Writing Robust Functions with stop() and warning()
Signal custom conditions from within your own functions.
Writing Robust Functions with stop() and warning() is a free R Academy lesson on CoddyKit — lesson 4 of 4. You can read the complete lesson below for free — then practise it hands-on in the browser with a built-in code editor and a 24/7 AI tutor. It is part of the R Academy learning path, one of 4 lessons in the course, and your progress syncs across the web and the CoddyKit app.
Why Input Validation Matters
Defensive functions check their inputs and fail loudly with informative errors rather than silently producing wrong results. A function that catches bad input at the boundary is far easier to debug than one that propagates garbage through a pipeline.
# Without validation: wrong result, no error
bad_mean <- function(x) sum(x) / length(x)
bad_mean('hello') # no error, just NA
# With validation: clear error
good_mean <- function(x) {
if (!is.numeric(x)) stop('x must be numeric')
sum(x) / length(x)
}
tryCatch(good_mean('hello'), error = function(e) cat(e$message, '\n'))Informative Error Messages
Generic errors like stop('bad input') are frustrating to debug. Use paste0() or sprintf() inside stop() to include the actual value and expected type in the message. This saves time when tracing errors in pipelines.
check_numeric <- function(x, arg_name = 'x') {
if (!is.numeric(x)) {
stop(sprintf(
"'%s' must be numeric, but got class '%s'",
arg_name, class(x)
))
}
}
tryCatch(check_numeric('abc', 'score'),
error = function(e) cat(e$message, '\n'))Validating Length and Range
Common validations: check that a vector is not empty (length(x) == 0), that a scalar is within range, or that a data frame has the expected columns. Each check should have its own tailored error message.
validate_prob <- function(p) {
if (!is.numeric(p)) stop(paste0('p must be numeric, got: ', class(p)))
if (length(p) != 1) stop(paste0('p must be length 1, got: ', length(p)))
if (p < 0 || p > 1) stop(paste0('p must be in [0,1], got: ', p))
p
}
cat(validate_prob(0.7), '\n')
tryCatch(validate_prob(1.5), error = function(e) cat(e$message, '\n'))Using warning() for Non-Fatal Issues
Use warning() when the function can still return a sensible result but the caller should know something unusual happened — like coercing a value, imputing a default, or detecting a borderline input.
clamp <- function(x, lo, hi) {
if (any(x < lo)) warning(paste0(sum(x < lo), ' value(s) below lo, clamped'))
if (any(x > hi)) warning(paste0(sum(x > hi), ' value(s) above hi, clamped'))
pmax(lo, pmin(hi, x))
}
result <- clamp(c(2, -5, 7, 100, 4), 0, 10)
cat('Result:', result, '\n')call. = FALSE in stop()
By default, stop() prepends Error in funcname(...): to the message. Pass call. = FALSE to suppress this prefix when the calling context is obvious or when the function is a user-facing utility.
# With call. = TRUE (default): shows calling function
f1 <- function(x) stop('bad input')
tryCatch(f1(1), error = function(e) cat(conditionMessage(e), '\n'))
# With call. = FALSE: cleaner message
f2 <- function(x) stop('bad input', call. = FALSE)
tryCatch(f2(1), error = function(e) cat(conditionMessage(e), '\n'))Custom Condition Classes for stop()
Create a custom error class using structure() so callers can catch just your function's errors without catching all errors. This is the professional way to write package-level errors.
value_error <- function(msg, call = sys.call(-1)) {
structure(
class = c('value_error', 'error', 'condition'),
list(message = msg, call = call)
)
}
check_positive <- function(x) {
if (x <= 0) stop(value_error(paste0('Expected positive, got: ', x)))
x
}
tryCatch(check_positive(-3),
value_error = function(e) cat('ValueError:', e$message, '\n'),
error = function(e) cat('Other error:', e$message, '\n'))Checking Data Frame Columns
When a function expects a data frame, validate that required columns exist before using them. Use %in% to check column names and setdiff() to report which columns are missing.
require_cols <- function(df, cols) {
missing <- setdiff(cols, names(df))
if (length(missing) > 0) {
stop(paste0('Missing columns: ', paste(missing, collapse = ', ')))
}
invisible(df)
}
df <- data.frame(x = 1:3, y = 4:6)
tryCatch(
require_cols(df, c('x', 'z', 'w')),
error = function(e) cat(e$message, '\n')
)stopifnot() for Concise Assertions
stopifnot(condition1, condition2, ...) throws an error if any condition is FALSE. It is a concise way to assert preconditions at the top of a function without writing separate if (!...) stop(...) blocks.
