asked on

round un number

Hello
here is my code to round un number

/********************************************************/
// Round : round(8.56475, 2) => 8.56
/********************************************************/
function roundFloat( value, n_precision ) {
      var tem = 1.*Math.pow(10, n_precision);
      var retval = Math.round(1.*value*tem)/tem;
      return retval;
}

It works well but I have one problem following

roundFloat( 256.905, 2 ) = 256.9 (false) instead 256.91 (true)

I dit verify and I see the reason : Javascript calculs "1.*256.905*100 = 25690.499999999996" (false) instead 25690.5

I dont know how I can correct it ?
thanks a lot
best regards
Hai

Rakafkaven

Javascript treats numbers as doubles, which can result in some bizarre miscalculations that aren't your fault at all. Silly as it seems, adding a large arbitrary number (say 50000) and then subtracting it again after the operation seems to fix the issue. Make your number large enough that it won't interfere with any conceivable rounding your function may do.

function roundFloat( value, n_precision ) {
value = value + 50000;
var tem = 1.*Math.pow(10, n_precision);
var retval = (Math.round(1.*value*tem)/tem) - 50000;
return retval;
}

ASKER CERTIFIED SOLUTION

Zvonko

membership

This solution is only available to members.

To access this solution, you must be a member of Experts Exchange.

Start Free Trial

ndhai

ASKER

hello Rakafkaven
thanks but It doestn't work :((
when I tried alert( roundFloat(256.905, 2) );, it showed 256.910000000149
Hai

ndhai

ASKER

>> alert( 256.905.toFixed(2) )

excellent response
thanks a lot
Hai

Rakafkaven

Not only do I waste power, the method I used doesn't work, since it the floating point operations still occur on the item. Rather than trying to avoid the operations, kicking them into an integer with a second Math.round and an additional decimal shifting operation works consistently. Of course, it wastes even MORE power, but it does return the answer without the potential for padding that toFixed has.

function roundFloat( value, n_precision ) {
value = value;
var tem = 1.*Math.pow(10, n_precision);
var retval = Math.round(Math.round(value*tem*100)/100)/tem;
return retval;
}

Zvonko

If you want to strip the zeros, then check this:

alert( roundFloat(256.905,2) )

function roundFloat( value, n_precision ) {
return value.toFixed(n_precision)*1;
}

Rakafkaven

If I didn't know better, I'd think that you knew what you were doing. :)

Zvonko

OK, there is nothing stripped in above number, but you know what I mean :)