0

I have a c++ ledger application in which floating point is used for calculations, now what should I do to convert to fix point arthimatic (considerably upto 4 digits past decimal point) without generating more bugs in the program. Any step wise process which I should adopt or tips to prevent errors? Please suggest some test cases which will be helpful

user3172833
  • 37
  • 6
  • 2
    Creating a suite of unit tests to cover all the current calculations in your application would be a good start. It would be a good to be able to swap between fixed and floating point easily in this scenario, so that you can compare results easily. – Simon Bosley Mar 12 '14 at 10:38

1 Answers1

1
  1. Introduce a type Currency used in relevant computations (if not done already)
  2. Make sure all relevant numbers are stored as Currency, not as double, or float
  3. Define Currency with a fixed-point real type. You can use existing implementations, like CodeF00's numeric::Fixed. See also What's the best way to do fixed-point math?
Community
  • 1
  • 1
Mikhail
  • 20,685
  • 7
  • 70
  • 146
  • Please suggest tests and tips to avoid performance issues as well.:) – user3172833 Mar 13 '14 at 03:22
  • I looked at your reference, sorry to be naive but it says that according to your architecture(bitsize) you can typedef fixed class. For eg typedef fixed<16,16> fixed. Will it affect if I shift my data to different architectures.Any way to keep it uniform. – user3172833 Mar 13 '14 at 03:56
  • Test tips are very application-specific. Cannot say anything non-obvious having no information on your case. Concerning your second comment - don't hesitate to use question marks. – Mikhail Mar 13 '14 at 07:01
  • Well I have used GMP lib for a double entry ledger system. I agree that you will need to know more features of the system to describe the tests. But a test which proves whether error exist due to current floating point arthimatic will be fine. I wanted tips on how is performance affected if i use fixed point arithematic on top of GMP vs MPFR of GMP. – user3172833 Mar 18 '14 at 12:03