Creating an instance of a class

[jimbo]

Dear all,

I am more or less new to c++ programing and therefore still have
problems with some fundamentals

At the moment I try to build a GUI-Application with Qt4. Sometimes I
have seen in the tutorials I have to instance a class with the new
command. Like QWidget QLabel *label = new QLabel("Hello World"); Then
I have created a pointer to a QLabel Object. The other way I have seen
is to just use QLabel label("Hello World"); Without a pointer and
without the new statement.

So to access the functions of the QLabel class I have to use the "."
for the part without pointer like label.show and for the pointer part:
label->show.

Can somebody explain me, why it is possible to use the two ways and
what is the difference in general.

Thank you a lot in advance.

jimbo

 

[=?iso-8859-1?q?Erik_Wikstr=F6m?=]

Dear all,

I am more or less new to c++ programing and therefore still have
problems with some fundamentals

At the moment I try to build a GUI-Application with Qt4. Sometimes I
have seen in the tutorials I have to instance a class with the new
command. Like QWidget QLabel *label = new QLabel("Hello World"); Then
I have created a pointer to a QLabel Object. The other way I have seen
is to just use QLabel label("Hello World"); Without a pointer and
without the new statement.

So to access the functions of the QLabel class I have to use the "."
for the part without pointer like label.show and for the pointer part:
label->show.

Can somebody explain me, why it is possible to use the two ways and
what is the difference in general.

You really should get yourself a good book if you haven't got one
already because it can take some time to learn those things and there
are a lot of things to keep in mind which I can't explain in one post
(but you are always welcome to ask questions).

The difference between using 'QLabel label;' and QLabel* label = new
QLabel();' is where the instance is stored and its scope. When not
using new the instance will be created on the stack and will be
destructed when it goes out of scope (this usually coincide with a
'}').

When using new the object is created on the heap and will remain there
until you call delete on a pointer pointing to it. This means that an
object instantiated in one function using new can be used in another
function (even if the first function has returned) if given a pointer
to it.

For every time you create an object with new you must (for each and
every path of execution) call delete on a pointer to that object or
your object will leak memory. To make things a bit more trickier this
is not always true when using Qt since a parent object will usually
delete/destruct all child objects on destruction.

An advice regarding dynamically allocated memory (using new) is to not
use it unless you have to.

 

[Piyo]

Dear all,

I am more or less new to c++ programing and therefore still have
problems with some fundamentals

At the moment I try to build a GUI-Application with Qt4. Sometimes I
have seen in the tutorials I have to instance a class with the new
command. Like QWidget QLabel *label = new QLabel("Hello World"); Then
I have created a pointer to a QLabel Object. The other way I have seen
is to just use QLabel label("Hello World"); Without a pointer and
without the new statement.

So to access the functions of the QLabel class I have to use the "."
for the part without pointer like label.show and for the pointer part:
label->show.

Can somebody explain me, why it is possible to use the two ways and
what is the difference in general.

Thank you a lot in advance.

jimbo

Given an object and a pointer to that object, you can access members
like so:

class foo
{
public:
int bar;
};

int
main()
{
foo* baz1 = new foo();
foo baz2;

baz2.bar; // cool
baz2->bar; // not cool
baz1.bar; // not cool
(*baz1).bar; // cool but too long
(baz1)->bar; // short hand version of the previous line
baz1->bar; // even shorter version of the previous line
}

In summary, "->" is a short hand notation to dereference the
pointer first and then access its member.

Good Luck!

 

[jimbo]

Given an object and a pointer to that object, you can access members
like so:

class foo
{
public:
int bar;

};

int
main()
{
foo* baz1 = new foo();
foo baz2;

baz2.bar; // cool
baz2->bar; // not cool
baz1.bar; // not cool
(*baz1).bar; // cool but too long
(baz1)->bar; // short hand version of the previous line
baz1->bar; // even shorter version of the previous line

}

In summary, "->" is a short hand notation to dereference the
pointer first and then access its member.

Good Luck!

