if operator breaks signal-slot

[DungeonLords]

if(true) makes my tim never starts... Why? If I remove if(true) then my tim again counts...

int main(int argc, char *argv[])
{
    QCoreApplication a(argc, argv);
    if(true) //remove me to work
        MainClass mainClass(QCoreApplication::instance());

    int ret= QCoreApplication::exec();
    return ret;
}

My tim code

MainClass::MainClass(QObject *parent)
    : QObject(parent)
{
    connect(&tim, &QTimer::timeout, &obj, &MyClass::mySlot);
    tim.start(1000);
}

void MyClass::mySlot(){
    static uint i;
    qDebug() << "hi" << i++;
}

[SimonSchroeder]

@DungeonLords said in if operator breaks signal-slot:

so why I should use delete or unique_ptr? Is it bad advice?

Using a parent with Qt is perfectly fine and the preferred way of handling this.

[Pl45m4]

@DungeonLords said in if operator breaks signal-slot:

if(true) //remove me to work
    MainClass mainClass(QCoreApplication::instance());

Think about the lifetime of mainClass.

This, by the way, has nothing to do with the if itself, any Signals & Slots, the timer or anything with Qt.
Not even a C++ issue, because other OOP languages also have scopes in which objects are "living" and "valid" before they are destructed.

[DungeonLords]

@Pl45m4 so

if(true) MainClass *mainClass(new MainClass(QCoreApplication::instance()));

is right solution for my case?

[Pl45m4]

@DungeonLords said in if operator breaks signal-slot:

MainClass *mainClass(new MainClass(QCoreApplication::instance()));

Why two instances of mainClass now?!
[Edit: I think I've misread that. It should also work.
I've seen two new in there. One inside the other as parent, instead of the pointer initialization]

 MainClass *mainClass = new MainClass(&a);

should be good.

Also fine:

 MainClass *mainClass;
 if ( /* condition */)
      mainClass = new MainClass;
 int ret = QCoreApplication::exec();
 if (mainClass)
     delete mainClass;
 return ret;

But why you want to put the creation inside the if scope?

[jeremy_k]

@Pl45m4 said in if operator breaks signal-slot:

Also fine:

 MainClass *mainClass;
[...]
 if (mainClass)
     delete mainClass;

Except for the uninitialized pointer access when /* condition */ is false...

[Pl45m4]

@jeremy_k

MainClass *mainClass = nullptr;
[...]
 if (mainClass != nullptr)
     delete mainClass;

:)

[aha_1980]

@Pl45m4 Still wrong ;)

MainClass *mainClass = nullptr;
[...]
delete mainClass;

Regards

[JonB]

@aha_1980
What's wrong? delete-ing a variable which has value nullptr? That's fine (i.e. guaranteed to accept and ignore nullptr; quite different from delete-ing an uninitialized variable). Whether you choose to write that or test for nullptr and not call delete is up to you.

[Pl45m4]

@aha_1980 said in if operator breaks signal-slot:

@Pl45m4 Still wrong ;)

My code inits the MainClass ptr to nullptr.
When it's properly heap allocated with new ( => mainClass != nullptr), it deletes it at the end of the code.

And if that part

if ( /* condition */)
      mainClass = new MainClass;

is skipped, mainClass is still a nullptr and you don't have to delete it?!

Explain please, what's exactly wrong with that?!

[Christian Ehrlicher]

@Pl45m4 There is nothing wrong - the check for != nullptr is just superfluous :)

[Pl45m4]

@Christian-Ehrlicher said in if operator breaks signal-slot:

the check for != nullptr is just superfluous

Ah well... yes. Because calling delete on a nullptr doesn't do anything.
So one could call delete mainClass regardless.

[SimonSchroeder]

This would be the perfect place for std::unique_ptr instead of a raw pointer (using std::make_unique<MainClass> to create the object). Then we wouldn't have all this discussion about delete... 😉

[DungeonLords]

@SimonSchroeder said in if operator breaks signal-slot:

std::unique_ptr instead of a raw pointer

In my code I use parent

MainClass *mainClass(new MainClass(QCoreApplication::instance()));

so why I should use delete or unique_ptr? Is it bad advice?

[SimonSchroeder]

@DungeonLords said in if operator breaks signal-slot:

so why I should use delete or unique_ptr? Is it bad advice?

Using a parent with Qt is perfectly fine and the preferred way of handling this.