Re: non-local static object initialization problems

[Ali]

Hello,

Thank you for your valuable reply. I got the following link from an
other group comp.lang.c++ (many thanks go to user "red floyd")

http://www.parashift.com/c++-faq-lite/ctors.html

My problem is discussed under 10.12-10.16, togehter with its solution.
According to 10.15

"Note that the ANSI/ISO C++ committee is working on this problem, but
compilers aren't yet generally available that handle these change"

I have tried g++ 4.1 but i got segmentation fault due to the
uninitilaized pointer.

Thanks,

Ali


Bernd Strieder írta:
> Ali wrote:
>
> > My problem is quite similar to Item 10 of Effective C++ (Scott
> > Meyers):
> >
> >
> http://www.awprofessional.com/content/downloads/meyerscddemo/DEMO/EC/EI10_FR.HTM
> >
> > I have Airplane objects declared in a namespace myNamespace in a file
> > main.cpp, the memPool is initialized correctly. Everything works fine
> > on Win32 platform with MinGW GCC 3.4.4, Borland C++ Builder 6, Visual
> > Studio 2003 and Visual Studio 2005.
> >
> > On Linux with GCC 3.3.5 and Intel C++ Compiler 8.1 i get segmentation
> > fault, i could debug the reason: memPool is not yet initialized when
> > the constructor is called for the Airplane objects in myNamespace.
> >
> > Could anyone explain why?
>
>
> There might be workarounds involving the order of definitions or the
> linking order.
>
> Maybe you could try a newer gcc release. gcc 3.4 and up include lots of
> fixes wrt. C++. The old Intel compiler might have had to behave exactly
> like the old gcc compiler to be compatible. This might explain, why it
> works with Mingw gcc 3.4.4.
>
> If this does not suffice, then please post minimal compileable code for
> reproducing the problem.
>
> Bernd Strieder