agentleFS
worldbankworldbank/geoPovLearn/.claude/skills/testing-r-packages/SKILL.md

testing-r-packages

Skill2 starsChanged 6 months agoMITView on GitHub

Best practices for writing R package tests using testthat version 3+. Use when writing, organizing, or improving tests for R packages. Covers test structure, expectations, fixtures, snapshots, mocking, and modern testthat 3 patterns including self-sufficient tests, proper cleanup with withr, and snapshot testing.

1 other project uses a version of this, 1 of them edited.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Use usethis::use_r("foofy") and usethis::use_test("foofy") to create paired files

    line 28
  2. Distinctive:

    Test descriptions: should read naturally and describe behavior, not implementation.

    line 48
  3. Very specific:

    Use withr to manage state changes:

    line 185
  4. Distinctive:

    Write tests assuming they will fail and need debugging:

    line 208
  5. Distinctive:

    Avoid hidden dependencies on earlier tests

    line 210
  6. Very specific:

    Use devtools::load_all() instead of library()

    line 220
  7. Distinctive:

    1. Constructor functions: - Create data on-demand:

    line 253
  8. Distinctive:

    When working with testthat 3 code, prefer modern patterns:

    line 397

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Testing R Packages with testthat

Modern best practices for R package testing using testthat 3+.

Initial Setup

Initialize testing with testthat 3rd edition:

usethis::use_testthat(3)

This creates tests/testthat/ directory, adds testthat to DESCRIPTION Suggests with Config/testthat/edition: 3, and creates tests/testthat.R.

File Organization

Mirror package structure:

  • Code in R/foofy.R → tests in tests/testthat/test-foofy.R
  • Use usethis::use_r("foofy") and usethis::use_test("foofy") to create paired files

Special files:

  • helper-*.R - Helper functions and custom expectations, sourced before tests
  • setup-*.R - Run during R CMD check only, not during load_all()
  • fixtures/ - Static test data files accessed via test_path()

Test Structure

Tests follow a three-level hierarchy: File → Test → Expectation

Standard Syntax

test_that("descriptive behavior", {
  result <- my_function(input)
  expect_equal(result, expected_value)
})

Test descriptions should read naturally and describe behavior, not implementation.

BDD Syntax (describe/it)

For behavior-driven development, use describe() and it():

describe("matrix()", {
  it("can be multiplied by a scalar", {
    m1 <- matrix(1:4, 2, 2)
    m2 <- m1 * 2
    expect_equal(matrix(1:4 * 2, 2, 2), m2)
  })

  it("can be transposed", {
    m <- matrix(1:4, 2, 2)
    expect_equal(t(m), matrix(c(1, 3, 2, 4), 2, 2))
  })
})

Key features:

  • describe() groups related specifications for a component
  • it() defines individual specifications (like test_that())
  • Supports nesting for hierarchical organization
  • it() without code creates pending test placeholders

Use describe() to verify you implement the right things, use test_that() to ensure you do things right.

See references/bdd.md for comprehensive BDD patterns, nested specifications, and test-first workflows.

Running Tests

Three scales of testing:

Micro (interactive development):

devtools::load_all()
expect_equal(foofy(...), expected)

Mezzo (single file):

testthat::test_file("tests/testthat/test-foofy.R")
# RStudio: Ctrl/Cmd + Shift + T

Macro (full suite):

devtools::test()    # Ctrl/Cmd + Shift + T
devtools::check()   # Ctrl/Cmd + Shift + E

Core Expectations

Equality

expect_equal(10, 10 + 1e-7)      # Allows numeric tolerance
expect_identical(10L, 10L)       # Exact match required
expect_all_equal(x, expected)    # Every element matches (v3.3.0+)

Errors, Warnings, Messages

expect_error(1 / "a")
expect_error(bad_call(), class = "specific_error_class")
expect_no_error(valid_call())

expect_warning(deprecated_func())
expect_no_warning(safe_func())

expect_message(informative_func())
expect_no_message(quiet_func())

Pattern Matching

expect_match("Testing is fun!", "Testing")
expect_match(text, "pattern", ignore.case = TRUE)

Structure and Type

expect_length(vector, 10)
expect_type(obj, "list")
expect_s3_class(model, "lm")
expect_s4_class(obj, "MyS4Class")
expect_r6_class(obj, "MyR6Class")      # v3.3.0+
expect_shape(matrix, c(10, 5))         # v3.3.0+

Sets and Collections

expect_setequal(x, y)           # Same elements, any order
expect_contains(fruits, "apple") # Subset check (v3.2.0+)
expect_in("apple", fruits)       # Element in set (v3.2.0+)
expect_disjoint(set1, set2)      # No overlap (v3.3.0+)

Logical

expect_true(condition)
expect_false(condition)
expect_all_true(vector > 0)      # All elements TRUE (v3.3.0+)
expect_all_false(vector < 0)     # All elements FALSE (v3.3.0+)

Design Principles

1. Self-Sufficient Tests

Each test should contain all setup, execution, and teardown code:

# Good: self-contained
test_that("foofy() works", {
  data <- data.frame(x = 1:3, y = letters[1:3])
  result <- foofy(data)
  expect_equal(result$x, 1:3)
})

# Bad: relies on ambient state
dat <- data.frame(x = 1:3, y = letters[1:3])
test_that("foofy() works", {
  result <- foofy(dat)  # Where did 'dat' come from?
  expect_equal(result$x, 1:3)
})

2. Self-Contained Tests (Cleanup Side Effects)

Use withr to manage state changes:

test_that("function respects options", {
  withr::local_options(my_option = "test_value")
  withr::local_envvar(MY_VAR = "test")
  withr::local_package("jsonlite")

  result <- my_function()
  expect_equal(result$setting, "test_value")
  # Automatic cleanup after test
})

Common withr functions:

  • local_options() - Temporarily set options
  • local_envvar() - Temporarily set environment variables
  • local_tempfile() - Create temp file with automatic cleanup
  • local_tempdir() - Create temp directory with automatic cleanup
  • local_package() - Temporarily attach package

3. Plan for Test Failure

Write tests assuming they will fail and need debugging:

  • Tests should run independently in fresh R sessions
  • Avoid hidden dependencies on earlier tests
  • Make test logic explicit and obvious

4. Repetition is Acceptable

Repeat setup code in tests rather than factoring it out. Test clarity is more important than avoiding duplication.

