Joining strings with separators is a common task that seems trivial until you need to handle spaces, special characters, or maintain portability across different Unix-like systems. The naive approaches often fail when dealing with real-world input.

Many quick solutions have limitations:

# Fails with spaces:
echo "foo bar baz" | sed 's/ /---/g'
# Output: foo---bar---baz (but breaks with input containing spaces)

# Problematic with special characters:
printf "%s---" foo "bar baz" quux
# Output: foo---bar baz---quux--- (extra trailing separator)

Using Parameter Expansion (Bash/Ksh)

For modern shells with array support:

join_by() {
  local d=${1-} f=${2-}
  if shift 2; then
    printf %s "$f" "${@/#/$d}"
  fi
}

join_by "---" foo "bar baz" quux
# Output: foo---bar baz---quux

POSIX-compliant IFS Approach

Works across most shells:

join_by() {
  sep="$1"
  shift
  old_ifs="$IFS"
  IFS="$sep"
  result="$*"
  IFS="$old_ifs"
  printf '%s\n' "$result"
}

join_by "---" foo "bar baz" quux
# Output: foo---bar baz---quux

Using awk for Complex Cases

When dealing with special characters or large inputs:

join_by() {
  awk -v sep="$1" 'BEGIN {
    for (i=2; i2 ? sep : ""), ARGV[i]
    }
    print ""
  }' "$@"
}

join_by "---" foo "bar baz" "qu'ux"
# Output: foo---bar baz---qu'ux

For large datasets (1000+ items), the awk solution is generally fastest, followed by the IFS approach. The parameter expansion method works well for moderate-sized arrays but may hit command-line length limits.

All presented solutions properly handle:

• Strings containing spaces
• Empty strings
• Special characters (including quotes and backslashes)
• No trailing separators

When working with Unix shell scripts, joining strings with separators seems trivial until you need to handle:

• Strings containing spaces
• Special characters in input
• Portability across different shells

# The sed approach (breaks with spaces)
echo "foo bar baz" | sed 's/ /---/g'
# Output: foo---bar---baz

# IFS trick (limited to single character separators)
join_by() {
  local IFS="$1"
  shift
  echo "$*"
}
join_by --- foo bar baz
# Output: foo---bar---baz

For handling all characters including spaces and special cases:

POSIX-compliant Function

strjoin() {
  sep="$1"
  shift
  first="$1"
  shift
  printf "%s" "$first"
  [ $# -gt 0 ] && printf "%s" "$sep"
  printf "%s" "$@"
}

Using printf with Argument Shifting

join_strings() {
  sep="$1"
  shift
  printf '%s' "$1"
  shift
  printf '%s' "${@/#/$sep}"
}

# Handling paths with spaces
join_by ":" "/path/with spaces" "/another/path" "/normal/path"

# Creating CSV from array
arr=("value 1" "value,2" "value\n3")
join_strings "," "${arr[@]}"

For large datasets (10,000+ items):

• Avoid subshells in loops
• Use built-in string operations when possible
• Consider awk for very large joins

# High-performance awk version
awk -v sep="---" '{for(i=1;i<=NF;i++){if(i>1)printf sep; printf $i} printf "\n"}'