renv

Load environment files in R

renv

R vs. Environment Variables

The twelve-factor app stores config in environment variables.

Adam Wiggins, Heroku Founder

If you want to set environment variables for your R session, you don't want to use .bashrc; use .Renviron (user) or .Renviron.site (system).

RStudio Support Ticket

Why not Renviron?

You might be already using bash (or other shell scripting) to manage environment settings, anyway.
Shared environment variables make them available across a range of processes.
Having multiple locations for identical configuration settings is annoying.

Drew's Law: If you ever have the same configuration data stored in multiple locations, at some point it will cause a headache.

Drew Lanenga

When to use?

Think: "When does my code contain configuration settings?"
- Data Location
  - Local directory, S3, FTP, Dropbox, remote database/tables
- Secure Credentials
  - API usernames, passwords, keys
- Staging Environments
  - Different API endpoints/credentials for staging/sandbox vs. production settings

Usage

# get variables from standard Sys.getenv call
renv::load_vars("~/.bashrc")
foo <- Sys.getenv("foo")

# alternatively, assign value at load
vars <- renv::load_vars("~/.bashrc")
foo <- vars$foo

Examples

Example: Plotly!

According to Plotly's Getting Started Guide, save this in your .Rprofile file:

Sys.setenv("plotly_username" = "your_plotly_username")
Sys.setenv("plotly_api_key" = "your_api_key")

Example: Plotly!

plotly.sh

export plotly_username=your_plotly_username
export plotly_api_key=your_api_key

plotly.R or .Rprofile

renv::load_vars("plotly.sh")

Example: Storing S3 Information

s3info.sh

export S3_ROOT=https://mybucket.s3.amazonaws.com
export S3_PATH=rawdata
export S3_URL=$S3_ROOT/$S3_PATH

read.R

vars <- renv::load_vars("s3info.sh")
mydata <- read.csv(paste0(vars$S3_URL, "/mydata.csv"))