Strings are not just arrays of characters

Strings look like character arrays and behave differently in three ways that cause real bugs.

They are immutable

Every "modification" builds a new string. Inside a loop that is quadratic:

// O(n²): each += copies the whole string so far.
let out = '';
for (const ch of text) out += transform(ch);

// O(n): collect, then join once.
const parts = [];
for (const ch of text) parts.push(transform(ch));
const result = parts.join('');

Modern engines optimise some += patterns with ropes, but the array-and-join version is the one that is portable and predictable. In Java it is StringBuilder; in Python ''.join(parts); in Go strings.Builder.

A character is not a code unit

JavaScript strings are UTF-16. Anything outside the basic plane — emoji, many CJK extensions — occupies two code units, and length counts units, not characters.

'😀'.length;             // 2
'😀'[0];                 // half a surrogate pair, meaningless alone
[...'😀'].length;        // 1 — the spread iterates code points

Reversing a string with split('').reverse().join('') corrupts every emoji in it. Use [...text] when correctness for real user input matters.

Comparison is not always byte comparison

Accented characters can be composed of a base plus a combining mark, or a single precomposed code point. The two render identically and are not ===. Normalise before comparing user text:

'é'.normalize('NFC') === 'é'.normalize('NFC');

For case-insensitive comparison, toLowerCase() is locale-dependent for a handful of languages — Turkish dotless i being the standard example.

Costs to keep in mind

Two anagrams share a sorted-character key, and grouping anagrams is exactly a hash map keyed on that. Which is the next chapter's territory.