I want arithmetic ops which return error instead of losing precision

[safinaskar]

I want lib for monetary calculations. decimal-rs sometimes lose precision, and I don't like that. For example, 1.0000000000000000000000000001 * 1.0000000000000000000000000001 is 1.0000000000000000000000000002 under decimal-rs (this is incorrect from mathematical point of view). (I used 0.1.36.) Even checked_mul gives same answer instead of returning error. I want some function (say, exact_mul and similar names for addition and subtraction), which returns Ok (or Some) when a result can be exactly represented and Err (or None) if not.

In my program (assuming my program is bug-free) such precision losing should never happen. But as we all know we can never be sure that there is no bugs! So I want this exact_mul. I would replace my checked_mul with exact_mul (with unwrap). And if I ever lose precision, I will discover this, because I will see panic

There is similar bug in rust-decimal paupino/rust-decimal#515

[davidli2010]

decimal-rs has 38 digits of precision, and the decimal will be rounded if it overflowed. This is a feature of decimal-rs and we have no plan to add exact_mul.

P.S.: pgnumeric has upto 1000 digits of precision, and 1.0000000000000000000000000001 * 1.0000000000000000000000000001 is 1.00000000000000000000000000020000000000000000000000000001.

[safinaskar]

Ok, you may close the bug