func subscript(x uint64) string {
	const w = len("₀") // all digits 0...9 have the same utf8 width
	var buf [32 * w]byte
	i := len(buf)
	for {
		i -= w
		utf8.EncodeRune(buf[i:], '₀'+rune(x%10)) // '₀' == U+2080
		x /= 10
		if x == 0 {
			break
		}
	}
	return string(buf[i:])

	// For %#U we want to add a space and a quoted character at the end of the buffer.
	if f.sharp && u <= utf8.MaxRune && strconv.IsPrint(rune(u)) {
		i--
		buf[i] = '\''
		i -= utf8.RuneLen(rune(u))
		utf8.EncodeRune(buf[i:], rune(u))
		i--
		buf[i] = '\''
		i--
		buf[i] = ' '
	}
	// Format the Unicode code point u as a hexadecimal number.
	for u >= 16 {
		i--
		buf[i] = udigits[u&0xF]
		prec--

		if b.Flush(); b.err != nil {
			return 0, b.err
		}
		n = b.Available()
		if n < utf8.UTFMax {
			// Can only happen if buffer is silly small.
			return b.WriteString(string(r))
		}
	}
	size = utf8.EncodeRune(b.buf[b.n:], r)
	b.n += size
	return size, nil
}

// WriteString writes a string.
// It returns the number of bytes written.
// If the count is less than len(s), it also returns an error explaining