How to call a constructor in another constructor
-
Dear all, I have 2 constructors of the same class: CTest(){i = j = 0;}; CTest(int t){j = t;}; It is possible to call the first constructor in the second constructor? Of course, I can make an Initialize() method and I can call it in the both constructors. Thank you.
-
Sorry, but that link is not correct. at least not completely. The way to call one constructor from another is to use "placement new". For example: Foo::Foo() { int i; new (this)Foo(i); // this will call the Foo::Foo(int) ctor without allocating memory } Foo::Foo(int) { } This begs the question of why you want to do this. If it is for "common" initialization code, the better answer is to have a separate initialization function, as in the following. Then call that init code either from each ctor, or perhaps better, to call it from the places where Foo is constructed. then maybe handle init failures there. bool Foo:init(some parameters here maybe) { // do init code here }
-
Sorry, but that link is not correct. at least not completely. The way to call one constructor from another is to use "placement new". For example: Foo::Foo() { int i; new (this)Foo(i); // this will call the Foo::Foo(int) ctor without allocating memory } Foo::Foo(int) { } This begs the question of why you want to do this. If it is for "common" initialization code, the better answer is to have a separate initialization function, as in the following. Then call that init code either from each ctor, or perhaps better, to call it from the places where Foo is constructed. then maybe handle init failures there. bool Foo:init(some parameters here maybe) { // do init code here }
Actually, you can call a constructor from the object itself. A constructor is just another function/method, but is called by the memory allocation scheme (IE: Operator new). You can also call the destructor any time you want, as a standard object method (EX: pMyObj->~MyObject(); ). Why would you want to do this? Consider the following: You want to allocate memory dynamically in a busy, multithreaded environment. Therefore, you want to avoid the heap. So, you decide dynamically allocate memory from the stack using _alloca(). In order to initialize this "object", you need to call the constructor. Also, when you are finished with this object, you MUST NOT "delete" it, but you should destroy it properly by calling the destructor. A non-tested example for thought purposes:
#define STACK_OBJECT_NEW(c, p) {p = (c *)_alloca(sizeof(c)); if (p) p->c(); }
#define STACK_OBJECT_DELETE(c, p) {if (p) p->~c(); }void MyFunc()
{
MyClass *p = NULL;STACK_OBJECT_NEW(MyClass, p);
if (p)
{
p->InvokeSomeMethod();
...STACK\_OBJECT\_DELETE(MyClass, p); }}
onwards and upwards...
-
Dear all, I have 2 constructors of the same class: CTest(){i = j = 0;}; CTest(int t){j = t;}; It is possible to call the first constructor in the second constructor? Of course, I can make an Initialize() method and I can call it in the both constructors. Thank you.
I would recommend going the way of the
Initialize()method. It will make your code much easier to debug two years from now on a late, dark, rainy night with a customer expecting a bug fix early the next morning. /ravi My new year's resolution: 2048 x 1536 Home | Articles | Freeware | Music ravib@ravib.com -
Sorry, but that link is not correct. at least not completely. The way to call one constructor from another is to use "placement new". For example: Foo::Foo() { int i; new (this)Foo(i); // this will call the Foo::Foo(int) ctor without allocating memory } Foo::Foo(int) { } This begs the question of why you want to do this. If it is for "common" initialization code, the better answer is to have a separate initialization function, as in the following. Then call that init code either from each ctor, or perhaps better, to call it from the places where Foo is constructed. then maybe handle init failures there. bool Foo:init(some parameters here maybe) { // do init code here }
ahz wrote: The way to call one constructor from another is to use "placement new". This is incorrect. Herb Sutter - the current chairman of the C++ standards committee described this 'anti-idiom' as 'an abomination' and 'abhorrent' in the article 'delegating constructors' C/C++ Users Journal May 2003. For any data members that are not POD, this approach will fail, because the destructor of that data members will run only once, while the constructors will have run twice To the initial poster. The only way to solve this problem is with an initialization function, but you must also ensure that that function is not virtual.
