I have a class that can throw an exception in its constructor. How can I declare an instance of that class in a try/catch block, while still making it available in the right scope?
try { MyClass lMyObject; }
catch (const std::exception& e) { /* Handle constructor exception */ }
lMyObject.DoSomething(); // lMyObject not in scope!
Is there an alternative way to accomplish this, while respecting the RAII idiom?
I'd prefer not to use an init() method for two-phased construction. The only other thing I could come up with was:
MyClass* lMyObject;
try { lMyObject = new MyClass(); }
catch (const std::exception& e) { /* Handle constructor exception */ }
std::shared_ptr<MyClass> lMyObjectPtr(lMyObject);
lMyObjectPtr->DoSomething();
Works OK, but I'm not happy with the raw pointer in scope and pointer indirection. Is this just another C++ wart?
If a constructor throws that means the object failed to initialize and hence it failed to start its existence.
MyClass* lMyObject;
try { lMyObject = new MyClass(); }
catch (std::exception e) { /* Handle constructor exception */ }
In the above if the constructor throws an exception, lMyObject is left uninitialized, in other words, the pointer contains an indeterminate value.
See classic Constructor Failures for a detailed explanation:
We might summarize the C++ constructor model as follows:
Either:
(a) The constructor returns normally by reaching its end or a return statement, and the object exists.
Or:
(b) The constructor exits by emitting an exception, and the object not only does not now exist, but never existed.
There are no other possibilities.
The best way of writing your code is this:-
MyClass lMyObject;
lMyObject.DoSomething();
No trys, catches, or pointers.
If the constructor throws, then DoSomething can't get called. Which is right: If the constructor threw, then object was never constructed.
And, importantly, don't catch (or even catch/rethrow) exceptions unless you have something constructive to do with them. Let exceptions do their job and ripple up until something that knows how to handle them can do its job.
Constructors are for putting an object into a consistent state and establishing class invariants. Allowing an exception to escape a constructor means that something in that constructor has failed to complete, so now the object is in some unknown weird state (which may include resource leaks now too). Since the object's constructor has not completed, the compiler will not cause it's destructor to be invoked either. Perhaps what you're looking for is to catch the exception inside the constructor. Assuming it's not rethrown, this would cause the constructor to complete execution, and the object is now fully-formed.
You don't need to use shared_ptr, use unique_ptr:
std::unique_ptr<MyClass> pMyObject;
try { pMyObject.reset(new MyClass()); }
catch (std::exception &e) { /* Handle constructor exception */ throw; }
MyClass &lMyObject = *pMyObject;
lMyObject.DoSomething();
Obviously, it's your responsibility to ensure that the program does not fall through the catch block without either initialising pMyObject, or exiting the function (e.g. via return or throw).
If available, you can use Boost.Optional to avoid using heap memory:
boost::optional<MyClass> oMyObject;
try { oMyObject.reset(MyClass()); }
catch (std::exception &e) { /* Handle constructor exception */ throw; }
MyClass &lMyObject = *oMyObject;
lMyObject.DoSomething();
You could set up MyClass's copy constructor to accept garbage input, thereby effectively declaring a pointer with your declaration of the object. Then you could manually call the default constructor within the try block:
MyClass lMyObject(null); // calls copy constructor
try {
new (lMyObject) MyClass(); // calls normal constructor
}
catch (const std::exception& e) { /* Handle constructor exception */ }
lMyObject.DoSomething();
Related
Lets consider having a class with a constructor throwing an exception like following:
class Class
{
public:
Class(type argument)
{
if (argument == NULL)
{
throw std::exception("Value cannot be null.\nParameter name: argument");
}
// Instructions
}
private:
// Properties
}
Since the class constructor might throw an exception we cannot declare an object directly.
Class obj(argument); // Harmful
Which means the constructor must be called must be using a try/catch
try
{
Class obj(argument);
}
catch (std::exception& ex)
{
std::cout << ex.what() << std::endl;
}
The problem is that we can only use the object inside the try block. The only way to use it outside of the try block is to declare a Class* pointer then use the new keyword to construct a new object then assign it's address to the previous pointer.
Class* pObj;
try
{
pObj = new Class(argument);
}
catch (std::exception& ex)
{
std::cout << ex.what() << std::endl;
}
So what is the standard way to define the previous class in order to create instances without using pointers or dynamic memory allocation?
Thank you in advance
Since the class constructor might throw an exception we cannot declare
an object directly.
Yes, you can. You only need to put it in a try block if you actually have a plan to deal with an exception right there, in the function. If you don't have such a plan, then just let the exception propogate (though you should catch it eventually, just to provide a report if nothing else).
