Right way to initialize python embedded in a multi-threaded application

[shooshx]

I'm embedding python in a multi-threaded C application.
I've taken care to wrap every call to the Python C API with

gstate = PyGILState_Ensure();
// call python code
PyGILState_Release(gstate);

But I'm stumped with what to do in the initialization.
Right after the call to Py_IsInitialized() I've added a call:

PyEval_InitThreads();

The docs say that this function leaves the GIL locked when it returns.
I do some more initializations like importing modules and then I call

PyEval_ReleaseLock();

This seems to cause a problem since not long after a call to
PyGILState_Release(gstate) that's made in a different thread crashes.
with

"Fatal Python error: This thread state must be current when releasing"

If I don't do the call to PyEval_ReleaseLock() in the main thread
right after initialization, the GIL seems to be released
after the first PyGILState_Ensure() - PyGILState_Release() pair.

So what am I doing wrong here?
What is the correct way of initializing a multi-threaded application?

[francis...@gmail.com]

Hi,

Maybe you already fixed the issue, but for the record, I've got the same
problem and finally it turned out that I was calling PyEval_InitThreads twice
and also after fixing that, I also had to move the call to PyEval_ReleaseLock(); at the end of the entire initialization (not just after PyEval_initThreads).

The key thing there is to follow:

<<at initialization thread>>
Py_Initialize ();
PyEval_InitThreads();
/* now call here to initialize all python code by loading external files
or internal module loading (i.e. Py_InitModule3) */
/* ..and now, once no more Python C/API call is required, release
the GIL so other threads can come into play */
PyEval_ReleaseLock ();

<<and now, from other threads, use>>
/* wait til gil acquired */
state = PyGILState_Ensure();
/* your code */
/* release GIL */
PyGILState_Release (state);

Hope it helps, Cheers!

[francis...@gmail.com]

Just clarify there's no problem about calling twice to PyEval_InitThreads ()
as indicated by Python's doc.