compute_area <- function(width, height) {
stopifnot(
is.numeric(width),
is.numeric(height),
width > 0,
height > 0
)
width * height
}
cat(compute_area(5, 3), '\n')
tryCatch(compute_area(-1, 3), error = function(e) cat(e$message, '\n'))Named stopifnot() for Better Messages
In R 4.0+, named expressions in stopifnot() replace the auto-generated message with your custom description. Use the expression text as the name to produce informative failures.
validate_age <- function(age) {
stopifnot(
'age must be numeric' = is.numeric(age),
'age must be positive' = age > 0,
'age must be under 150' = age < 150
)
invisible(age)
}
tryCatch(validate_age(-5), error = function(e) cat(e$message, '\n'))
tryCatch(validate_age('x'), error = function(e) cat(e$message, '\n'))Combining Validation and Business Logic
A well-structured function separates validation from logic. Validate at the top, then perform the operation. This makes the function easier to read and ensures errors are detected before any computation begins.
discount_price <- function(price, pct) {
if (!is.numeric(price) || price <= 0)
stop(paste0('price must be positive numeric, got: ', price))
if (!is.numeric(pct) || pct < 0 || pct > 100)
stop(paste0('pct must be in [0,100], got: ', pct))
if (pct > 50) warning('discount > 50% is unusual')
price * (1 - pct / 100)
}
cat(discount_price(100, 20), '\n')
cat(suppressWarnings(discount_price(100, 60)), '\n')tryCatch Inside Utility Functions
Sometimes a utility function should handle errors internally and return a fallback rather than propagating them. Wrap the core logic in tryCatch() and return an agreed-upon sentinel like NA or NULL on failure.
safe_log <- function(x) {
tryCatch({
if (!is.numeric(x)) stop('not numeric')
if (x <= 0) stop('must be positive')
log(x)
}, error = function(e) {
warning(paste0('safe_log failed for x=', x, ': ', e$message))
NA_real_
})
}
results <- sapply(list(10, -1, 'a', 100), safe_log)
cat(results, '\n')Quick Check
What does call. = FALSE do when passed to stop()?
Robust Functions: Key Takeaways
Key takeaways for writing robust functions:
- Validate inputs at the top; fail loudly with informative
stop()messages - Include the actual value in error messages:
paste0('Expected numeric, got: ', class(x)) - Use
warning()for non-fatal issues where computation can continue call. = FALSEsuppresses the calling function prefix in error messagesstopifnot()for concise assertions; named form for custom messages (R 4.0+)- Custom condition classes enable selective catching by callers
robust_divide <- function(x, y) {
stopifnot('x must be numeric' = is.numeric(x),
'y must be numeric' = is.numeric(y))
if (y == 0) stop('Division by zero', call. = FALSE)
if (abs(y) < 1e-10) warning('y is very small; result may be inaccurate')
x / y
}
cat(robust_divide(10, 2), '\n')
tryCatch(robust_divide(10, 0), error = function(e) cat(e$message, '\n'))Frequently asked questions
Is the “Writing Robust Functions with stop() and warning()” lesson free?
Yes — the full text of “Writing Robust Functions with stop() and warning()” is free to read here on the web, and the R Academy course includes 4 lessons in total. To practise it interactively (a built-in code editor and a 24/7 AI tutor) and unlock the rest of the R Academy course, upgrade to CoddyKit PRO.
What will I learn in “Writing Robust Functions with stop() and warning()”?
Signal custom conditions from within your own functions. You practise R Academy with hands-on code you run directly in the browser, and a 24/7 AI tutor answers your questions as you work through the lesson.
Do I need any experience to start R Academy?
No prior experience is required. R Academy on CoddyKit is structured for beginners through advanced learners; this is — lesson 4 of 4, so you can start here or from the beginning and move at your own pace.
How long does the “Writing Robust Functions with stop() and warning()” lesson take?
Most CoddyKit lessons take about 5–10 minutes. Each one is bite-sized and interactive, so you make steady progress and pick up exactly where you left off across the web and the app.
Can I write and run code in this R Academy lesson?
Yes. Every R Academy lesson includes a built-in code editor, so you write and run real code right in your browser and get instant AI feedback — no local setup required.
All lessons in this course
- Errors, Warnings, and Messages in R
- tryCatch() for Error Recovery
- withCallingHandlers() and Restarts
- Writing Robust Functions with stop() and warning()