[cfe-commits] r157138 - in /cfe/trunk: lib/CodeGen/CGExprCXX.cpp test/CodeGenCXX/cxx0x-delegating-ctors.cpp

Jordan Rose jediknil at belkadan.com
Sat May 19 21:34:11 PDT 2012


On May 20, 2012, at 0:20, Richard Smith wrote:

> On Sat, May 19, 2012 at 6:43 PM, Jordy Rose <jediknil at belkadan.com> wrote:
>> I am not a standardista, but this doesn't seem right to me. The target constructor finishes before the delegating constructor runs, and presumably the target constructor has to be able to initialize the entire object by itself. I don't see anything in the standard that says 'x' is zero-initialized by the default constructor on its own, the default constructor as a target, or the delegating constructor.
>> 
> [class.base.init]/7: The expression-list or braced-init-list in a mem-initializer is used to initialize the designated subobject (or, in the case of a delegating constructor, the complete class object) according to the initialization rules of [dcl.init] for direct-initialization.
> [dcl.init]/10: An object whose initializer is an empty set of parentheses, i.e., (), shall be value-initialized.
> [dcl.init]/7: To value-initialize an object of type T means: [...] if T is a (possibly cv-qualified) non-union class type without a user-provided constructor, then the object is zero-initialized [...]

But X /does/ have a user-provided constructor -- X::X(int). Or it will after we finish defining it. Thus we should just call the implicit default constructor, which will not initialize 'x'.

That's a technicality, but one that feels like C++ to me. Otherwise you have to pay for the zero-initialization whether you use it or not.

(I do see now that X() counts as value-initialization, which I had missed earlier.)

> [dcl.fct.def.default]/4: A member function is user-provided if it is user-declared and not explicitly defaulted or deleted on its first declaration.
> 
> Thus X() performs zero-initialization before calling the default constructor. Even when used in a delegating initialization.

Jordy



More information about the cfe-commits mailing list