Re: assignment and coyp constructor

From:

Victor Bazarov <v.Abazarov@comAcast.net>

Newsgroups:

comp.lang.c++

Date:

Mon, 14 Dec 2009 14:28:56 -0500

Message-ID:

<hg63mt$duu$1@news.datemas.de>

samhas wrote:

I have the following class definition:

class test2{
public:
  int y;
  test2(int ui) : y(ui){}
  test2(const test2& t) : y(5){}
  test2& operator= (const test2& t){ y = 6; return *this; }
};

In an expression like

test2 t2(test2(20)); or
test2(t2) = test2(20);

neither the copy-constructor nor the assgnment operator get executed.
t2 end up with y .

But in the following case the copy constructor gets called:

test2 t = test2(20);
test2 t2(t); // or test2 t2 = t;

Can someone explain to me what's going on there?

Most likely optimisation of some kind. The compiler is allowed to
create code that would omit the use of the copy-constructor (and to
construct the object directly) in certain cases, even if the copy c-tor
has side effects. The fact that the copy c-tor actually is supposed to
produce a different object than what you end up having, looks like a bug
to me. However, whose bug it is is rather debatable. When you write a
copy constructor and instead creating a unique element that has really
nothing in common with the object of which it's supposed to be a copy,
then it could be construed as *your* bug, and you're just reaping what
you've sown.

V
--
Please remove capital 'A's when replying by e-mail
I do not respond to top-posted replies, please don't ask