Development Workflow

This guide covers the efficient development workflow for R package development.

The Fast Iteration Cycle

IMPORTANT: Follow this workflow to develop efficiently.

During Development (Fast Iteration Cycle)

Run these commands repeatedly while developing:

  1. devtools::document() - Generate documentation from roxygen2 comments
    • Updates man/ directory
    • Updates NAMESPACE with exports and imports
    • Takes seconds to run
  2. devtools::load_all() - Load your package into memory
    • Equivalent to library(yourpackage) but uses development version
    • Loads all functions, even unexported ones
    • Takes seconds to run
  3. devtools::test() - Run all tests
    • Runs testthat test suite
    • Takes seconds to minutes depending on test count
    • For single file: devtools::test_active_file('R/yourfile.R')

This cycle is fast (seconds) and gives you immediate feedback.

Why This Works

# Typical workflow:
devtools::document()   # Update docs and namespace
devtools::load_all()   # Load updated code
devtools::test()       # Verify it works

# Test your function interactively
your_function(test_data)

# Find issue, fix code, repeat

This tight feedback loop lets you: - Catch errors immediately - Test interactively - Iterate quickly on design

Final Validation (Run Once at End)

  1. devtools::check() - Full R CMD check

WARNING: Do NOT run check() during iteration.

Why avoid check() during development?

  • Takes 1-2 minutes (sometimes longer)
  • Runs many validations unnecessary during development
  • Interrupts your flow
  • Everything it checks is validated by the fast cycle

When to run check()

Only run check() once at the very end: - Before submitting to CRAN - Before creating a release - Before final code review - After all development is complete

check() validates: - Package structure - Documentation completeness - Code quality - Examples run correctly - Tests pass - No NOTEs or WARNINGs

Interactive Development

Testing individual functions

# Load your package
devtools::load_all()

# Test your function
result <- your_function(mtcars, mpg, hp)
result

# Check intermediate values
debug(your_function)
your_function(mtcars, mpg, hp)

Using browser() for debugging

your_function <- function(data, ...) {
  # Add browser() where you want to pause
  browser()

  # Rest of function
}

# Run function - it will pause at browser()
your_function(test_data)

Common Workflow Patterns

Pattern 1: New Function Development

# 1. Write function in R/your-file.R
# 2. Add roxygen documentation
# 3. Document and load
devtools::document()
devtools::load_all()

# 4. Test interactively
your_function(test_data)

# 5. Write formal tests in tests/testthat/test-your-file.R
# 6. Run tests
devtools::test()

# 7. Fix issues, go to step 3

Pattern 2: Fixing a Bug

# 1. Write a failing test that demonstrates the bug
# 2. Run tests to confirm failure
devtools::test()

# 3. Fix the code
# 4. Document and load
devtools::document()
devtools::load_all()

# 5. Verify test now passes
devtools::test()

Pattern 3: Updating Documentation

# 1. Update roxygen comments in R code
# 2. Regenerate documentation
devtools::document()

# 3. Preview documentation
?your_function

# 4. Repeat until satisfied

Keyboard Shortcuts (RStudio)

Speed up your workflow with keyboard shortcuts:

  • Cmd/Ctrl + Shift + D - devtools::document()
  • Cmd/Ctrl + Shift + L - devtools::load_all()
  • Cmd/Ctrl + Shift + T - devtools::test()
  • Cmd/Ctrl + Shift + E - devtools::check()

Best Practices

DO:

  • Run the fast cycle (document, load, test) frequently
  • Test interactively in console
  • Write tests as you develop
  • Fix issues immediately when found
  • Use browser() for complex debugging

DON’T:

  • Run check() during development (only at end)
  • Skip writing tests “for later” (write them now)
  • Ignore warnings from document() or test()
  • Make many changes before testing
  • Commit broken code to git

Troubleshooting

“could not find function”

Problem: Function exists in R code but not loaded

Solution: Run devtools::load_all()

“object not found in namespace”

Problem: Missing @export in roxygen or namespace not updated

Solution:

# Add @export to your roxygen block
devtools::document()  # Update namespace
devtools::load_all()  # Load updated package

Tests fail but function works interactively

Problem: Test environment differs from interactive environment

Solution: - Check test data setup - Verify all required packages loaded in tests - Use devtools::load_all() before interactive testing

“no visible binding for global variable”

Problem: Using NSE without proper imports

Solution: See package-imports.md for proper namespace usage

Performance Tips

Speed up test runs

# Run only one test file
devtools::test_active_file()

# Run only tests matching a pattern
devtools::test(filter = "your_function")

Skip slow tests during development

test_that("slow integration test", {
  skip_if_not(interactive(), "Slow test - run manually")

  # Expensive test here
})

Use testthat snapshots for complex output

test_that("complex output matches", {
  expect_snapshot(your_function(data))
})

Git Workflow (Source Development)

If you’re contributing to tidymodels packages (recipes, yardstick, etc.), you’ll also need basic git workflow.

Initial Setup

# Clone the repository
git clone https://github.com/tidymodels/recipes.git
cd recipes

# Create a feature branch
git checkout -b feature/add-step-name

During Development

# Your fast iteration cycle (unchanged)
devtools::document()
devtools::load_all()
devtools::test()

Committing Changes

# Check what changed
git status
git diff

# Add your changes
git add R/step_name.R tests/testthat/test-step_name.R

# Commit with descriptive message
git commit -m "Add step_name() for [brief description]"

# Push to your fork
git push origin feature/add-step-name

Before Submitting PR

  1. Run full check: devtools::check()
  2. Update NEWS.md with your changes
  3. Ensure all tests pass
  4. Push final commits

Common Git Commands

# See current branch
git branch

# Switch branches
git checkout main

# Update from upstream
git pull origin main

# View commit history
git log --oneline

Note: This is a minimal git guide. For complete git documentation, see the official Git documentation or GitHub guides.

Next Steps