reprex: Help me help you

How to create a minimal reproducible example using reprex

While teaching cadres of students at University of British Columbia in Stat 545, Jenny Bryan found herself trying to parse snippets of code from GitHub issues where students had posted questions. Some questions were easy to address, but other questions were not clear about what was being asked. She found that it was not straightforward for her students to create reproducible examples. That experience led her to create the R package reprex.

By running code with the reprex, you can create a reproducible example to post to Stack Overflow, Slack, GitHub, or an email to your collaborator.

Using reprex

One important aspect of reprex is that it works best with whatever is currently saved on your clipboard (that is, highlight with your mouse and use command-c or control-c). The reprex vignettes provide great examples that are definitely worth looking through!

Below we work through a few examples of using reprex to debug some code. We include the code-to-debug directly into the reprex() command, but again, you may want to work entirely within copy-and-paste.

Example: first try

library(reprex)
data(iris)

reprex(iris %>% summarize(Petal.Length))

Note that the error message given is that the library(dplyr) command is missing. reprex is checking to make sure all the data and packages are appropriately loaded! The information above is now automatically stored on your clipboard, and you can paste it directly to Stack Overflow, Slack, or GitHub issues as needed.

Example: second try

library(dplyr)
data(iris)

reprex(iris %>% summarize(Petal.Length))

Wait, I thought I loaded dplyr??? But the library(dplyr) command is not contained in the code which was submitted to reprex, so reprex didn’t know that library(dplyr) command had been provided previously.

Example: third try

data(iris)

reprex({library(dplyr); iris %>% summarize(Petal.Length)})

For the third try, we make sure that the library(dplyr) command is part of the reproducible example. Now it seems like there is a different error; we forgot the mean function in summarize.

Example: fourth try

data(iris)

reprex({library(dplyr); iris %>% summarize(Petal.Length)})

On the fourth try, it works! The code runs as we thought it would.

Copy and paste

It’s worth pointing out one more time that reprex works extremely well with code that is on your clipboard. So instead of writing the code inside the reprex() call (as done above), a more typical use would be to copy the code with your mouse and then type reprex() in the console of RStudio. The output (seen above) is then automatically copied onto your clipboard (replacing the code) and you can paste into Stack Overflow, Slack, or GitHub as needed.

In the classroom

As with all things in life, practice is key. reprex is surprisingly straightforward, despite possibly seeming like one more thing to learn. If your class uses a slack channel or piazza for classroom communication, a very short semester-long assignment would be a requirement of posting at least one reprex-generated technical question to the class. Our experience is that if students get used to posting technical questions to the class, the process immediately becomes nonthreatening and useful.