How to avoid integer overflow?

Question

In the following C++ code, 32767 + 1 = -32768.

#include <iostream>
int main(){
short var = 32767;
var++;
std::cout << var;
std::cin.get();
}

Is there any way to just leave "var" as 32767, without errors?

Do you get whats going on here? You've hit the ceiling for an integer (short), so adding one more flips the signs to the maximum negative amount for an integer. — blu
– blu, Commented Jun 14, 2010 at 23:47
If you want a different mode where integers don't automatically wrap around, it doesn't exist, sorry. — dmazzoni
– dmazzoni, Commented Jun 14, 2010 at 23:48
This is called saturation arithmetic. It can be highly illogical, so be warned. What's 32767+1-1? — MSalters
– MSalters, Commented Jun 15, 2010 at 8:58

Mehrdad Afshari · Accepted Answer · 2010-06-14 23:56:23Z

31

Yes, there is:

if (var < 32767) var++;

By the way, you shouldn't hardcode the constant, use numeric_limits<short>::max() defined in <limits> header file instead.

You can encapsulate this functionality in a function template:

template <class T>
void increment_without_wraparound(T& value) {
   if (value < numeric_limits<T>::max())
     value++;
}

and use it like:

short var = 32767;
increment_without_wraparound(var); // pick a shorter name!

answered Jun 14, 2010 at 23:44

Mehrdad Afshari

424k93 gold badges864 silver badges796 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

Khaled Alshaya · Accepted Answer · 2010-06-14 23:47:20Z

0

#include <iostream> 
int main(){ 
unsigned short var = 32767; 
var++; 
std::cout << var; 
std::cin.get(); 
}

answered Jun 14, 2010 at 23:47

Khaled Alshaya

97.5k41 gold badges183 silver badges237 bronze badges

This answers the question that was asked, but not, I suspect, the question that was intended.

technomage · Accepted Answer · 2010-06-14 23:53:50Z

-1

use 'unsigned short int' or 'long int'

#include <iostream>
int main(){
long int var = 32767;
var++;
std::cout << var;
std::cin.get();
}

answered Jun 14, 2010 at 23:53

technomage

5254 silver badges15 bronze badges

Then x = 32676; x++ will result in x == 32677, not 32676.

OK, so how do you avoid the overflow on an unsigned 64 bit integer?