But, assuming you do have a plan to handle the exception right there in the function, then the solution is simple.
try {
Class obj(argument);
// use obj here, inside the try block
}
catch(...) { ... }
// not here, outside the try block
Edit: By your comment below, either you are misunderstanding me, or I am misunderstanding you. Perhaps a concrete exaqmple is required. Let's say that this is your function which uses your class:
void Foobar(type argument)
{
Class obj(argument);
obj.method1(1,2,3);
obj.method2(3,4);
int x = Wizbang(obj);
gobble(x);
}
Now, you want to handle the exception that the Class constructor might throw. What I'm suggesting is putting all of that junk in the function, into a try block, thusly:
void Foobar(type argument)
{
try
{
Class obj(argument);
obj.method1(1,2,3);
obj.method2(3,4);
int x = Wizbang(obj);
gobble(x);
}
catch(std::exception& e)
{
std::cout << e.what() << std::endl;
}
}
If you can't do this, please explain why. You've said "I need access to use the object later", but you've provided no reason why "later" cannot mean "later within the same try block where the object was created". As such, your requirements are incomplete.
What you want is not (reasonably) possible. Suppose there were some method of doing what you want, that is, creating an object obj of type Class that was still in scope outside any relevant try{}catch{} construct used to handle exceptions in the constructor. Thw question to ask yourself is, if an exception were to be thrown during construction of the obj, and you somehow handled the exception and then execution were to continue in the block where the obj was still in scope, what exactly would the obj contain? It was never constructed, so just what would be in the obj????
Please have a look at demo code :
class myError
{
const char* str;
public:
myError():str(NULL) {}
myError(const char* temp)
{
str = temp;
}
const char* what()
{
return str;
}
};
class ab
{
int x;
public:
ab() try :x(0)
{
throw myError("error occured in the constructor of class ab");
}
catch(myError& temp)
{
std::cout<<"Handler no. 1 of ab constructor"<<std::endl;
}
};
int main () try
{
ab bb;
cout << "Resumed execution!" << endl;
return 0;
}
catch(myError& temp)
{
std::cout<<"Handler below the main function"<<std::endl;
std::cout<<"And the error is :" <<temp.what();
}
My Questions :
Why only function try block's handler of ctor and dtor only rethows the exception? ,
and when you simply throw exception inside ctor , its handler doesn't rethrows the object? i.e
Ctor::Ctor()
{
try{
throw Excep1();
}
catch(Excep1& temp) {
std::cout<<"Doesn't rethrows the exception object";
}
}
I wanna know that how to resume the control back to cout << "Resumed execution!" << endl; , after handling rethrown object?
why is it often said that we shouldn't place function try block on dtor of base class?
The usual rule is that a catch block doesn't rethrow unless you ask it
to. How would you stop the exception from propagating otherwise. In
the case of a constructor, however, if something in the initialization
list throws, then you haven't got a fully constructed object; there is
nothing you could do with the object, not even call the destructor on
it. And if the function catch block of the constructor doesn't
rethrow, what is it going to do, since it cannot simply return (and
leave the variable on the stack)?
In all other cases, it's up do the function containing the catch block
to know what to do. In the case of your main, for example, you could
write:
try {
ab bb;
} catch (...) {
}
std::cout << "Resumed execution!" << std::endl;
What you can't do is execute code where bb would be in scope and
accessible, but not have been correctly constructed.
As for why you shouldn't place a function try block on the destructor of a
base class, I've never heard that rule. In general, destructors
shouldn't throw, so there's no point in wrapping them in a try block,
period.
For the second question, destructors shouldn't be throwing period. Consider a case where your destructor is freeing a lot of memory through delete. What would happen if your destructor threw an error before finishing the clean up? You now have a memory leak. If your destructor is causing a runtime error, then you probably have problems elsewhere in your code that you need to fix.
As usually, Herb Sutter knows and explains everything:
If the handler body contained the statement "throw;" then the catch block would obviously rethrow whatever exception A::A() or B::B() had emitted. What's less obvious, but clearly stated in the standard, is that if the catch block does not throw (either rethrow the original exception, or throw something new), and control reaches the end of the catch block of a constructor or destructor, then the original exception is automatically rethrown.
More in his article
I have a simple C++ object that I create at the start of function F() to ensure two matched functions (OpDo, OpUndo) are called at the start and return of the F(), by using the object's constructor and destructor. However, I don't want the operation to be undone in case an exception was thrown within the body of F(). Is this possible to do cleanly? I have read about std::uncaught-exception, but its use does not seem to be recommended.
