[Development] Should qMetaTypeId<const T>() == qMetaTypeId<T>()?

[Olivier Goffart]

On Mittwoch, 31. August 2016 08:00:53 CEST Jedrzej Nowacki wrote:
> I love to be proven wrong, but this time I would not call it easy [?]  Try
> this:

We were talking about "T" vs. "const T";  not "T *" vs "const T *".
So in your example you should compare "F *" and "F * const"

The question whether F* and const F* should have the same metatype is another 
question worth asking. Since their normalized type does not compare the same 
anyway, i guess it's ok if they are different.


> 
> 
> struct F{};
> Q_DECLARE_METATYPE(F*);
> Q_DECLARE_METATYPE(const F*);
> Q_DECLARE_METATYPE(F);
> // Q_DECLARE_METATYPE(const F); doesn't compile, that is a bug!
> 
> class O : public QObject
> {
>     Q_OBJECT
> signals:
>     void foo1(const F*);
>     void foo2(F*);
> };
> 
> int main(int argc, char **argv) {
>     qDebug() << QMetaMethod::fromSignal(&O::foo1).methodSignature();
>     qDebug() << QMetaMethod::fromSignal(&O::foo2).methodSignature();
>     return 0;
> }
> 
> 
> 
> ________________________________
> From: Development <development-bounces+jedrzej.nowacki=qt.io at qt-project.org>
> on behalf of Olivier Goffart <olivier at woboq.com> Sent: Saturday, August 27,
> 2016 9:30:21 AM
> To: development at qt-project.org
> Subject: Re: [Development] Should qMetaTypeId<const T>() ==
> qMetaTypeId<T>()?
> On Mittwoch, 24. August 2016 16:06:50 CEST J?drzej Nowacki wrote:
> > On onsdag 24. august 2016 15.23.01 CEST Marc Mutz wrote:
> > > Hi,
> > > 
> > > Currently, it's not, which doesn't make much sense, does it?
> > > 
> > > So,
> > > 
> > >   template <typename T>
> > >   int qMetyTypeId<T>() {
> > >   
> > >     return qMetaTypeIdHelper<typename std::remove_cv<T>::type>();
> > >   
> > >   }
> > > 
> > > ?
> > > 
> > > (There's of course a lot more involved in this, registration should
> > > discard const, too, e.g.).
> 
> I think it make sens, but it should be done at the QMetaTypeId2 Level.
> We already do this for const T&, See commit
> 03b25125986f083a260b421abeed1f1d088d27b3
> In fact, the same should be done for const T
> 
> > From C++ perspective const T and T are kind of separate types, metatype
> > mimics that. QMetaType is used in many places and in some constness
> > matters.
> > 
> > Consider that example:
> >  QMetaType::typeName(qMetaTypeId<const QString>())
> > 
> > and
> > 
> >  QMetaType::typeName(qMetaTypeId<QString>())
> > 
> > The function is used in metaobject to compute signatures of invokables.
> 
> I can prove you wrong easily,  both have the same result, since
> QMetaType::normalizedType("const QString") == "QString"
> 
> They are different C++ type, yes,  but from a practical purpose, their use
> in the metatype system is exactly the same.  You allocate, destroy or copy
> a const QString the same way you do a QString.
> 
> --
> Olivier
> 
> Woboq - Qt services and support - https://woboq.com - https://code.woboq.org
> _______________________________________________
> Development mailing list
> Development at qt-project.org
> http://lists.qt-project.org/mailman/listinfo/development