Functions and settings in your Rprofile can be dangerous for reproducibility, but can afford some nice workflow tools.


October 6, 2021

.warning {
    display: inline-flex;
    margin: 30px;
    font-weight: 700;
    font-size: 16px;
    /* font-size: 1rem; */
    line-height: 1.25;
    color: #0b0c0c;

.warning:before {
    display: inline-block;
    margin: 10px;
    content: "!";
    color: white;
    background: #d4351c;
    font-weight: 700;
    left: 50px;
    min-width: 35px;
    max-width: 35px;
    max-height: 35px;
    min-height: 35px;
    margin-top: -7px;
    border-radius: 50%;
    font-size: 30px;
    line-height: 35px;
    text-align: center;

In the R programming environment, you have access to configuration files, typically your .Rprofile and your .Renviron. RStudio has a more thorough write-up available at this link. On the start of a new R session your Environment file is available through a Sys.getenv call. Rprofiles are basically “sourced” in the active R session on start and everything in the Rprofile is available within the session. Generally speaking, your environment file is for definitions (like a typical dotfile) which variables given values in the form of:


You can then access this file with the following function call in R


If I had something like:

options(stringsAsFactors = TRUE)

bleh <- function(){

In my R session, strings would be treated as factors and I would have access to the function bleh.

This is where the reproducibility piece comes into play: unless you include the .Rprofile in whatever you share, you collaborators will likely not have access to these setting and functions.

Using a function that is only available to you is not very being very nice to your collaborators or your future self.

So Why am I Writing This

Putting helper functions for your convenience in your Rprofile can be handy. This includes helpers to create files for you (related to project workflow) or tasks that relate to the workflow and not to the reproducibility of the work. They aren’t needed to re-create the outputs of the analysis.

There is an argument to be made that these functions should be wrapped in a package for others to be use. Putting these functions in a package also allows for better testing, continuous integration, and documentation; not to mention that packages can be more easily shared. So why put the function in your Rprofile? Sometimes I use R for quick data visualizations and don’t want to create project and my typical formal project structure. I might read a paper and it elicits a questions that I want to explore further (but then I throw away the script).

A script I have in my R file for exactly this purpose is shown below. The gist is it create an R script in the R directory by default that includes some of the packages that I use frequently. In this way I get a script with a header I will likely use.

cr <- function(x, dir_use = "R"){
  loc <- getwd()

  dirloc <- file.path(loc, dir_use)

  filnm <- sprintf("%s.R", x)

    cli::cli_alert_warning("File exists.")
  } else {
    file.create(file.path(dirloc, filnm))

    cli::cli_alert_success("Created {filnm} in {dir_use}")

    cat("#' Purpose:\nlibrary(tidyverse)\nlibrary(data.table)\n",  
    file = file.path(dirloc, filnm))
    rstudioapi::navigateToFile(file.path(dirloc, filnm))



In theory I could put this function in a package and then export it with the Rprofile so that it is already available. Maybe that represents the best of both worlds.

The key take away is that this function produces something that is indepedent of the main analysis and not needed to reproduce what is in the R script.

As another note, you can edit your Rprofile with:




BibTeX citation:
  author = {Michael DeWitt},
  title = {RProfiles},
  date = {2021-10-06},
  url = {https://michaeldewittjr.com/programming/2021-10-06-rprofiles},
  langid = {en}
For attribution, please cite this work as:
Michael DeWitt. 2021. “RProfiles.” October 6, 2021. https://michaeldewittjr.com/programming/2021-10-06-rprofiles.