Most people have used std::uncaught_exception() to try to tell if an exception is pending, so they can throw an exception from a destructor if there isn't one already. That is generally considered Not A Good Idea.
If you want to not undo an operation if an exception has thrown, it should do the trick.
Remember that the destructor is your last chance to release any resources an object has, because after the destructor ends the object does not exist, and any resources it held are now permanently leaked. If OpDo() allocates any memory or file handles or whatever, you do need to deal with that in the destructor no matter what.
You can subvert the Scope Guard idiom. Instead of not doing something in the destructor when no exception is thrown, we invert that and only do something if no exception is thrown:
class DoUndoRAII{
public:
DoUndoRAII()
: noexcept_(false)
{
// your stuff here
}
~DoUndoRAII(){
if(noexcept_){
// do whatever you need to do
}
}
void no_exception(){
noexcept_ = true;
}
private:
bool noexcept_;
};
void func(){
DoUndoRAII do_undo;
// last line
do_undo.no_exception();
}
When an exception is thrown, do_undo.no_exception() will never be called and thus never set the noexcept_ value to true. :) An example can be found here on Ideone.
Let's assume that your F returns some class Helper:
Helper F()
{
MyClass doUndoWrapper;
}
When flow is normal - Helper is created. When exception is raised no copy of Helper is created. Try use this semantic by placing to private region constructor of Helper and declaring F as friend - so no one could create helper.
class Helper
{
private:
friend Helper F();
Helper(){ //place there OpDo semantic - first entry
// construct this class
Helper(const Helper& copy){ //this must present to allow stack operations
// copy constructor will be called at end of `F` to return value
// so place OpUndo semantic there to mark success without exception
Will the following append() in the catch cause the rethrown exception to see the effect of append() being called?
try {
mayThrowMyErr();
} catch (myErr &err) {
err.append("Add to my message here");
throw; // Does the rethrow exception reflect the call to append()?
}
Similarly, if I rewrite it this way, will bit slicing occur if the actual exception is derived by myErr?
try {
mayThrowObjectDerivedFromMyErr();
} catch (myErr &err) {
err.append("Add to my message's base class here");
throw err; // Do I lose the derived class exception and only get myErr?
}
In both cases, since you catch by reference, you are effectively altering the state of the original exception object (which you can think of as residing in a magical memory location which will stay valid during the subsequent unwinding -- 0x98e7058 in the example below). However,
In the first case, since you rethrow with throw; (which, unlike throw err;, preserves the original exception object, with your modifications, in said "magical location" at 0x98e7058) will reflect the call to append()
In the second case, since you throw something explicitly, a copy of err will be created then thrown anew (at a different "magical location" 0x98e70b0 -- because for all the compiler knows err could be an object on the stack about to be unwinded, like e was at 0xbfbce430, not in the "magical location" at 0x98e7058), so you will lose derived-class-specific data during the copy-construction of a base class instance.
Simple program to illustrate what's happening:
#include <stdio.h>
struct MyErr {
MyErr() {
printf(" Base default constructor, this=%p\n", this);
}
MyErr(const MyErr& other) {
printf(" Base copy-constructor, this=%p from that=%p\n", this, &other);
}
virtual ~MyErr() {
printf(" Base destructor, this=%p\n", this);
}
};
struct MyErrDerived : public MyErr {
MyErrDerived() {
printf(" Derived default constructor, this=%p\n", this);
}
MyErrDerived(const MyErrDerived& other) {
printf(" Derived copy-constructor, this=%p from that=%p\n", this, &other);
}
virtual ~MyErrDerived() {
printf(" Derived destructor, this=%p\n", this);
}
};
int main() {
try {
try {
MyErrDerived e;
throw e;
} catch (MyErr& err) {
printf("A Inner catch, &err=%p\n", &err);
throw;
}
} catch (MyErr& err) {
printf("A Outer catch, &err=%p\n", &err);
}
printf("---\n");
try {
try {
MyErrDerived e;
throw e;
} catch (MyErr& err) {
printf("B Inner catch, &err=%p\n", &err);
throw err;
}
} catch (MyErr& err) {
printf("B Outer catch, &err=%p\n", &err);
}
return 0;
}
Result:
Base default constructor, this=0xbfbce430
Derived default constructor, this=0xbfbce430
Base default constructor, this=0x98e7058
Derived copy-constructor, this=0x98e7058 from that=0xbfbce430
Derived destructor, this=0xbfbce430
Base destructor, this=0xbfbce430
A Inner catch, &err=0x98e7058
A Outer catch, &err=0x98e7058
Derived destructor, this=0x98e7058
Base destructor, this=0x98e7058
---
Base default constructor, this=0xbfbce430
Derived default constructor, this=0xbfbce430
Base default constructor, this=0x98e7058
Derived copy-constructor, this=0x98e7058 from that=0xbfbce430
Derived destructor, this=0xbfbce430
Base destructor, this=0xbfbce430
B Inner catch, &err=0x98e7058
Base copy-constructor, this=0x98e70b0 from that=0x98e7058
Derived destructor, this=0x98e7058
Base destructor, this=0x98e7058
B Outer catch, &err=0x98e70b0
Base destructor, this=0x98e70b0
Also see:
Scope of exception object in C++
Throwing ... "by reference"
This question is rather old and has an answer appropriate to the time it was asked.
