If the memory layout is fixed and fields are untyped then every field must be at least 8 bytes to potentially hold a double precision floating point value. Ther

If the memory layout is fixed and fields are untyped then every field must be at least 8 bytes to potentially hold a double precision floating point value. There would clearly be value in adding typing to restrict field values to 1 or 2 or 4 byte integers to allow packing those fields. But I can see that it would add complexity.