parseFloat function seems to have a bug when summing? help!

[ajordan]

hello!
i think i have a bug when i use parseFoat function:

i have the following operation :

parseFloat("10.664062") + parseFloat("10.664062")
+parseFloat("10.664062");

the result of following statement must be 31.992186. Does somebody
explain to me why the result returned is 31.992185999999997.

strange no?

thanks a lot for your help.

[Tim Fooy]

No, it is not a bug. The reason for this result is the fact that the
internal representation of a float (which follows IEEE 754 standard) is
limited to 7 correct digits, because of the limitation to 32 bits. See e.g.
http://research.microsoft.com/~holla...ieeefloat.html (that
looks correct to me, i haven't read it all)

If you want a higher precision, you'll have to use a double precision
floating point variable, but i don't know whether this exists in javascript.

Tim

[DrewM]

Tim Fooy wrote:

No, it is not a bug. The reason for this result is the fact that the
internal representation of a float (which follows IEEE 754 standard) is
limited to 7 correct digits, because of the limitation to 32 bits. See e.g.
http://research.microsoft.com/~holla...ieeefloat.html (that
looks correct to me, i haven't read it all)

If you want a higher precision, you'll have to use a double precision
floating point variable, but i don't know whether this exists in javascript.

You can use the toFixed() method to convert a Number object to a fixed
number of decimal places.

e.g.

(1.159).toFixed(2)

would return 1.16

ref: http://tinyurl.com/3ys5x
drew.

[Lasse Reichstein Nielsen]

"Tim Fooy" <th*****@invalid.com> writes:

No, it is not a bug. The reason for this result is the fact that the
internal representation of a float (which follows IEEE 754 standard) is
limited to 7 correct digits, because of the limitation to 32 bits.
Actually, Javascript numbers are 64 bit (double precission) IEEE 754
floating point numbers, so it has 53 significant bits. The problem is
not the number of significant digits itself, but merely the fact that it
is finite. There are some numbers that look simple in base ten that
cannot be represented exactly in base two. The small error on the
53rd bit accumulates when you do arithmetic, until it becomse visible.

Try something as simple as
alert(0.1 + 0.1 + 0.1);
which alerts 0.30000000000000004 in my browser.
If you want a higher precision, you'll have to use a double precision
floating point variable, but i don't know whether this exists in javascript.

It does, it's all there is.
/L
--
Lasse Reichstein Nielsen - lr*@hotpop.com
DHTML Death Colors: <URL:http://www.infimum.dk/HTML/rasterTriangleDOM.html>
'Faith without judgement merely degrades the spirit divine.'

[Dr John Stockton]

JRS: In article <40***********************@news.skynet.be>, seen in
news:comp.lang.javascript, Tim Fooy <th*****@invalid.com> posted at Wed,
28 Apr 2004 13:25:13 :

No, it is not a bug. The reason for this result is the fact that the
internal representation of a float (which follows IEEE 754 standard) is
limited to 7 correct digits, because of the limitation to 32 bits. See e.g.
http://research.microsoft.com/~holla...ieeefloat.html (that
looks correct to me, i haven't read it all)

If you want a higher precision, you'll have to use a double precision
floating point variable, but i don't know whether this exists in javascript.

What you write would be true if Numbers in javascript were represented
by IEEE Singles. They are represented by IEEE Doubles.

Although the OP only gets 7 digits as desired, the error is actually of
the order of 1 in 1E16.
I cannot explain why the OP gets 31.992185999999997 ; ISTM that one gets
31.992185999999996.

The OP could have tested +"10.664062" + +"10.664062" + +"10.664062" or
10.664062 + 10.664062 + 10.664062 or 3*10.664062 all of which
give me 31.992185999999996 - which shows that the matter has nothing
to do with parseFloat.

The OP should have read the FAQ, specifically Sec 4.7, and what it links
to and/or my site and/or many others.

Try 0.2 + 0.4 and 0.1 + 0.7.

--
© John Stockton, Surrey, UK. ?@merlyn.demon.co.uk Turnpike v4.00 IE 4 ©
<URL:http://jibbering.com/faq/> Jim Ley's FAQ for news:comp.lang.javascript
<URL:http://www.merlyn.demon.co.uk/js-index.htm> jscr maths, dates, sources.
<URL:http://www.merlyn.demon.co.uk/> TP/BP/Delphi/jscr/&c, FAQ items, links.

[Randy Webb]

DrewM wrote:

You can use the toFixed() method to convert a Number object to a fixed
number of decimal places.

e.g.

(1.159).toFixed(2)

would return 1.16

And then wonder why you get errors for .007 ? Its well documented in
the c.l.j FAQ, http://www.jibbering.com/faq/#FAQ4_6
--
Randy
Chance Favors The Prepared Mind
comp.lang.javascript FAQ - http://jibbering.com/faq/