Is throw by value and catch by reference the law of thumb?

Question

I am so confused and fed up with C++: As I have been taught "passing by reference or pointer" overcomes getting copies of parameters passed in a function.

So in an exception-handling scenario according to the law of thumb: throw by value, catch by reference.

Why Stroustrup in his "A Tour of C++" catches by value? eg:

void test()
{
    try {
        std::vector v(−27);
    }
    catch (std::length_error) {
        // handle negative size
    }
    catch (std::bad_alloc) {
        // handle memory exhaustion
    }
} 

This example is copied from the book. Page 59. (Invariants).

Answer 1

Why Stroustrup in his "A Tour of C++" catches by value?

Unless you want to copy the exception for some reason (e.g. store a custom exception type), you should always catch by const&. This might have been an oversight from Bjarne, or in that particular snippet it didn't really matter whether or not the exception was copied.

Related:

• "Always catch exceptions by reference"

• C++ catch blocks - catch exception by value or reference?

---

Why C++ is changing inconsistently? I mean from time to time everything changes. eg: in C++ 11:

There is no inconsistency here - static_assert merely became more flexible, as it doesn't force the user to provide an error message anymore. This is useful for cases where the boolean condition itself is self-explanatory and an error message would not add anything useful.