Today i learned a bit about how floating point numbers are expressed binary. Well, i can see why the IEE-754 technical standard is smart for storing a huge range of numbers using relatively few bits. Also i see why it's useful to sometimes to express each symbol as four 4-bits(Like you can do i Cobol, right?), when you want very precise numbers.
Anyways, my question is, isn't it more convenient to "just" store floating point numbers as two integers? For example, you could express the number before the comma as a 34 bit integer and the number after the comma as a 30 bit integer. Then'd you be able to write any number in the range of +-8,5 billion with approximately 9 decimals, right?
It's still a relatively small range of numbers compared to double precision floating points, but as far as i can see it's better than the Cobol way?
Thank you.