current state of KDE_DEPRECATED ?

[Nicolas Goutte]

On Thursday 16 March 2006 04:37, Nicolas Goutte wrote:
> On Thursday 16 March 2006 00:48, Nicolas Goutte wrote:
> > On Wednesday 15 March 2006 23:43, Nicolas Goutte wrote:
> > > On Wednesday 15 March 2006 19:38, Alexander Neundorf wrote:
> > > > Hi,
> > > >
> > > > I get a lot of compiler errors due to KDE_DEPRECATED used in front of
> > > > a ctor with gcc 3.3.6 .
> > > > Is this the intended usage ?
> > >
> > > I have looked closer and can indeed confirm the problem with gcc 3.3.1.
> > > Using Q_DECL_CONSTRUCTOR_DEPRECATED does not seem to help. (So
> > > currently I wonder why QT3_SUPPORT_CONSTRUCTOR seems to work in Qt4.)
> >
> > I have understood now. QT3_SUPPORT_CONSTRUCTOR is empty, as QT3_SUPPORT
> > is defined while building Qt4.
> >
> > So in fact this is a bug in Qt4, which should not define
> > Q_DECL_CONSTRUCTOR_DEPRECATED for gcc 3.3.x like for gcc 3.4.x and
> > higher.

For the archive: Qt Issue N107077

>
> Now I see that I get deprecated warnings for KDE code using Qt deprecated
> functions. So somehow it works in Qt, however why does it not work in KDE?

I have found the subtle difference: it is the keyword "explicit".

That works:
  __attribute__((deprecate)) explicit

That does not:
  __attribute__((deprecated))

>
> > > > Should KDE_DEPRECATED simply be defined to nothing with gcc < 3.4 ?
> > >
> > > Currently I think that we should create a KDE_CONSTRUCTOR_DEPRECATED
> > > and indeed disableit for gcc 3.3.x.
> > >
> > > > Bye
> > > > Alex
> > >
> > > Have a nice day!

Have a nice day!