class X {
public:
class XExc {
public:
XExc() { cout << "Constructor X::XExc" << endl; }
// XExc(const XExc&) { cout << "Copy X::XExc" << endl; }
~XExc() { cout << "Destuctor X::XExc" << endl; }
};
void f() {
cout << "Inside f" << endl;
throw XExc(); }
};
int main() {
X x;
try {
x.f();
} catch(X::XExc&) {
cout << "Exception handler" << endl;
}
}
Exceptions : Something I'm missing ?[SOLVED]
Author
Hi.
I am now studying exceptions in C++. I wrote code to test destruction of thrown exception objects and when they occur. The output is somewhat strange.
Here is the code:
The output is :
Inside f
Constructor X::XExc
Destuctor X::XExc
Exception handler
Destuctor X::XExc
So the first destructor destroys the original XExc and the second destroys the copy made when throw is encountered.(at least that's how i see it)
But, notice I commented the copy-constructor. If I explicitly specify one, there's no Destructor before the handler, but there's no copy constructor output either. I am a little confused here.
[Edited by - Mocanu Razvan on April 14, 2006 4:29:41 AM]
Someone please correct me if I am wrong but I am not sure you should be catching a reference like that.
Quote:Original post by EasilyConfused
Someone please correct me if I am wrong but I am not sure you should be catching a reference like that.
Why not? Though you should probably catch it as const reference...
To the OP: It seems like it's able to optimize away the temporary when the copy constructor is user-defined. Not sure exactly why it won't optimize it when it's compiler-generated...
Throw by value, catch by reference.
I come to the same conclusion as RDragon. You might try making your example a little less contrived (throw from a deeper point, so more unwinding occurs, give your class some actual data to copy, et cetera) and see how things change.
I come to the same conclusion as RDragon. You might try making your example a little less contrived (throw from a deeper point, so more unwinding occurs, give your class some actual data to copy, et cetera) and see how things change.
Quote:Original post by RDragon1
To the OP: It seems like it's able to optimize away the temporary when the copy constructor is user-defined. Not sure exactly why it won't optimize it when it's compiler-generated...
Perhaps it's because he is compiling without optimizations? When run in "Release" mode in VC++2005, it behaves "correctly".
jfl.
Author
Yes indeed. I forgot I was using the Debug configuration. Compiling in Release mode solves the problem. And yes, it seems to be optimizing away the temporary.
Thanks.
Thanks.
This topic is closed to new replies.
Advertisement
Popular Topics
Advertisement
