Fredrik Lundh wrote: > [EMAIL PROTECTED] wrote: > > > I've been working on an external C module for Python in order to use > > some of the functionality from Ethereal. Right now I'm getting > > segfaults originating from within the Ethereal source code, but the > > same code works fine when used normally (i.e. through Ethereal or > > Tethereal). I'm wondering if there's some sort of memory limitation the > > Python interpreter imposes on the C code it calls that I might be > > running into > > no.
Well, that answers that. I know memory limits can be an issue with the Java VM, at least it was back when I was taking Java classes in college and using version 1.1. So I thought there might be something similar going on with Python, but I guess not. > I'd suspect, in order: > > you've haven't initialized/configured the Ethereal library correctly > you're using the Ethereal API in the wrong way > you're accidentally overwriting some memory area > you're messing up the reference count for some python object > you're using the Python API in the wrong way > some other bug in your code > a bug in the part of the ethereal API you're using > a bug in your code > a bug in the Python API you're using > a bug in your code I was able to modify my C code so that instead of being a Python module, it runs as a standalone binary, and it works as it should. Calling it with os.spawn* works. The two versions are essentially the same, the primary differences being the necessary difference in how arguments and return values are handled. This will work if necessary, but I would think having it as a Python module would be slightly more elegant and efficient since we avoid the overhead of setting up new processes. In case it helps anyone, I'm running Python 2.4.1 on Fedora Core 4, and working with Ethereal 0.10.14. I'm using the Tethereal source code as a guide to how to call into the Ethereal code. - David -- http://mail.python.org/mailman/listinfo/python-list
