> Instead, you'd make the trie of an encoding, probably UTF-8 There are some new implications you need to account for if you do this, as you no longer have one

> Instead, you'd make the trie of an encoding, probably UTF-8 There are some new implications you need to account for if you do this, as you no longer have one node per letter, but a path per letter. E.g. subtrees will no longer correctly represent substrings, making autocompletion slightly trickier.