? About data types...

[slashdot]

I notice that there are different data types. And I am assuming there fore storing data (obviously), but my question is, if you need to store a piece of data that doesn't require a short data integer (signed: -32768 to 32767
unsigned: 0 to 65535), but could be put into a char (signed: -128 to 127
unsigned: 0 to 255) which could take less space. Would the program be faster if you be more precise? Or is it just the calculations that make the program faster or slower? If it totaly flew by my head, and you know it, could you give me a better definition on how it works?

[DJMaze]

it's memory.

char = 8 bit
short = 16 bit
int = 32 bit
int64 = 64 bit

So depending on how big the number can be you decide how much memory you want it to use.
If your app is small it has no real use to specificaly specify data.
However if you do huge calculations or have an array with millions of entries then you do notice the size differences.

Strings and characters are a bigger concern (like in wordprocessors) because each character is 8bit.

PHP Code:

int number = 65535; // 32 bits
char *string = "65535"; // 48 bits (including NULL terminator) 


[Valmont]

Would the program be faster if you be more precise?

Most likely not. Most likely for standard PC's even slower. The standard didn't provide a formal definition for that. Typically for such questions, one needs to test. In almost every case, the answer is the same: test.

There is another issue.
If there are no special requirements to meet when it comes to saving memory, then don't do too much work. For example, if you implement a variable using unsigned char, the coder that depends on your code, might end up with nasty surprises because he expected a plain integer.

For your conveniance, assume the int is the standard integer type, and double is the default floating point precision type. The ISO says so by the way. Only when special requirement need to be met, deviate.

Examples.
- Salary is usual up to 2 decimals accurate (cents). float seems to suffice. Nevertheless, implement double:

double MyHourSalary = 12.04;

- Index for an array or matrix. Often these containers can hold quite a bit of data. A normal int often doesn't suffice. unsigned seems is ok, but might be problematic as well. To make sure we have type-wise a valid index, we should use std::size_t :

#include <cstdlib>

std::size_t Idx = 12;
Double TheArray[Idx];
//Fill the array here. Then we iterate:
for(std::size_t i = 0; i < Idx; ++i}
{
   cout<<TheArray[i]<<endl;
}

Was a good question.