C / C++ / MFC

12-May-05 20:53

To display the ranges of values an unsigned _int64 can have,this is what I wrote:

void myDlg::ShowRanges()
{
unsigned _int64 uValue;
CString uValueStr;
uValue=(unsigned _int64)pow(2,sizeof(unsigned _int64)*8) ;
////////////////////////////////////////////////////////
uValueStr.Format("0 to %I64u",uValue-1);
AfxMessageBox(uValueStr);
}
///////////////////////////////////////////
This is displaying:
0 to 9 223 372 036 854 775 807
instead of the correct answer:
0 to 18 446 744 073 709 551 614

Obviously the answer is truncated by some statment.But where ? In the formatting perhaps?
Thanks for any help.
NB: I added the spaces between the digits to facilitate reading.

Graham Bradshaw12-May-05 21:18

Graham Bradshaw

12-May-05 21:18

I see two problems here:

1) pow is a floating point function, returning a double, and doubles won't maintain precision to the 20 digits you are expecting.

2) it wouldn't work anyway. You can't represent the range of an integer type as you are doing. For example, think about a byte. That has a sizeof 1, but the range is not 0 to 256, but 0 to 255. Note that 256 will not fit in a byte data type.

The range is actually 0 to 2 to the power of (number of bits) - 1

cmk12-May-05 21:31

cmk

12-May-05 21:31

If you look in limits.h

#if _INTEGRAL_MAX_BITS >= 64
/* minimum signed 64 bit value */
#define _I64_MIN (-9223372036854775807i64 - 1)
/* maximum signed 64 bit value */
#define _I64_MAX 9223372036854775807i64
/* maximum unsigned 64 bit value */
#define _UI64_MAX 0xffffffffffffffffui64
#endif

Sooooo,
uValueStr.Format("0 to %I64u", _UI64_MAX);

...cmk

Save the whales - collect the whole set

fardak12-May-05 23:09

12-May-05 23:09

And that works perfectly.
Thank you so much for your time.

PJ Arends13-May-05 9:45

13-May-05 9:45

You could also do :

unsigned _int64 uValue = (unsigned _int64)~0;

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

"Obviously ??? You're definitely a superstar!!!" mYkel - 21 Jun '04

Within you lies the power for good - Use it!

Honoured as one of The Most Helpful Members of 2004

fardak13-May-05 21:39

13-May-05 21:39

You are right.
I wrote :
void CMyHexDDlg::ShowRangeOfULONGLONG()
{unsigned _int64 uValue = (unsigned _int64)~0;
CString result;
result.Format("0 to %I64u",uValue);
AfxMessageBox(result);
}
and the output is ( as expected ) :
0 to 18446744073709551615

Very elegant ,and fast solution.
Thank you so much for your time.

PJ Arends13-May-05 22:08

13-May-05 22:08

I use this little template class:

template <typename T>
class Limits
{
public:
   static T min() { return (T)~0 > 0 ? 0 : (T)1 << (sizeof(T) * 8 - 1); }
   static T max() { return (T)~0 > 0 ? (T)~0 : (T)~0 & ~((T)1 << (sizeof(T) * 8 - 1)); }
};
 
// example usage
 
unsigned _int64 ui64Max = Limits<unsigned _int64>::max();
_int64 i64Min = Limits<_int64>::min();
 
char cMin = Limits<char>::min();

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

"Obviously ??? You're definitely a superstar!!!" mYkel - 21 Jun '04

Within you lies the power for good - Use it!

Honoured as one of The Most Helpful Members of 2004

fardak13-May-05 23:01

13-May-05 23:01

This is a new subject to me.
My hint is taht the class you are defining can be used for all kind of data types(unsigned and signed) to find the minimum and maximum values that can be stored in them.
But if you have time,could you please tell me what header file I must include to use your class ?
Thanks for any help.

PJ Arends14-May-05 8:03

14-May-05 8:03

fardak wrote:
could you please tell me what header file I must include to use your class ?

It is plain C++, no header file required.

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

"Obviously ??? You're definitely a superstar!!!" mYkel - 21 Jun '04

Within you lies the power for good - Use it!

Honoured as one of The Most Helpful Members of 2004

fardak14-May-05 20:16

14-May-05 20:16

I copied the template and class Limits and insert it (as is) in one of my header files of my code
(D:\myPrograms\myHexD\MyDTypeDlg.h)
,and the compiler displayed this list of warnings and errors.(By the way I am using VC++6.0).

