Re: [Libreoffice] WaE on vcl because of qt4 module

On Friday 25 of November 2011, julien2412 wrote:
> Hello,
>
> By compiling vcl module (Debian testing x86-32, WaE, debug and symbols
> enabled), I've got this :
> In file included from /usr/include/qt4/QtCore/qdatastream.h:45:0,
>                  from /usr/include/qt4/QtCore/qpair.h:45,
>                  from /usr/include/qt4/QtCore/qhash.h:49,
>                  from
> /home/maryline/compile-libreoffice/libo/vcl/unx/kde4/KDEXLib.hxx:35,
>                  from
> /home/maryline/compile-libreoffice/libo/vcl/unx/kde4/KDEData.cxx:31:
> /usr/include/qt4/QtCore/qscopedpointer.h: In static member function ‘static
> void QScopedPointerDeleter<T>::cleanup(T*)’:
> /usr/include/qt4/QtCore/qscopedpointer.h:55:5: error: declaration of
> ‘pointer’ shadows a global declaration [-Werror=shadow]

 A really helpful error message from gcc. The argument of the Qt function 
shadows an unknown global declaration and gcc does not feel like saying where 
it is. As long as you don't compile the file with just -E and find in the 
preprocessed source where the global 'pointer' comes from, this is just 
guessing, but I think it's rather good guessing that Qt is innocent here and 
the problem is wherever the global 'pointer' is. I do not see any 
global 'pointer' indentifier in vcl, so presumably something pollutes the 
global namespace and that is what should get fixed.

 As for the Qt bugreport you've made, please close it. First of all, nobody 
will fix it simply because it's most probably invalid and definitely 
unhelpful (do you realize that you are asking Qt developers to rename a 
function argument to something that does not clash with anything in the whole 
of LibreOffice and everything it uses?). Moreover, in Qt-based land -Whadow 
is (AFAICT) usually considered a nuisance rather than help (and this shows 
why), so I expect nobody will even seriously look at it anyway, for a reason.

-- 
 Lubos Lunak
 l.lunak@suse.cz