Linux Basic — Array vs String Performance

Array vs String Performance

Arrays are faster for indexed access and avoiding repeated splits. Strings are faster for small one-off operations and use less memory.

---

When Arrays Win

1. Reading many lines

# SLOW — one fork per line
while IFS= read -r line; do
    lines+=("$line")
done < file

# FAST — single builtin
mapfile -t lines < file

mapfile (alias readarray) is a Bash 4+ builtin — one operation to slurp a file into an array.

---

2. Indexed lookups

# SLOW — linear search in string
list="a b c d e"
if [[ "$list" =~ (^| )c($| ) ]]; then

# FAST — O(1) indexed access
list=("a" "b" "c" "d" "e")
echo "${list[2]}"   # c

---

3. Avoiding repeated splits

# SLOW — splits on every access
files="a.txt b.txt c.txt"
for f in $files; do  # splits again each loop
    process "$f"
done

# FAST — split once
files=(a.txt b.txt c.txt)
for f in "${files[@]}"; do
    process "$f"
done

---

When Strings Win

Small datasets

For < 100 items, the memory overhead of an array isn’t worth it.

One-time append

# strings are fine for building once
output=""
output+="line 1\n"
output+="line 2\n"
echo -e "$output"

Serialisation

If you’re passing data to an external command via stdin, a string is simpler:

data="line1\nline2\nline3"
echo -e "$data" | process

---

Memory Considerations

Arrays have per-element overhead. For very large datasets (10k+ items), consider:

• Streaming with while read (constant memory)
• External tools (awk, sort) that handle large files
• Writing to a temp file instead of holding in memory

---

Benchmark

# string append (10k lines): ~2 seconds
result=""
for i in {1..10000}; do
    result+="$i\n"
done

# array (10k lines): ~0.5 seconds
result=()
for i in {1..10000}; do
    result+=("$i")
done

Arrays are faster because each += on a string copies the entire string. Arrays just append a reference.

---

Quick Reference

Task	Use
Read file into list	mapfile -t arr < file
Indexed access	array
Random access by position	array
One-time build + output	string
> 10k items	stream or external tool

---

Next → Profiling with time