Exception handling in Visual Studio 2005

[ahammad]

When using the try catch block in Visual Studio 2005, we cannot have:

try{

[something]

} catch [B](...)[/B] {

[something]

}

For some reason, we cannot use the (...) in the catch part for exception handling. Is there a way to do this in Visual Studio 2005 so that the code in the catch part would be run for every exception?

[weaknessforcats]

This code:

try{

//[something]

} catch (...) {

//[something]

}

compiles with no errors using Visual Studio.NET 2005.

Have you enabled exception handling in your project properties??

[Silent1Mezzo]

When using the try catch block in Visual Studio 2005, we cannot have:

try{

[something]

} catch [B](...)[/B] {

[something]

}

For some reason, we cannot use the (...) in the catch part for exception handling. Is there a way to do this in Visual Studio 2005 so that the code in the catch part would be run for every exception?

Couldn't you just catch the super class ?

try{

[something]

} catch [B](Exception e)[/B] {

[something]

}

[ahammad]

Sure it compiles, but whatever is in the catch block doesn't get executed in my case.

	int nums[1];
	nums[0] = 5;
	try
	{
		int x = nums[99];
		int y = 2;
	}
	catch (...)
	{
		TRACE("Catch\n");
	}

The program will handle the error, but the TRACE command does not get executed. It's like the program handles the error but skips the catch block altogether.

[weaknessforcats]

This code:

int nums[1];
nums[0] = 5;
try
{
int x = nums[99];
int y = 2;
}
catch (...)
{
TRACE("Catch\n" );
}

Does not throw an exception. It just corrupts memory.

Without a throw, there can be no catch.

Also, when you catch a base class object be certain that you catch it by reference or by pointer. If you don't, and it's really a derived obect, all the derived part will be sliced off. So all you will see in the catch block is the base part of your object.

[code=cpp]
try{

[something]

} catch (Exception e) { <<<<Exception & or Exception* !!

[something]

}
[/code]

[ahammad]

when I do something like this:

int num[2];
num[0] = 5;
num[1] = 6;

int testvalue;


try
{
	testvalue = num[9];
}
catch (...)
{
	std::cout << "EXCEPTION caught \n";

}

int x;
cin >> x;

I get an error and the console application crashes. Shouldn't the program catch the error?

[RedSon]

when I do something like this:

int num[2];
num[0] = 5;
num[1] = 6;

int testvalue;


try
{
	testvalue = num[9];
}
catch (...)
{
	std::cout << "EXCEPTION caught \n";

}

int x;
cin >> x;

I get an error and the console application crashes. Shouldn't the program catch the error?

what is in your catch parens?

[ahammad]

what is in your catch parens?

Nothing except the three dots, what you see is what you get. It is supposed to mean "handle all exceptions".

The reason I did it like that was because I was told that this should be perfectly legal, and that it doesn't work with Microsoft compilers, but it does work with other compilers. I am not sure if that is true or not.

Funny thing is, if I use regular arrays, it doesn't work, but if I use a class like vectors, it does. This code for example, works perfectly, and both catch blocks are executed:

    std::vector<int> v;

    try
    {
        v.at(10) = 1;
    }
    catch (std::out_of_range)
    {
        std::cout << "caught std::out_of_range\n";
    }

    try
    {
        v.at(10) = 1;
    }
    catch (...)
    {
        std::cout << "EXCEPTION: " << e.what() << "\n";
    }

The point is, if it is legal in other C++ compilers, then there has to be a similar command instead of (...)...but why does it work for classes and not arrays?

[plemoine]

This is normal: C "arrays" are dumb and not protected. An array subscript is simply a pointer, an iterator into memory that you can increment or decrement at will... until you corrupt something else in memory.

On the other hand, STL vectors do bound checking and throw an exception from operator[] if you blow the limit. But this is not built-in in the language itself, it is in STL.

Nothing except the three dots, what you see is what you get. It is supposed to mean "handle all exceptions".

The reason I did it like that was because I was told that this should be perfectly legal, and that it doesn't work with Microsoft compilers, but it does work with other compilers. I am not sure if that is true or not.

Funny thing is, if I use regular arrays, it doesn't work, but if I use a class like vectors, it does. This code for example, works perfectly, and both catch blocks are executed:

    std::vector<int> v;

    try
    {
        v.at(10) = 1;
    }
    catch (std::out_of_range)
    {
        std::cout << "caught std::out_of_range\n";
    }

    try
    {
        v.at(10) = 1;
    }
    catch (...)
    {
        std::cout << "EXCEPTION: " << e.what() << "\n";
    }

The point is, if it is legal in other C++ compilers, then there has to be a similar command instead of (...)...but why does it work for classes and not arrays?

[weaknessforcats]

I get an error and the console application crashes. Shouldn't the program catch the error?

Let's try this again.

Unless there is a throw, there is no exception. If there is no exception, then there is nothing to catch.

This example:

int num[2];
num[0] = 5;
num[1] = 6;

int testvalue;


try
{
testvalue = num[9];
}
catch (...)
{
std::cout << "EXCEPTION caught \n";

}

int x;
cin >> x;

just corrupts memory. Look at the code. Do you see a throw instruction??
No. Hence, no exception. Hence, nothing to catch.

As to
[code=cpp]
testvalue = num[9];
[/code]

All this does is load an int (testvalue) with the contents of element 9 of the array num. The fact that num has only two elements doesn't matter. The compiler just takes the address of the array plus 9 times the sizeof(int) and plops the values at that memory address into testvaue. You are responsible for correct memory management, not the compiler.

[gnanapoongothai]

Sorry to to start a new query here, if try catch is used to handle exception then what about HANDLES in .could u ppl can explain it what is doing exactly

[weaknessforcats]

Please start a new discussion.