Number Formatter Question

[David Dickinson]

Hi,

I have a NumberFormatter set up with Maximum and Minimum Fraction Digits set to 2.

Everythings worked well until my latest test where for some reason when I print the output to the screen my double (after being formatted) appears as 0.00 when it should really be 9.094947017729282E-13 this also happens when the number is negative -9.094947017729282E-13

Does anyone know why this is? I've tried altering the fractional digits but this doesn't solve it and I don't want any more than two digits following the decimal point.

Thanks!

[Mike Gershman]

9.094947017729282E-13 to 2 decimal places is 0.00.

You are expecting NumberFormatter to give you scientific notation? I didn't know it could do that. Please show your code.

[David Dickinson]

Originally posted by Mike Gershman:
You are expecting NumberFormatter to give you scientific notation? I didn't know it could do that. Please show your code.

I didn't expect the result in scientific notation, is this performed automatically by the double type when a value becomes to long? Do you know of any way to force it to store the numeric value?

Thank you!

[Jeroen Wenting]

The formatted number will be 0.00 when rounded to two decimal places if it's between -0.0049* and 0.0049* .
The UNformatted number (so your double) doesn't change.

You can't have it both ways, either you get it printed rounded to two decimal places or you don't...

[David Dickinson]

Well in my code I evaluate the double:

if (res > 0) {}
else if (res < 0) {}
else {}

But because the actual value of the double is 0.000000000000001 (or something very similar) it falls into the second condition which I don't want it to.

Can you think of any way of stopping this? In my JUnit tests I just set the delta to 0.01 which effectively skips any minute differences.

Basically i'd be happier if the value was .01 different from 0 for it to fall into either condition.

So am I right in thinking the test simply needs changing to:

if (res > 0.01) {}
else if (res < -0.01) {}
else {}

[Layne Lund]

A common way to compare doubles is to use a delta value (as you see in JUnit). You can implement this same logic with something like this:

if (Math.abs(expected - actual) < DELTA) {
  // expected and actual close enough to be considered "equal"
} else {
  // expected and actual differ enough to be considered "not equal"
}

I take it from your code above, that you actually have three conditions: zero, negative, and positive. You may want to use something like the logic I gave above to test for "equality" with zero first and then test for positive and negative values.

HTH

Layne
[ December 08, 2004: Message edited by: Layne Lund ]

[David Dickinson]

Layne,

Only problem with that code is I don't know what the value is going to be, the method exists to determine whether the value is positive/negative and act accordingly.

I've tried understanding the purpose of Math.abs but failed miserably.

I would have thought it would be easy to compare a double (with a fractional precision of 2) with zero?

Can anyone offer any advice? I'm not particuarly strong with maths and can't think of any formula which I could use

[David Dickinson]

I should add that the reason I thought I could use the following expression is res represents a currency figure and therefore anything less than a penny or 0.01 really isn't worth worrying about for this program.

if (res < -0.009) {}
else if (res > 0.009) {}
else {}

So would this code be suitable or do you have a more suitable alternative?
[ December 09, 2004: Message edited by: David Dickinson ]