demo/
directory contains larger scale demos, that use many features of the package. -
the
inst/doc/
directory is used for larger scale documentation, like vignettes, long-form documents which show how to combine multiple the parts of your package to solve problems.
Other optional files and directories are part of good development practice:
-
a
NAMESPACE
file describes which functions are part of the formal API of the package and are available for others to use. See namespaces for more details. -
tests/
andinst/tests/
contains unit tests which ensure that your package is operating as designed. In this book, we'll focus on thetestthat
package for writing tests. -
the
data/
directory contains.rdata
files, used to include sample datasets (or other R objects) with your package.
There are other directories that we won't cover. You might see these in other packages you download from CRAN, but these topics are outside the scope of this book.
-
src/
: C, C++ and fortran source code -
exec/
: executable scripts -
po/
: translation files
When creating a package the first thing (and sometimes the most difficult) is to come up with a name for it. There's only one formal requirement:
- The package name can only consist of letters and numbers, and must start with a letter.
But I have a few additional recommendations:
-
Make the package name googleable, so that if you google the name you can easily find it. This makes it easy for potential users to find your package, and it's also useful for you, because it makes it easier to find out who is using it.
-
Avoid using both upper and lower case letters: they make the package name hard to type and hard to remember. For example, I can never remember if it's
Rgtk2
orRGTK2
orRGtk2
.
Some strategies I've used in the past to create packages names:
-
Use abbreviations:
lvplot
(letter value plots),meifly
(models explored interactively) -
Add an extra R:
stringr
(string processing),tourr
(grand tours),httr
(HTTP requests),helpr
(alternative documentation view) -
Find a name evocative of the problem and modify it so that it's unique:
plyr
(generalisation of apply tools),lubridate
(makes dates and times easier),mutatr
(mutable objects),classifly
(high-dimensional views of classification)
Once you have a name, create a directory with that name, and inside that create an R
subdirectory and a DESCRIPTION
file (note that's there's no extension, and the file name must be all upper case).
The R/
directory contains all your R code, so copy in any existing code.
It's up to you how you arrange your functions into files. There are two possible extremes: all functions in one file, and each function in it's own file. I think these are both too extreme, and I suggest grouping related functions into a single file. My rule of thumb is that if I can't remember which file a function lives in, I probably need to split them up into more files: having only one function in a file is perfectly reasonable, particularly if the functions are large or have a lot of documentation. As you'll see in the next capture, often the code for the function is small compared to it's documentation (it's much easier to do something than it is to explain to someone else how to do it.)
The next step is to create a DESCRIPTION
file that defines package metadata.
A minimal description file (this one is taken from an early version of plyr) looks like this:
Package: plyr
Title: Tools for splitting, applying and combining data
Description:
Version: 0.1
Authors@R:
Maintainer: Hadley Wickham <[email protected]>
Author: Hadley Wickham <[email protected]>
License: MIT
This is the critical subset of package metadata: what it's called (Package
), what it does (Title
, Description
), who's allowed to use and distribute it (License
), who wrote it (Author
), and who to contact if you have problems (Maintainer
). Here I've left the Description
blank to illustrate that if you haven't decided what the correct value is yet, it's ok to leave it blank.
Again, the six required elements are:
-
Package
: name of the package. Should be the same as the directory name. -
Title
: a one line description of the package. -
Description
: a more detailed paragraph-length description. -
Version
: the version number, which should be of the the formmajor.minor.patchlevel
. See?package_version
for more details on the package version formats. I recommended following the principles of semantic versioning. -
Maintainer
: a single name and email address for the person responsible for package maintenance. -
License
: a standard abbreviation for an open source license, likeGPL-2
orBSD
. A complete list of possibilities can be found by runningfile.show(file.path(R.home(), "share/licenses/license.db"))
. If you are using a non-standard license, putfile LICENSE
and then include the full text of the license in aLICENSE
.
A more complete DESCRIPTION
(this one from a more recent version of plyr
) looks like this:
Package: plyr
Title: Tools for splitting, applying and combining data
Description: plyr is a set of tools that solves a common set of
problems: you need to break a big problem down into manageable
pieces, operate on each pieces and then put all the pieces back
together. For example, you might want to fit a model to each
spatial location or time point in your study, summarise data by
panels or collapse high-dimensional arrays to simpler summary
statistics. The development of plyr has been generously supported
by BD (Becton Dickinson).
