Question

Possible Duplicate:
Is JavaScript’s Math broken?

I wrote some simple C# code that runs Python code dynamically (already implemented):

string code = @"100 * 2 + 4 / 3";
ScriptEngine engine = Python.CreateEngine();
ScriptSource source = engine.CreateScriptSourceFromString(code, SourceCodeKind.Expression);
int res = source.Execute<int>();
Console.WriteLine(res);

And then I thought about Javascript, and that there are core differences between C# and JS. For example:

In JS:

var t=1.02+1.01 = 2.0300000000000002;

And then I tried this via Jint:

var script = @"
  function add( ) {
    return 1.02 + 1.01;
  };
  return add();";
  var result = new JintEngine().Run(script);
  Console.WriteLine(result);

The result was:

Maybe I don’t see the whole picture, but if a programmer on the other side of the world sends me his script file, I (and him) expect the result to be consistent! (Let’s ignore the problematic base 2 representation for now, I’m talking about consistency).

If I was mistaken, in what scenario would I use running other code on .Net? (I will have to be very very suspicious for every line of code…)

Am I right ?

another Example :

   var script = @"
                               function show( )
                               {
                                      return  parseInt('123asd'); //in js it's 123
                               };
                               return show();";

                var result = new JintEngine().Run(script);
                Console.WriteLine(result);

result :

How can I trust a script to yield the same expected result ? ( or am I getting this whole situation wrong…?)

Answer 1

I don’t see any “inconsistency”. That’s just how floating point numbers work!

1) The value – by definition – is seldom “exact”

2) The representation (e.g. from Printing out to a string) can be totally misleading if you try to print out more digits than your value has precision 🙂

Sample C code

#include <stdio.h>

int 
main ()
{
  double d = 1.02 + 1.01;
  float f = 1.02 + 1.01;
  printf ("d=%lf, f=%f\n", d, f);
  printf ("d=%25.20lf, f=%25.20f\n", d, f);
  return 0;
}

Sample output:

d=2.030000, f=2.030000
d=   2.03000000000000020000, f=   2.02999997138977050000

Excellent article:

• What Every Computer Scientist Should Know About Floating Point Arithmetic

Finally, please read this discussion:

• Is floating point math broken?

The problem is NOT Python, it’s NOT C#, it’s NOT Javascript.

I assure you 🙂