> It's usually only needed when adding billions of values together and the accumulated truncation errors would be at an unacceptable level. OTOH, it’s easy to

> It's usually only needed when adding billions of values together and the accumulated truncation errors would be at an unacceptable level. OTOH, it’s easy to implement, so I have a couple of functions to do it easily, and I got quite a lot of use out of them. It’s probably overkill sometimes, but sometimes it’s useful.