Hi to all of you,

first of all, thanks a lot for the explanation. I didn´t know that
this whole topic is related to where the memory is allocated. Now with
heap and stack I can imagine a little bit better what I am doing

But to come back to my own little example. When i am already in a
main() function of my app and i create the instances there without the
new operator then i will have no problems concerning the use of the
instance in a different scope because i am already in main and main is
not ended yet, or?

I think i need to take it more carefully when i implement an own class
and use functions there.

Thx for your help.

Best regards,

jimbo

 

[Dizzy]

But to come back to my own little example. When i am already in a
main() function of my app and i create the instances there without the
new operator then i will have no problems concerning the use of the
instance in a different scope because i am already in main and main is
not ended yet, or?

After main ends the destructors of "static" storage objects will run. Make
sure that none of those will reference objects from main's stack.

I think i need to take it more carefully when i implement an own class
and use functions there.

Yeah, I think too many examples of Qt's documentation use dynamic memory
allocation when they could use "auto" memory allocation just fine. Not to
mention they have some strange system of keeping track of dynamic memory
objects themselves (if derived from QObject I think) which tends to lead to
a coding style where one forgets to use delete (it may be OK with QObject
derived objects but certainly it's not a good coding style to apply outside
of Qt).

 

[=?iso-8859-1?q?Erik_Wikstr=F6m?=]

Hi to all of you,

first of all, thanks a lot for the explanation. I didn´t know that
this whole topic is related to where the memory is allocated. Now with
heap and stack I can imagine a little bit better what I am doing

But to come back to my own little example. When i am already in a
main() function of my app and i create the instances there without the
new operator then i will have no problems concerning the use of the
instance in a different scope because i am already in main and main is
not ended yet, or?

Yes, the scope of main is a superset of almost all other scopes
(static being the only exception that I can think of right now).

I think i need to take it more carefully when i implement an own class
and use functions there.

It's not so much when implementing a class as when using it. But since
we are on the topic of implementing classes and dynamic memory
allocation I can say that if you implementation uses new instantiate
members (perhaps in the constructor) you should probably delete them
in the destructor, like so:

class Foo {
int* arr; // An array
public:
Foo(int i) : arr(new int { } // * allocate an array of size i
~Foo() { delete[] arr; } // Free the memory used
};

This way a user of the class will never have to worry about how the
class manages its resources, it will just work.

* If you don't understand the ': arr(new int { }'-bit it's an
initializer-list and does the same thing as '{ arr = new int; }',
for a more detailed explanation see http://www.parashift.com/c++-faq-lite/ctors.html#faq-10.6

 

[jimbo]

Hi to all of you,

first of all, thanks a lot for the explanation. I didn´t know that
this whole topic is related to where the memory is allocated. Now with
heap and stack I can imagine a little bit better what I am doing

But to come back to my own little example. When i am already in a
main() function of my app and i create the instances there without the
new operator then i will have no problems concerning the use of the
instance in a different scope because i am already in main and main is
not ended yet, or?

Yes, the scope of main is a superset of almost all other scopes
(static being the only exception that I can think of right now).

I think i need to take it more carefully when i implement an own class
and use functions there.

It's not so much when implementing a class as when using it. But since
we are on the topic of implementing classes and dynamic memory
allocation I can say that if you implementation uses new instantiate
members (perhaps in the constructor) you should probably delete them
in the destructor, like so:

class Foo {
int* arr; // An array
public:
Foo(int i) : arr(new int { } // * allocate an array of size i
~Foo() { delete[] arr; } // Free the memory used

};

This way a user of the class will never have to worry about how the
class manages its resources, it will just work.

* If you don't understand the ': arr(new int { }'-bit it's an
initializer-list and does the same thing as '{ arr = new int; }',
for a more detailed explanation seehttp://www.parashift.com/c++-faq-lite/ctors.html#faq-10.6

Thx for the explanation. I added your link to my favourites...so I
will look there the next time before I post

Have a nice day.

Best regards,

jimbo