I'm trying to write a C++ extension for Python (3.x) that allows me to use a specific hardware shield for the Raspberry Pi.
I haven't got any experience in writing C and/or C++, but using the documentation on the Python website, I actually got things working. What I would like to do is make my type a Singleton, but if the init-method returns with an error (-1), the second call to the constructor will mess things up. If the init returns without errors, it actually works as expected.
The important parts of my code:
static PyObject *
MyType_new(PyTypeObject *type, PyObject *args, PyObject *kwds)
{
static MyType *self = NULL; /* static to create a Singleton */
if (self == NULL) {
self = (MyType *)type->tp_alloc(type, 0);
self->is_initialized = false;
}
else {
Py_XINCREF(self);
}
return (PyObject *)self;
}
static int
MyType_init(MyType *self, PyObject *args, PyObject *kwds)
{
const char *i2c_addr;
self->rpi_model = RPI_MODEL_B;
self->enabled_components = COMPONENTS_ALL_BIT;
static char *kwlist[] = {"rpi_model", "enabled_components", NULL};
if (self != NULL && !self->is_initialized) {
if (!PyArg_ParseTupleAndKeywords(args, kwds, "|ii", kwlist, &self->rpi_model, &self->enabled_components)) {
return -1;
}
}
/*
DO SOME INITIALIZATION STUFF
*/
if (!self->component_A->testConnection()) {
PyErr_SetString(InitializationException, "Unable to communicate with component_A.");
return -1;
}
self->is_initialized = true;
return 0;
}
I also installed the following funtion in the tp_dealloc member:
static void
MyType_dealloc(PyObject *self)
{
Py_TYPE(self)->tp_free((PyObject*)self);
}
I put some print-statements in these methods (for debugging) and it seems that in the second call to MyType_new, self != NULL, although the MyType_dealloc method has been executed upon failure of the initialization.
Is anybody able to point me in the right direction?
