[omniORB] omniORBpy build problem on Solaris 2.7

[Harri Pasanen]

omniORB2_8_develop branch seems to have working gcc support Solaris 7,
thank you Duncan for notifying me of this.

I still had the "_Py_NoneStruct: referenced symbol not found" problem
below though.  After a lot of digging I finally isolated this to be a
problem with Python's autoconf on Solaris 7 when gcc 2.95.2 is used.  It
failed to pass the "-Xlinker --export-dynamic" to gcc when linking
Python itself. After fixing this all seems to be well.


Harri Pasanen wrote:
> 
> Solaris 2.7
> 
> Looks like the omniORBpy makefiles (dir.mk)'s on SunOS are geared to use
> Sun's C++ compiler.
> 
> We have gcc 2.95.2 installed, which builds omniORB2 itself without
> problems.
> 
> I figured that the Linux target would be closer than Sunos in
> omniORB_280/src/lib/omniORBpy/omniidl/cxx/dir.mk if gcc is used.
> Indeed, it builds _omniidlmodule.so without problems, but at runtime I
> get:
> 
> bash-2.03$ python
> Python 1.5.2 (#20, Jan 21 2000, 09:31:23)  [GCC 2.95.2 19991024
> (release)] on sunos5
> Copyright 1991-1995 Stichting Mathematisch Centrum, Amsterdam
> >>> import _omniidlmodule
> Traceback (innermost last):
>   File "<stdin>", line 1, in ?
> ImportError: ld.so.1: python: fatal: relocation error: file
> /work/harri/src/omniORB_280/lib/sun4_sosV_5.7/_omniidlmodule.so: symbol
> _Py_NoneStruct: referenced symbol not found
> >>>
> 
> My Python's main is not compiled with g++, rather with gcc.  I'm not
> sure if this is related to that.
> 
> Before I pursue the debugging further, I'd like to know if someone has
> already been there, done that?
> 
> Regards,
> 
> Harri