>>>>> "[EMAIL PROTECTED]" == [EMAIL PROTECTED] com <[EMAIL PROTECTED]> writes:
[EMAIL PROTECTED]> whenever you are using a package that leaks memory.
[EMAIL PROTECTED]> it can be appropriate to use Rpyc
[EMAIL PROTECTED]> (http://rpyc.wikispaces.com/) to run the leaking
[EMAIL PROTECTED]> code in a different process, and restart it from
[EMAIL PROTECTED]> time to time. I've been using this method to avoid
[EMAIL PROTECTED]> the leaks of matplotlib.
The only known leak in matplotlib is in the tkagg backend which we
believe comes from tkinter and is not in matplotlib proper. There are
a variety of ways to make it look like matplotlib is leaking memory,
eg overplotting when you want to first clear the plot, or failing to
close figures properly. We have unit tests to check for leaks, and
they are passing. Perhaps you can post some code which exposes the
problem.
JDH
--
http://mail.python.org/mailman/listinfo/python-list