-
Actually, you can call a constructor from the object itself. A constructor is just another function/method, but is called by the memory allocation scheme (IE: Operator new). You can also call the destructor any time you want, as a standard object method (EX: pMyObj->~MyObject(); ). Why would you want to do this? Consider the following: You want to allocate memory dynamically in a busy, multithreaded environment. Therefore, you want to avoid the heap. So, you decide dynamically allocate memory from the stack using _alloca(). In order to initialize this "object", you need to call the constructor. Also, when you are finished with this object, you MUST NOT "delete" it, but you should destroy it properly by calling the destructor. A non-tested example for thought purposes:
#define STACK_OBJECT_NEW(c, p) {p = (c *)_alloca(sizeof(c)); if (p) p->c(); }
#define STACK_OBJECT_DELETE(c, p) {if (p) p->~c(); }void MyFunc()
{
MyClass *p = NULL;STACK_OBJECT_NEW(MyClass, p);
if (p)
{
p->InvokeSomeMethod();
...STACK\_OBJECT\_DELETE(MyClass, p); }}
onwards and upwards...
-
you don't like my "hack"? :omg: Â onwards and upwards...
-
ahz wrote: The way to call one constructor from another is to use "placement new". This is incorrect. Herb Sutter - the current chairman of the C++ standards committee described this 'anti-idiom' as 'an abomination' and 'abhorrent' in the article 'delegating constructors' C/C++ Users Journal May 2003. For any data members that are not POD, this approach will fail, because the destructor of that data members will run only once, while the constructors will have run twice To the initial poster. The only way to solve this problem is with an initialization function, but you must also ensure that that function is not virtual.
It may be frowned upon, and may even be abhorent, but is it not incorrect. The original poster asked a simple direct technical question and I gave a simple direct techinically correct answer. It is correct C++ syntax, it is technically correct, and it works --- albeit for simple cases. I wouldn't use it, and don't recommend it --- for Herb Sutters reasons --- and for many of my own reasons that I won't belabor here. However, I agree the best way to solve the issue is with an initialization function. However, I don't think it necessarily needs to be non-virtual. Depends on where you are calling the function from. If from the ctor, then yes make sure it's non-virtual. If from outside the ctor (the method I recommend, then it's ok to be virtual).
-
It may be frowned upon, and may even be abhorent, but is it not incorrect. The original poster asked a simple direct technical question and I gave a simple direct techinically correct answer. It is correct C++ syntax, it is technically correct, and it works --- albeit for simple cases. I wouldn't use it, and don't recommend it --- for Herb Sutters reasons --- and for many of my own reasons that I won't belabor here. However, I agree the best way to solve the issue is with an initialization function. However, I don't think it necessarily needs to be non-virtual. Depends on where you are calling the function from. If from the ctor, then yes make sure it's non-virtual. If from outside the ctor (the method I recommend, then it's ok to be virtual).
ahz wrote: It may be frowned upon, and may even be abhorent, but is it not incorrect. The original poster asked a simple direct technical question and I gave a simple direct techinically correct answer. Wrong, wrong, wrong :( :mad: X| From: http://www.parashift.com/c++-faq-lite/ctors.html#faq-10.3[^] "BTW do NOT try to achieve this via placement new. Some people think they can say new(this) Foo(x, int(x)+7) within the body of Foo::Foo(char). However that is bad, bad, bad. Please don't write me and tell me that it seems to work on your particular version of your particular compiler; it's bad. Constructors do a bunch of little magical things behind the scenes, but that bad technique steps on those partially constructed bits. Just say no."
-
ahz wrote: It may be frowned upon, and may even be abhorent, but is it not incorrect. The original poster asked a simple direct technical question and I gave a simple direct techinically correct answer. Wrong, wrong, wrong :( :mad: X| From: http://www.parashift.com/c++-faq-lite/ctors.html#faq-10.3[^] "BTW do NOT try to achieve this via placement new. Some people think they can say new(this) Foo(x, int(x)+7) within the body of Foo::Foo(char). However that is bad, bad, bad. Please don't write me and tell me that it seems to work on your particular version of your particular compiler; it's bad. Constructors do a bunch of little magical things behind the scenes, but that bad technique steps on those partially constructed bits. Just say no."
I guess you can't read. It is correct. Whether you should do it is another thing altogether. LEARN TO READ!