anybody here have Visual Studio .NET?

brc64 · Oct 14, 2003, 05:58 PM

A programmer friend of mine just got .NET from his school and decided to give it a try. After spending almost an entire day trying to figure out a problem he was having in C++, he finall figured out that apparently float is broken in .NET.

If somebody who has a copy of .NET reads this, could you try this simple program?

Code:

#include <iostream>

using namespace std;
int main(){
   float x = 0;
   x += .02;
   x += .02;
   x += .02;

   if(x < .06)
      cout << "wtf?" << endl << x << endl;
   else
      cout << "works fine" << endl;

   return 0;
}

The problem my friend is running into is that after adding .02 a few times the value ends up being around .05999998. double has the same problem. He tried the exact same code in Visual Studio 6 and it worked correctly. I'm trying to figure out if this is just a strange problem that only he has, or if float and double are actually "broken" in .NET (which seems difficult to believe).

- Me

Jeff · Oct 15, 2003, 10:42 PM

I get 0.06. However I am using VS.NET2003, not the original (2002).

brc64 · Oct 16, 2003, 03:47 AM

He tells me he's using the 2003 version as well. It's a very strange problem. I don't have any idea what could be causing it.

- Me

The_Neon_Cowboy · Oct 16, 2003, 05:27 AM

arg ...olny got Visual Studio 6

thats a wierd problem i'd contact micosoft

|-|ahaha · Oct 20, 2003, 04:34 AM

It is not a problem with VS .NET it is a problem with all computers! Simply put: the computer cannot represent real numbers (floating point) without an error. Just to be sure I compiled it under dev-c++ (gcc) and it gave the same kind of behaviour. The problem is atenuated by using double precision (double), but it is still there.

brc64 · Oct 20, 2003, 04:54 PM

Quote:

Originally posted by |-|ahaha
It is not a problem with VS .NET it is a problem with all computers! Simply put: the computer cannot represent real numbers (floating point) without an error. Just to be sure I compiled it under dev-c++ (gcc) and it gave the same kind of behaviour. The problem is atenuated by using double precision (double), but it is still there.

That thought had occured to me, but that doesn't explain why the same program works perfectly fine in Visual Studio 6. Obviously SOMETHING is different between VS6 and VS.NET, but we can't figure out what.

- Me

UberLord · Oct 20, 2003, 05:20 PM

Interesting.

GCC 3.3.1 reports

Code:

wtf?
0.06

If it really holds a value other than 0.06 then the cout float is borked or the boolean evaluation of x < 0.06 is borked.

I also suffer similar problems when using reading floats from SQL server and storing them in doubles in VB 6 (vb doesn't know what a float is) so I think that the assumption that computers hate floats is correct

FluffyChicken · Oct 20, 2003, 10:10 PM

A quick serch got me this http://www.codeguru.com/mfc/comments/4387.shtml
(using google and Visual Studio .NET float problem)
Which shows the problem is there, not just with you.

Go here for (hopefuly answers)http://msdn.microsoft.com/support/
or try here as well
Newsgroups: microsoft.public.dotnet.languages.vc

bocybo · Mar 5, 2004, 11:05 AM

I know this is an *old* thread, but this is a very commonly come across problem. When comparing a double(or float in this case) to a number, be it 0, 5, or .06 you always need to have an epsilon value to counteract the floating point inaccuracy. So you could do this:

if((x<e+.06)&&(x>.06-e)) say e is like .0001
...

This should fix the incarrucacy. It is definatly not just .net, but has to do with the way that the PC stores floating point digits.

