Forums Feature Suggestions More exactness in numbers close to 0

mIRC can handle numbers up to 1 * (10 ^ 4932)? But, going closer to zero, mIRC only goes to 1 * (10 ^ -6) = 0.000001. I'd *LOVE* to have mIRC handle numbers close to zero bette, the better to calcualte me geometric means. And it seems inconsistant that mIRC handles numbers so large but doesn't even come close to being able to handle numbers closer to 0 well.

Hmmm. Yeah, I am pretty sure IEEE 754 floating point numbers have a special additions and a bit or so specify denormalized values. So the for numbers between 0 and epsolin where episolin is the standard difference between two . Mostly to prevent rounding to zero during division. Although, I think the suggestion is good, It might take some coaxing to do. Though, frankly I have always thought that mIRC numbers should actually map onto C/C++ numbering directly, and dynamicly cast like LISP. But, then again, making it OO might be nice too. Yeah, denormalize values are a pretty good idea.

Actually, mIRC seems to be capable of only 19 digits for whole numbers, and 20 digits (including the '.') for floating points, but only up to 6 decimal positions.

9999999999999999999
999999999999999999.9
99999999999999999.99
9999999999999999.999
9999999999999999.999
999999999999999.9999
99999999999999.99999
9999999999999.999999

Beyond these, it starts rounding dropping the lest significant number.

- Raccoon

mIRC to my knowledge doesn't even come close to following IEEE Floating Point.