5. Use devtools::load_all() Workflow

During development:

  • Use devtools::load_all() instead of library()
  • Makes all functions available (including unexported)
  • Automatically attaches testthat
  • Eliminates need for library() calls in tests

Snapshot Testing

For complex output that's difficult to verify programmatically, use snapshot tests. See references/snapshots.md for complete guide.

Basic pattern:

test_that("error message is helpful", {
  expect_snapshot(
    error = TRUE,
    validate_input(NULL)
  )
})

Snapshots stored in tests/testthat/_snaps/.

Workflow:

devtools::test()                    # Creates new snapshots
testthat::snapshot_review('name')   # Review changes
testthat::snapshot_accept('name')   # Accept changes

Test Fixtures and Data

Three approaches for test data:

1. Constructor functions - Create data on-demand:

new_sample_data <- function(n = 10) {
  data.frame(id = seq_len(n), value = rnorm(n))
}

2. Local functions with cleanup - Handle side effects:

local_temp_csv <- function(data, env = parent.frame()) {
  path <- withr::local_tempfile(fileext = ".csv", .local_envir = env)
  write.csv(data, path, row.names = FALSE)
  path
}

3. Static fixture files - Store in fixtures/ directory:

data <- readRDS(test_path("fixtures", "sample_data.rds"))

See references/fixtures.md for detailed fixture patterns.

Mocking

Replace external dependencies during testing using local_mocked_bindings(). See references/mocking.md for comprehensive mocking strategies.

Basic pattern:

test_that("function works with mocked dependency", {
  local_mocked_bindings(
    external_api = function(...) list(status = "success", data = "mocked")
  )

  result <- my_function_that_calls_api()
  expect_equal(result$status, "success")
})

Common Patterns

Testing Errors with Specific Classes

test_that("validation catches errors", {
  expect_error(
    validate_input("wrong_type"),
    class = "vctrs_error_cast"
  )
})

Testing with Temporary Files

test_that("file processing works", {
  temp_file <- withr::local_tempfile(
    lines = c("line1", "line2", "line3")
  )

  result <- process_file(temp_file)
  expect_equal(length(result), 3)
})

Testing with Modified Options

test_that("output respects width", {
  withr::local_options(width = 40)

  output <- capture_output(print(my_object))
  expect_lte(max(nchar(strsplit(output, "\n")[[1]])), 40)
})

Testing Multiple Related Cases

test_that("str_trunc() handles all directions", {
  trunc <- function(direction) {
    str_trunc("This string is moderately long", direction, width = 20)
  }

  expect_equal(trunc("right"), "This string is mo...")
  expect_equal(trunc("left"), "...erately long")
  expect_equal(trunc("center"), "This stri...ely long")
})

Custom Expectations in Helper Files

# In tests/testthat/helper-expectations.R
expect_valid_user <- function(user) {
  expect_type(user, "list")
  expect_named(user, c("id", "name", "email"))
  expect_type(user$id, "integer")
  expect_match(user$email, "@")
}

# In test file
test_that("user creation works", {
  user <- create_user("test@example.com")
  expect_valid_user(user)
})

File System Discipline

Always write to temp directory:

# Good
output <- withr::local_tempfile(fileext = ".csv")
write.csv(data, output)

# Bad - writes to package directory
write.csv(data, "output.csv")

Access test fixtures with test_path():

# Good - works in all contexts
data <- readRDS(test_path("fixtures", "data.rds"))

# Bad - relative paths break
data <- readRDS("fixtures/data.rds")

Advanced Topics

For advanced testing scenarios, see:

  • references/bdd.md - BDD-style testing with describe/it, nested specifications, test-first workflows
  • references/snapshots.md - Snapshot testing, transforms, variants
  • references/mocking.md - Mocking strategies, webfakes, httptest2
  • references/fixtures.md - Fixture patterns, database fixtures, helper files
  • references/advanced.md - Skipping tests, secrets management, CRAN requirements, custom expectations, parallel testing

testthat 3 Modernizations

When working with testthat 3 code, prefer modern patterns:

Deprecated → Modern:

  • context() → Remove (duplicates filename)
  • expect_equivalent() → expect_equal(ignore_attr = TRUE)
  • with_mock() → local_mocked_bindings()
  • is_null(), is_true(), is_false() → expect_null(), expect_true(), expect_false()

New in testthat 3:

  • Edition system (Config/testthat/edition: 3)
  • Improved snapshot testing
  • waldo::compare() for better diff output
  • Unified condition handling
  • local_mocked_bindings() works with byte-compiled code
  • Parallel test execution support

Quick Reference

Initialize: usethis::use_testthat(3)

Run tests: devtools::test() or Ctrl/Cmd + Shift + T

Create test file: usethis::use_test("name")

Review snapshots: testthat::snapshot_review()

Accept snapshots: testthat::snapshot_accept()

Find slow tests: devtools::test(reporter = "slow")

Shuffle tests: devtools::test(shuffle = TRUE)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.