On Thu, Sep 25, 2008 at 5:10 AM, Markus Werle <numerical.simulation@web.de> wrote:
Robert Dailey <rcdailey <at> gmail.com> writes:

> However, the above results in the following exception if I pass a pointer
> type to the 'global' parameter in my AddGlobal function. Note that the
> pointer type is that of a user defined class, like Interface*:First-chance
> exception at 0x7d4e2366 in Crusades_Debug.exe: Microsoft C++ exception:
> boost::python::error_already_set at memory location 0x0012ed44..
> Anyone know why this is happening? I haven't exposed the class in question
> (class Interface) through class_(). Could this be the problem?

Please try to compile in release mode to see if it works.
If it only crashes in Debug mode then you are a victim of
the decision to have boost::python Debug link to Python release
version and the trouble that follows from this.
See <http://www.nabble.com/-python--partially-resolved:-the-never-ending-story-
of-vc8-win32-linker-settings-desaster-td18945799.html
> about the details.

Carefully check the verbose linker output to see what happens.

We specifically compiled Boost 1.36 to link against the debug python libraries when boost debug is built. I used dependency walker to ensure it was not loading the release library.