C / C++ / MFC

3-Sep-06 3:14

This is the same with any data type, when 1 large number is added to another and the result is larger than the data type can hold, it wraps around. i.e.

ULONG size1 = 0xFFFFFF00;<br />
ULONG size2 = 0xFFFFFF00;<br />
size1 += size2; // == 0xFFFFFE00

The output is less than the input. I want to check for this problem and if it's there I would prefer the output to be the max value the data type can hold. What is the best way to approach this?

Ahmed Ismail Mohamed3-Sep-06 3:29

Ahmed Ismail Mohamed

3-Sep-06 3:29

waldermort wrote:
ULONG size1 = 0xFFFFFF00;
ULONG size2 = 0xFFFFFF00;

if you divede both size1 and size by a big number that make you sure that you can add and you still in the range of long.
size1 = size1/n;
size2 = size2/n;

waldermort wrote:
size1 += size2; // == 0xFFFFFE00

adding in this line will not make a problem

then finally, when you want to output the final value , make the value the program do it time the number you devide by it:
ex: textbox1.text = Tostring(size2 * n) // n is the number you devided by it.

I hope it work properly with you.
regards; Smile | :)

** Ahmed Ismail **

Waldermort3-Sep-06 3:42

3-Sep-06 3:42

That doesn't work, it produces exactly the same result as simply adding them:

ULONG max1 = 0xFFFFFF00;<br />
ULONG max2 = 0xFFFFFF00;<br />
max1 = max1/4;<br />
max2 = max2/4;<br />
max2 += max1;<br />
max2 *= 4; // == 0xFFFFFE00

The only way I have found is to cast them to a larger data type and check the value using an if else block

if ( ((UINT64)max1 + (UINT64)max2) > 0xFFFFFFFF )<br />
    max2 = 0xFFFFFFFF;<br />
else<br />
    max2 += max1;

But I'm sure there must be a more accurate way of doing it.

Waldermort3-Sep-06 8:26

3-Sep-06 8:26

Ahmed Ismail Mohamed wrote:
i think that you should make max2 *4 only when you output the resualt.

What you are saying makes no sense. Why divide and add when it is taking the result I want further away.

Nader Elshehabi3-Sep-06 4:40

Nader Elshehabi

3-Sep-06 4:40

Hello

Well, if you have some assembly background you'd remember that there is a bit in the status register called overflow bit to check for this error. Anyway, I know how to check for overflow in Assembly and C#, but sorry, my C++ is too rusty to remember how to do it in C++ Big Grin | :-D

.

You are right, there is a better way to detect overflow, yet I will provide you with a simpler way of detecting it. Usualy the overflow result is smaller than any of the two operands. So, try this:

ULONG size1 = 0xFFFFFF00;
ULONG size2 = 0xFFFFFF00;
size1 += size2; // == 0xFFFFFE00

if( size1 < size2)
    size1 = 0xFFFFFFFF;

Remember, maximum value of each type is totally OS dependant. Keep that in mind. Smile | :)

If someone got a better approach, please post. I'll try to remember how to check overflow in C++. If I got anything, I'll keep you posted.

Regards

Waldermort3-Sep-06 7:39

3-Sep-06 7:39

That probably is a better way of doing it. Thinking about it, the output would always be smaller. If you remember the better way, please post it. I would love to know.

toxcct3-Sep-06 5:24

toxcct

3-Sep-06 5:24

use the header <LIMITS.h>

with this, you can know the types max and min values, and some other informations for floating points types.

TOXCCT >>> GEII power

[VisualCalc 3.0 updated ][Flags Beginner's Guide new! ]

Rick York3-Sep-06 16:06

Rick York

3-Sep-06 16:06

You could do something like this :

_int64 sum = (_int64)size1 + (_int64)size2;
if( sum > ULONG_MAX )
TRACE( "an overflow occurred\n" );
else
size1 = (ULONG)sum;

Stephen Hewitt3-Sep-06 16:16

Stephen Hewitt

3-Sep-06 16:16

Try this:

ULONG size1 = 0xFFFFFF00;
ULONG size2 = 0xFFFFFF00;
ULONG sizeR = size1 + size2; // == 0xFFFFFE00
if (sizeR < max(size1, size2))
{
    // Overflow.
}

Steve

CString to wchar_t*!!!!!!

MayankT3-Sep-06 17:08

MayankT

3-Sep-06 17:08

try this:

<br />
if (MAXVAL - size1 < size2)<br />
{<br />
//overflow<br />
}<br />
else<br />
{<br />
size1 += size2;<br />
}<br />

MAXVAL can be obtained from limits.h as toxcct said.

----------------------
Mayank Thakore
Learning C++ - since 1998

They didn't print my card right; so I resigned.

mostafa_pasha3-Sep-06 3:13

mostafa_pasha

3-Sep-06 3:13

Hi,
how can i convert CString to wchar_t*? for example
CString m_strFilename;
how can i convert m_strFilename to wchar_t*?

i use L"" for string such that L"c:\filename.cpp", but is
there any function or macro do this?

Best Regards.
MJM

Waldermort3-Sep-06 3:25

3-Sep-06 3:25

MultiByteToWideChar() is the function you want, but why not just use CStringW?

mostafa_pasha3-Sep-06 3:33

mostafa_pasha

3-Sep-06 3:33

thanks , I use CStringW !

messages3-Sep-06 7:50

messages

3-Sep-06 7:50

See

<br />
USES_CONVERSION;<br />
CString str;<br />
str="Test";<br />
<br />
wchar_t* m_wchar = T2W(str.GetBuffer());<br />
str.ReleaseBuffer();<br />
<br />

Waldermort3-Sep-06 8:23

3-Sep-06 8:23

Which is almost exactly the same as what I told him. It's ok to use the conversion macros if you are doing a lot of converting, but for a single time it's pointless to add the extra header required.