Help rolling own lambda connect for Qt 4.8

[KroMignon]

@SPlatten Your welcome, I'm curious if this has been useful to you.

[SPlatten]

@KroMignon Trying it now...will let you know when I've tested it.

[SPlatten]

@KroMignon , it could the very old tools that this company is using:

gcc (Gentoo 4.6.3 p1.13,pie-0.5.2) and Qt 4.8.4

When I try to compile including the <functional> header which is present I get:

../clsLambda.h:29:5: error: 'function' in namespace 'std' does not name a type
../clsLambda.h:32:28: error: expected ')' before '<' token
../clsLambda.h: In member function 'void clsLambda::slot()':
../clsLamdba.h:37:24: error: 'mpFN' was not declared in this scope
make: *** [moc_clsLambda.o] Error 1
12:27:31 The process "/usr/bin/make" exited with code 2

The header:

#include <functional>
#include <QObject>

class clsLambda : public QObject {
    Q_OBJECT

private:
    std::function<void()>mpFN;                                    //This is line 29

public:
    clsLambda(std::function<void()> pFN, QObject* pParent = 0)    //This is line 32
            : mpFN(pFN) {
    }
public slots:
    void slot() { if ( mpFN ) mpFN(); }                           //This is line 37
};

[SPlatten]

I manged to remove the compiler warnings and compile by changing the class to:

#include <QObject>

class clsLambda : public QObject {
    Q_OBJECT

private:
    void(*mpFN)();

public:
    clsLambda(void(*pFN)()) : mpFN(pFN) {
    }
public slots:
    void slot() { if ( mpFN ) mpFN(); }
};

[KroMignon]

@SPlatten said in Help rolling own lambda connect for Qt 4.8:

When I try to compile including the <functional> header which is present I get:

Did you enable C++11 support?
I guess you have to add -std=c++11 in CXXFLAGS.
With qmake, you can add in pro file QMAKE_CXXFLAGS += -std=c++11
Or it may also be -std=c++0x, depending on GCC version.

[J.Hilk]

@SPlatten you could use verdigris
https://github.com/woboq/verdigris

that should work, but no guarantee :D

[SPlatten]

@KroMignon , I've modified the pro file:

QT += core gui
CONFIG += c++11
CONFIG += sdk_no_version_check
QMAKE_CXXFLAGS += -std=c++11
...

When I try to build I get:

ccplus: error: unrecognized command line option '-std=c++11'

[SPlatten]

@J-Hilk , I cannot access that site as its blocked by the corporate firewall.

[J.Hilk]

@SPlatten GitHub ?

[SPlatten]

@J-Hilk I clicked on link and was met with:

Error Message: Access to the requested URL has been denied (blocked-category-message)

[KroMignon]

@SPlatten said in Help rolling own lambda connect for Qt 4.8:

When I try to build I get:
ccplus: error: unrecognized command line option '-std=c++11'

You are using GCC 4.6 (which is very old!), so you should use -std=c++0x to enabled C++11 support (cf. https://gcc.gnu.org/gcc-4.6/cxx0x_status.html)

[J.Hilk]

@SPlatten well, here's the general website:
https://woboq.com/blog/verdigris-qt-without-moc.html

[SPlatten]

@KroMignon , its not working, with the changes to the pro file so builds without problems and the clsLambda now like:

class clsLambda : public QObject {
    Q_OBJECT

private:
    std::function<void()>mpFN; 

public:
    clsLambda(std::function<void()> pFN) : mpFN(pFN) {
    }
public slots:
    void lambdaSlot() { if ( mpFN ) mpFN(); }
};

The test code:

clsLambda objLambda([&]() {
    qDebug() << "HERE";
});
bool blnRC(QObject::connect(pobjAct, SIGNAL(triggered()), &objLambda, SLOT(lambdaSlot())));
qDebug() < blnRC;

blnRC is always true, but when a menu is selected there is no sign of HERE in the Application Output.

[Christian Ehrlicher]

@SPlatten said in Help rolling own lambda connect for Qt 4.8:

clsLambda objLambda(& {
qDebug() << "HERE";
});

C++ basics - how long does this object live?

[SPlatten]

@Christian-Ehrlicher , the lambda is an inline function, what do you mean?

[SPlatten]

@Christian-Ehrlicher , sorry, I know exactly what you meant and I will change the code to allow for a life of the object outside of the initial set-up, the example on the original link:
https://silmor.de/qtstuff.lambda.php

Is IMHO wrong and I agree with you.

[KroMignon]

@SPlatten said in Help rolling own lambda connect for Qt 4.8:

the lambda is an inline function, what do you mean?

with [&](){} you are capturing any current variable by reference... it'is not the best/cleanest way to do ;)

[jsulm]

This post is deleted!

[SPlatten]

@Christian-Ehrlicher , thank you for the nudge in the right direction:

Added to the class:

QList<clsLambda*> mlstLambdas;

In the code:

clsLambda* pobjLambda(new clsLambda([pobjAct]() {
    const QString cstrMID(pobjAct->property(clsMainWnd::mscszPropertyMenuID));
    qDebug() << cstrMID;
}));
mlstLambdas.append(pobjLambda);
QObject::connect(pobjAck, SIGNAL(triggered()), pobjLambda, SLOT(lambdaSlot())));

It works really well.

[SimonSchroeder]

@SPlatten Since clsLambda already inherits from QObject I would use this as parent during creation. Then you don't have to keep a list and think about when to delete and remove lambdas from the list.

Also, you have so much code to write just to use a lambda. This would push me to write the slots in the original way. Maybe you want to implement a macros that will do all the extra work (I'm not entirely sure if you could easily pass a lambda as argument to the macro). I would prefer to write this:

CONNECT_LAMBDA(pobjAck, SIGNAL(triggered()), [pobjAct]() {...});

In order to pick a unique name for pobjLambda you can use __LINE__ appended to your name which makes the name of the lambda object unique within each file (which is sufficient for a variable with local scope).

From the top of my head (untested):

#define CONNECT_LAMBDA(OBJ, SIG, LAMBDA) \
clsLambda* pobjLambda##__LINE__ (new clsLambda(LAMBDA)); \
mlstLambdas.append(pobjLambda##__LINE__) \
QObject::connect(OBJ, SIG, pobjLambda##__LINE__, SLOT(lambdaSlot()));

I believe that __LINE__ is the line the macro is used in, such that even the macro is multiple lines it always evaluates to the same line.