However, I just want to add a note on how to do proper exception handling since C++11 and I believe this corresponds very well to what you were trying to achieve with your append function:
Use std::nested_exception and std::throw_with_nested
It is described on StackOverflow here and here, how you can get a backtrace on your exceptions inside your code without need for a debugger or cumbersome logging, by simply writing a proper exception handler which will rethrow nested exceptions.
Since you can do this with any derived exception class, you can add a lot of information to such a backtrace!
You may also take a look at my MWE on GitHub, where a backtrace would look something like this:
Library API: Exception caught in function 'api_function'
Backtrace:
~/Git/mwe-cpp-exception/src/detail/Library.cpp:17 : library_function failed
~/Git/mwe-cpp-exception/src/detail/Library.cpp:13 : could not open file "nonexistent.txt"
Yes, rethrowing rethrows the original exception object, which you have modified by a reference. You can also catch a base class reference, modify by it and still be able to rethrow the original derived exception type by throw;.
for first question, yes.
but for second, refer to Vlad answer.
you will need to carefully design your exception object to handle copy ctor. by convention, base class doesn't recognize its child so you will most likely lose the additional data carried by derived class.
The C++2003 Standart, was at the time the active standard when that question was asked. Please be aware that C++ Language has a standard.
https://www.open-std.org/JTC1/SC22/WG21/docs/papers/2005/n1905.pdf
p.351, p.15.4:
When the handler declares a non-constant object, any changes to that object will not affect the temporary object that was initialized by execution of the throw-expression. When the handler declares a reference to a non-constant object, any changes to the referenced object are changes to the temporary object initialized when the throw expression was executed and will have effect should that object be rethrown.
So you can be sure that any proper C++ compiler of C++2003 will not create any extra copies...
Another way you can play with code snippets like presented below and observe the exact address of the object, and as each object has a unique address in C++, that implies that the object is the same. But it only enhances your confidence.
The only way to be 100% sure - look into Standart for Programming Language.
#include <iostream>
using std::cout;
class A{};
void f1() {
throw A();
}
void f2()
{
try {
f1();
}
catch(A& obj) {
cout << "f1 obj: " << &obj << "\n";
throw;
}
}
void f3()
{
try {
f2();
}
catch(A& obj) {
cout << "f3 obj: " << &obj << "\n";
throw;
}
}
int main()
{
try {
f3();
}
catch(A& obj)
{
cout << "main obj: " << &obj;
}
return 0;
}
Is it possible to handle exceptions in these scenarios:
thrown from constructor before entering main()
thrown from destructor after leaving main()
You can wrap up your constructor withing a try-catch inside of it.
No, you should never allow exception throwing in a destructor.
The funny less-known feature of how to embed try-catch in a constructor:
object::object( int param )
try
: optional( initialization )
{
// ...
}
catch(...)
{
// ...
}
Yes, this is valid C++. The added benefit here is the fact that the try will catch exceptions thrown by the constructors of the data members of the class, even if they're not mentioned in the ctor initializer or there is no ctor initializer:
struct Throws {
int answer;
Throws() : answer(((throw std::runtime_error("whoosh!")), 42)) {}
};
struct Contains {
Throws baseball;
Contains() try {} catch (std::exception& e) { std::cerr << e.what() << '\n'; }
};
Yes: don't use dangerous global objects!
It might be possible to set an exception handler before construction / destruction of the objects in question, that one should be able to handle those exceptions.
For constructors, there's some weird new syntax that allows exceptions to be caught within the constructor. Not sure how that works, and it's not commonly implemented in many compilers.
For destructors, you have to wrap the content of the destructor in a try { code(); } catch(...) {} block. Which may not always be the desired behavior, depending on what you want to achieve in that destructor.
Short answer: no.
Any global object that throws an exception in its constructor will cause an unhandled exception (that is, terminate be called).
Any class that throws an exception in its destructor is a broken class.
Using the singleton pattern rather than globals will give you more options.