PyErr_SetString() != raise ?

[Donn Cave]

In article <bevkb3$mg8$1 at solaria.cc.gatech.edu>,
 David Eger <eger at cc.gatech.edu> wrote:

> When I use PyErr_SetString() in an extension, python just keeps on
> chugging instead of acting as though a Python exception had been
> 'raise'd.  Why is this, and do I really have to write the Python code
> to raise an exception manually after I've used PyErr_SetString()?
> 
> Most recently, writing an iterator -- I write:
> 
>     if (nextEl == NULL) {
>       PyErr_SetString(PyExc_StopIteration, "Array index out of bounds");
>       return NULL;
>       }
> 
> in my iterator extension, but when I use my extension, I get an infinite loop:
...
> (and so on, and so on...)  The thing is, even though I *set* the
> exception with PyErr_SetString, to the interpreter, it doesn't get
> raised.  Am I just misinterpretting how PyErr_SetString is supposed 
> to work?

Sounds like it.  PyErr_SetString installs the exception, but it
doesn't raise it.  The most common way a C function causes an
exception to be raised is to return 0 (or NULL, if you prefer)
where a non-zero (PyObject *) is expected.  So you are doing that,
but I don't know if it's effective in that particular context.
Some functions have other conventions, for example type initialization
returns 0 for success.

More research required.  At any rate, don't worry about whether
the exception has been initialized, at the moment that is not
your problem.

   Donn Cave, donn at u.washington.edu