set -u
set -u: Treat Unset Variables as Errors
What it does: Makes bash exit with an error when you reference an undefined variable, catching typos and uninitialized variables.
set -u
echo "$TYPO_VARIABLE" # Error: unbound variable (instead of silently empty)
The Main Gotcha
You can't use traditional empty-checking patterns:
set -u
# 💥 FAILS - errors before the test runs
if [[ -z "$MAYBE_UNSET" ]]; then
echo "It's empty"
fi
# ✅ WORKS - use parameter expansion with default
if [[ -z "${MAYBE_UNSET:-}" ]]; then
echo "It's empty or unset"
fi
The :- provides a default value (empty string) if the variable is unset, preventing the error.
When to Use It
Good for new scripts: Starting fresh with set -u enforces discipline - you explicitly handle all variables from the start.
Hard for existing code: Retrofitting set -u into existing scripts/environments is painful. You'll need to:
- Add
:-to hundreds of variable expansions - Initialize all variables that might be unset
- Fix assumptions that unset equals empty
- Handle positional parameters (
$1,$2) defensively
Common Patterns
# Positional parameters
echo "${1:-default}" # Default value if no argument
# Required variables
: "${REQUIRED:?Error: REQUIRED must be set}"
# Optional variables
OPTIONAL="${OPTIONAL:-}" # Initialize as empty if unset
Bottom line: Great for catching bugs in new code, but retrofitting it into existing systems often isn't worth the effort.
Backlinks