warning C4003: not enough actual parameters for macro 'min'
D:\myPrograms\myHexD\MyDTypeDlg.h(17) : warning C4003: not enough actual parameters for macro 'max'
D:\myPrograms\myHexD\MyDTypeDlg.h(15) : error C2059: syntax error : 'function-style cast'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(15) : error C2334: unexpected token(s) preceding ':'; skipping apparent function body
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2143: syntax error : missing ')' before '}'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2059: syntax error : ')'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2143: syntax error : missing ';' before '}'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2238: unexpected token(s) preceding ';'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(17) : error C2143: syntax error : missing ';' before 'static'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
myHexDDlg.cpp
D:\myPrograms\myHexD\MyDTypeDlg.h(15) : warning C4003: not enough actual parameters for macro 'min'
D:\myPrograms\myHexD\MyDTypeDlg.h(17) : warning C4003: not enough actual parameters for macro 'max'
D:\myPrograms\myHexD\MyDTypeDlg.h(15) : error C2059: syntax error : 'function-style cast'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(15) : error C2334: unexpected token(s) preceding ':'; skipping apparent function body
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2143: syntax error : missing ')' before '}'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2059: syntax error : ')'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2143: syntax error : missing ';' before '}'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(16) : error C2238: unexpected token(s) preceding ';'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled
D:\myPrograms\myHexD\MyDTypeDlg.h(17) : error C2143: syntax error : missing ';' before 'static'
D:\myPrograms\myHexD\MyDTypeDlg.h(19) : see reference to class template instantiation 'Limits<t>' being compiled

Thanks for any help.

PJ Arends15-May-05 10:25

15-May-05 10:25

Sorry, but that is my fault. The functions are named Min and Max, I forgot to capitalize them when I typed them in here. In lower case the way they are now, the compiler substites the function names for the min and max macros defined in WinDef.h so it is trying to compile a line that looks like this:

static T (((a) < (b)) ? (a) : (b)) { return (T)~0 > 0 ?...

and it can not find the variables a and b.

So capitalize min() and max() to Min() and Max() and it will work fine.

Sorry about that.

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

"Obviously ??? You're definitely a superstar!!!" mYkel - 21 Jun '04

Within you lies the power for good - Use it!

Honoured as one of The Most Helpful Members of 2004

fardak16-May-05 1:58

16-May-05 1:58

Pls don't worry.
This is what I wrote and got the (correct) result:
void CMyHexDDlg::Temp()
{ CString result;
unsigned _int64 ui64Max=Limits<unsigned _int64="">::Max();
result.Format("Max:%I64u",ui64Max);
AfxMessageBox(result);
_int64 i64Min=Limits<_int64>::Min();
result.Format("Min:%I64i",i64Min);
AfxMessageBox(result);
char cMin=Limits<char>::Min();
result.Format("Min:%i",cMin);
AfxMessageBox(result);

}
///////////////////////////////////////////
The output of the three message boxes is:
Max: 18446744073709551615
Min: -9223372036854775808
Min:-128

The code is very subtle and very interesting.
May I add that you are a supestar alright.
Thank you so much for your time and effort.

fardak16-May-05 23:05

16-May-05 23:05

I have difficulty with one part of this statement and I hope you give me a hand there.
static T Max()
{ return (T)~0 > 0 ? (T)~0 : (T)~0 & ~((T)1 << (sizeof(T) * 8 - 1)); }
/////////////////////////////////////////
Say I have a WORD:
00000000 00000000
If it is a UWORD (HOB set to 1) return
11111111 11111111
If it is a WORD (HOB set to -)
transform it to
11111111 11111111
and AND it with what exactly ?
Bearing in mind that: 1<< (sizeof(T)*8-1) is:
10000000 00000000
Thanks for any help.

PJ Arends17-May-05 6:49

17-May-05 6:49

http://www.codeproject.com/cpp/bitbashing.asp[^]

In a signed value the high order bit(HOB) is the sign bit, it is zero for positive numbers and one for negative numbers. In unsigned values the HOB has no special significance.

So what is done is to first check if we are dealing with a signed or unsigned value by checking if ~0 (inverse zero) is greater than zero. For signed values ~0 will give you -1 (minus one), for unsigned values it will give you the largest possible value.

So for unsigned values we can simply return ~0.

But for signed values, we have to clear the HOB (set it zero) to get the largest possible value

ie:

BYTE bMax = Limits<BYTE>::Max();

will do the following calculations:

(T)~0 > 0 ?
   00000000 inversed is 11111111
   is 11111111 greater than zero?
   No, it is minus one, so it is less than zero.

So we have to return: (T)~0 & ~((T)1 << (sizeof(T) * 8 - 1))
   00000000 inversed is 11111111

   00000001 left shifted 1 * 8 - 1 (or 7) is 10000000
   10000000 inversed is 01111111

   11111111 bitwise and 01111111 is 01111111

So we can return 01111111, the largest possible signed value.

And now that I have gone through the exercise of explaining it this way, I realize that I can simplify the function a little. The last step with the bitwise and can be eliminated. The function can be

static T Max() { return (T)~0 > 0 ? (T)~0 : ~((T)1 << (sizeof(T) * 8 - 1)); }

Thanks for putting me through this exercise Big Grin | :-D

"You're obviously a superstar." - Christian Graus about me - 12 Feb '03

"Obviously ??? You're definitely a superstar!!!" mYkel - 21 Jun '04

Within you lies the power for good - Use it!

Honoured as one of The Most Helpful Members of 2004

fardak17-May-05 21:47