Uncaught Exception can be hard to detect

[LamacunLama]

With the following little program, an uncaught exception is provoked.

#include "single_header/standalone/fakeit.hpp"

using namespace fakeit;

class NotDefaultConstructible
{
    public:
        NotDefaultConstructible(int a){}
};

class ToMock
{
    public:
        virtual NotDefaultConstructible MyFunc() = 0;
};

int main()
{
    Mock<ToMock> myMock;
    Fake(Method(myMock, MyFunc));
    myMock.get().MyFunc();
}

If this exception is thrown when fakeit is used in combination with BOOST Test, the exception text of boost says: unknown type and with unknown type the exception type "DefaultValueInstatiationException" is meant. I don't know who is trying to catch such an exception type. It would be much more useful in this context to see a proper error description in the print output, which is already included in the .what().

I don't know if the following statement covers all use cases, but for me it would be easier to have a failing static_assertion if the return type of a faked function has no default constructor. This would therefore force the user to use "When" and to manually return something.
The check for this would look similar to this:

using ReturnType = decltype(myMock.get().MyFunc());
static_assert(std::is_constructible<ReturnType>());

[FranckRJ]

The issue is that FakeIt exceptions can't inherit std::exception because we don't want them to be caught by user code (it may not be the wisest choice, but that's how the library works for now). The static assert is a good idea but I'll need to search where to write it to be able to over all cases without false positives, because it must only be checked if we try to return a default-constructed object for a type that isn't default-constructible, so the check must be placed somewhere where it will be only evaluated in this case. Or maybe a static assert that cover only some cases is good enough and it can be placed in the Fake method, I don't know.