On Mon, Oct 11, 2010 at 1:43 PM, R. David Murray <report@bugs.python.org> wrote:
R. David Murray <rdmurray@bitdance.com> added the comment:
Can you reproduce the problem using a pure Python 'corrupt' module?
----------
 
When in python.exe, and you type help() then modules, aren't you really asking for packages?
I don't know how to make a package let alone a corrupt one.
 
If help() modules is in fact looking for just a Python module file with extension _.py in lib\site-packages I don't know what Python help() modules looks for in that source code _.py file to be considered a module worth reporting back to the user. Certain module attributes?
 
I do not know enough atm about how PythonWin is packaged. I'll look into it but I'm a newbie here and hopes for well informed facts on PythonWin may be ... well leave it at that.
 
Perhaps an idea here too: If python.exe help() modules crashes when it passes control to a module or package that is not pure Python (source code or byte code) but some kind of C SWIG thing, then perhaps to make python.exe more robust it should simulate what idle does, as idle does not crash and it reports error when finished with calling python.exe help() modules.
 
In other words, perhaps the Python interpreter can call an external routine (another program) that behaves like help() modules, perhaps using something like os.system() or subprocess.Popen(). Call it help.py or help.exe or something. If help.exe returns, all is good. If help crashes just like Python did at least Python can handle the crashing of help.exe gracefully. This idea is purely with the interest to make python.exe more robust.