Unicode
Simple explanation
Source files are bytes; humans think in characters and lines. The runtime keeps byte offsets exact (everything indexes bytes) while tracking row/column points and decoding UTF-8 only where character meaning matters.
Technical explanation
unicode/utf8.zig implements strict decoding (rejecting truncated, overlong, surrogate, and out-of-range sequences) plus encoding, code-point counting, and continuation-byte helpers. unicode/tables.zig classifies whitespace/letter/digit/word code points beyond ASCII. Points advance per byte with \n starting a new row, matching reference-runtime column semantics. The lexer substitutes the replacement character on invalid bytes and continues rather than failing the parse.