12_R_Packages.RmdToday, folks, we’re going to take a look at the work we’ll be conducting for project three. If you need a refresher, take a look at project three on the course site. tl;dr: It’s an R package!
The nice thing is that creating R packages has never been easier due
to packages like devtools and usethis. Let’s
load them both before we continue:
Now that this is done, we will create the skeleton of your R package. Please make sure you’re wherever you want that package to be located.
create_package(".")Visually verify that this has worked. In your R package directory, you will now have some new files: DESCRIPTION, NAMESPACE, R/, and vignettes.
Next, please take the function we wrote in class and make sure it is
saved in the R/ folder. We’re going to begin by writing some
documentation. Open the file containing your function. Click on the
magic wand icon, and scroll down to
Insert Roxygen Skeleton. This will create the necessary
documentation lines.
The title should be a short, snappy description of what your file
does. The params should be one-sentence descriptions of the inputs
needed to make your function go brrrrrrrrr. The return should be a
description of the output type. A researcher should be able to look at
your return statement to understand if their outputs are correct. If you
have technical details, you can add a @details to provide
more information to your users. When you have completed this, save and
close your file.
Now, we will edit the DESCRIPTION file. This is the package metadata that enables R to install your package. Minimally, you must fill in the Package, Title, Authors, and Description lines today.
Now that we have this, we can try documenting our package. Type
document()Then visually check: you should now have a man folder. In it should be manual files for any R functions you documented. Open adn preview these - they look just like the documentation for real R functions! Because they are!
Next, we can try installing our R package:
install()This is where we may run into problems. If we import functions from packages, we need to tell R that users should have those packages installed, and to do the importing. Adding dependencies is a two-step process. First, we add them to the DESCRIPTION, like so:
use_package("dplyr")If you used the pipe:
use_pipe()This can be tricky. We often work with tidyverse, which is a metapackage, or group of packages. R does not want that to be a dependency. If you can’t figure out what package within tidyverse a function comes from, look at its helpfile. It will be listed at the top.
The second part of this is ensuring our imports within the function
are in package::function format. Please do that now.
When you are done, document, check, and install your package.
Lastly, we’re going to put these on github. If you’re having github PAT issues, please do the following:
create_github_token()
set_github_pat()
edit_r_environ()You’ll add the following line to your R environ file:
GITHUB_PAT=TOKEN HERENow, add, commit and push your package files. Get together with a
partner and try installing their R package with
devtools::install_github().
check() and install() frequently to catch
errors before they snowball