URL: http://had.co.nz/plyr
Version: 1.3
Maintainer: Hadley Wickham <[email protected]>
Author: Hadley Wickham <[email protected]>
Depends: R (>= 2.11.0)
Suggests: abind, testthat (>= 0.2), tcltk, foreach
Imports: itertools, iterators
License: MIT
This DESCRIPTION
includes other components that are optional, but still
important:
-
Depends
,Suggests
,Imports
andEnhances
describe the which packages this package needs. They are described in more detail in namespaces. -
URL
: a url to the package website. Multiple urls can be separated with a comma or whitespace.
Instead of Maintainer
and Author
, you can Authors@R
, which takes a vector of person()
elements. Each person object specifies the name of the person and their role in creating the package:
-
aut
: full authors who have contributed much to the package -
ctb
: people who have made smaller contributions, like patches. -
cre
: the package creator/maintainer, the person you should bother if you have problems
Other roles are listed in the help for person. Using Authors@R
is useful when your package gets bigger and you have multiple contributors that you want to acknowledge appropriately. The equivalent Authors@R
syntax for plyr would be:
Authors@R: person("Hadley", "Wickham", role = c("aut", "cre"))
There are a number of other less commonly used fields like BugReports
, KeepSource
, OS_type
and Language
. A complete list of the DESCRIPTION
fields that R understands can be found in the R extensions manual.
So far we've just described the structure of a source package: the development version of a package that lives on your computer. There are also two other type of package: bundled packages and binary packages.
A package bundle is a compressed version of a package is a single file. By convention, package bundles in R use the extension .tar.gz
. This is linux convention indicating multiple files have been collapsed into a single file (.tar
) and then compressed using gzip (.gz
). The package bundle is useful if you want to manually distribute your package to another R package developer. It is not OS specific. You can use devtools::build()
to make a package bundle.
If you want to distribute your package to another R user (i.e. someone who doesn't necessarily have the development tools installed) you need to make a binary package. Like a package bundle, a binary package is a single file, but if you uncompress it, you'll see that the internal structure is a little different to a source package:
-
a
Meta/
directory contains a number ofRds
files. These contain cached metadata about the package, like what topics the help files cover and a parsed versions of theDESCRIPTION
files. (If you want to look at what's in these files you can usereadRDS
) -
a
html/
directory contains some files needed for the html help. -
there are no
.R
files in theR/
directory - instead there are three files that store the parsed functions in an efficient format. This is basically the results of loading all the R code and then saving the functions withsave
, but with a little extra metadata to make things as fast as possible. -
If you had an code in the
src/
directory there will now be alibs/
directory that contains the results of compiling that code for 32 bit (i386/
) and 64 bit (x64
)
Binary packages are platform specific: you can't install a windows binary package on a mac or vice versa. You can use devtools::build(binary = TRUE)
to make a package bundle.
An installed package is just a binary package that's been uncompressed into a package library, described next.
A library is a collection of installed packages. You can have multiple libraries on your computer and most people have at least two: one for the recommended packages that come with a base R install (like base
, stats
etc), and one library where the packages you've installed live. The default is to make that directory dependent on which version of R you have installed - that's why you normally lose all your packages when you reinstall R. If you want to avoid this behaviour, you can manually set the R_LIBS
environmental variable to point somewhere else. .libPaths()
tells you where your current libraries live.
This makes When you use library(pkg)
to load a package, R looks through each path in .libPaths()
to see if a directory called pkg
exists.
Package installation is the process whereby a source package gets converted into a binary package and then installed into your local package library. There are a number of tools that automate this process:
-
install.packages()
installs a package from CRAN. Here CRAN takes care of making the binary package and so installation from CRAN basically is equivalent to downloading the binary package value and unzipping it in.libPaths()[1]
(but you should never do this by hand because the process also does other checks) -
devtools::install()
installs a source package from a directory on your computer. -
devtools::install_github()
installs a package that someone has published on their github account. There are a number of similar functions that make it easy to install packages from other internet locations:install_url
,install_gitorious
,install_bitbucket
, and so on.
(to be integrated throughout the chapter)
-
Go to CRAN and download the source and binary for XXX. Unzipped and compare. How do they differ?
-
Download the source packages for XXX, YYY, ZZZ. What directories do they contain?
-
Where is your default library? What happens if you install a new package from CRAN?