Oct 14, 2003, 05:58 PM	#1
brc64 Zeeky H. Bomb Join Date: May 2002 Location: Evansville, IN Posts: 826 Rep Power: 0	anybody here have Visual Studio .NET? A programmer friend of mine just got .NET from his school and decided to give it a try. After spending almost an entire day trying to figure out a problem he was having in C++, he finall figured out that apparently float is broken in .NET. If somebody who has a copy of .NET reads this, could you try this simple program? Code: #include <iostream> using namespace std; int main(){ float x = 0; x += .02; x += .02; x += .02; if(x < .06) cout << "wtf?" << endl << x << endl; else cout << "works fine" << endl; return 0; } The problem my friend is running into is that after adding .02 a few times the value ends up being around .05999998. double has the same problem. He tried the exact same code in Visual Studio 6 and it worked correctly. I'm trying to figure out if this is just a strange problem that only he has, or if float and double are actually "broken" in .NET (which seems difficult to believe). - Me

Oct 15, 2003, 10:42 PM	#2
Jeff Yarr... I be blind! Join Date: May 2002 Location: Calgary, Canada Posts: 3,191 Rep Power: 80	I get 0.06. However I am using VS.NET2003, not the original (2002). __________________

Oct 16, 2003, 05:27 AM	#4
The_Neon_Cowboy HardwareHeaven Extreme Member Join Date: Dec 2002 Location: U.S.A. Posts: 16,009 Rep Power: 92 System Specs	arg ...olny got Visual Studio 6 thats a wierd problem i'd contact micosoft __________________

Oct 20, 2003, 05:20 PM	#7
UberLord A Legend in Underwear Join Date: May 2002 Location: Unknown Posts: 5,255 Rep Power: 0	Interesting. GCC 3.3.1 reports Code: wtf? 0.06 If it really holds a value other than 0.06 then the cout float is borked or the boolean evaluation of x < 0.06 is borked. I also suffer similar problems when using reading floats from SQL server and storing them in doubles in VB 6 (vb doesn't know what a float is) so I think that the assumption that computers hate floats is correct __________________ Gentoo Linux - Developer (baselayout) Read my blog "I contend that we are both atheists. I just believe in one fewer god than you do. When you understand why you dismiss all the other possible gods, you will understand why I dismiss yours." Stephen Roberts

Oct 20, 2003, 10:10 PM	#8
FluffyChicken DriverHeaven Addict Join Date: Nov 2002 Location: England Posts: 320 Rep Power: 0	A quick serch got me this http://www.codeguru.com/mfc/comments/4387.shtml (using google and Visual Studio .NET float problem) Which shows the problem is there, not just with you. Go here for (hopefuly answers)http://msdn.microsoft.com/support/ or try here as well Newsgroups: microsoft.public.dotnet.languages.vc __________________ [COLOR=#455b64]Team mauisun.org [/COLOR]

Oct 16, 2003, 03:47 AM	Thread Starter #3
brc64 Zeeky H. Bomb Join Date: May 2002 Location: Evansville, IN Posts: 826 Rep Power: 0	He tells me he's using the 2003 version as well. It's a very strange problem. I don't have any idea what could be causing it. - Me

Oct 20, 2003, 04:34 AM	#5
\|-\|ahaha DriverHeaven Newbie Join Date: Oct 2003 Posts: 2 Rep Power: 0	It is not a problem with VS .NET it is a problem with all computers! Simply put: the computer cannot represent real numbers (floating point) without an error. Just to be sure I compiled it under dev-c++ (gcc) and it gave the same kind of behaviour. The problem is atenuated by using double precision (double), but it is still there.

Mar 5, 2004, 11:05 AM	#9
bocybo DriverHeaven Junior Member Join Date: Jul 2002 Posts: 89 Rep Power: 0	I know this is an old thread, but this is a very commonly come across problem. When comparing a double(or float in this case) to a number, be it 0, 5, or .06 you always need to have an epsilon value to counteract the floating point inaccuracy. So you could do this: if((x<e+.06)&&(x>.06-e)) say e is like .0001 ... This should fix the incarrucacy. It is definatly not just .net, but has to do with the way that the PC stores floating point digits. __________________ Barton 2500 @ 11.0x215 Epox 8RDA+ 1GB Corsair PC3200 ATI Radeon 9800 Pro @ 420/380 SB Audigy Lite-on 52x32x52 Maxtor 120GB ATA133