STINNER Victor <victor.stin...@haypocalc.com> added the comment:
> I just looked at the code for PyFloat_AsDouble and it checks for NULL pointers
It's just a little help for developers: it will raise the ugly TypeError("bad
argument type for built-in operation") and returns -1.
It's never a good idea to pass NULL to a function of the Python C API. I think
that most (or all?) functions will fail.
> IMHO, the C-API documentation lacks information on how it behaves
> in error conditions (this might be a bit over generalized).
I didn't read the documentation, but you HAVE TO always check the result of a
Python function.
If the function returns a PyObject*, you have to check that the result is not
NULL. The same rule applies for Pyxxx_AsXxx(): you also have to check the
result (the error value depends on the function). Eg. PyInt_AsLong() returns -1
on error: use (result == -1) && PyErr_Occured()) in the caller to detect an
error.
----------
nosy: +haypo
_______________________________________
Python tracker <rep...@bugs.python.org>
<http://bugs.python.org/issue8113>
_______________________________________
_______________________________________________
Python-bugs-list mailing list
Unsubscribe:
http://mail.python.org/mailman/options/python-bugs-list/archive%40mail-archive.com
