I am quoting from the upstream bug: Currently, it's impossible to use the usual pickle mechanisms to pickle a dynamically created class, even if the user requests a different pickling mechanism via copy_reg. --------- As mentioned in the upstream bug, this behavior is needed for the proper working of sage [bug #201321]. We have tested the patch attached to the upstream bug in the sage-on-gentoo overlay without any ill effect. Of course it solved a good number of issues we had with sage.
So sage tries to use Python features that Python doesn't (yet) provide through an internally shipped and patched Python interpreter? I don't see why the python team needs to know this, but OK...
(In reply to comment #1) > So sage tries to use Python features that Python doesn't (yet) provide through > an internally shipped and patched Python interpreter? I don't see why the > python team needs to know this, but OK... > I talked to the person handling it in sage and who filled the upstream bug. He consider this more like a bug rather than a feature. This should be the behavior. Of course it is a corner case which doesn't matter for most people. Now, if we want to integrate sage into portage in the future, we want it as split components coming from the system rather than duplicating stuff in a monolithic build. python is one such component, along with quite a few python packages, it is essential for us to have this functionality or we may just as well give up. So we currently have a patched ebuild in our overlay but it would be nice if upstream and Gentoo themselves were moving it along. As mentioned it is a corner case and doesn't appear to have any side effects.
We won't accept any patches with such new features for dev-lang/python, which aren't backports from upstream repository. Please reopen this bug, when this patch has been applied in upstream repository.
This issue has now been reviewed upstream and will be included in python-2.7.3 and the next python-3.2.xx release.
2.7: http://hg.python.org/cpython/rev/64053bd79590 3.2: http://hg.python.org/cpython/rev/760ac320fa3d
2.7.3 is hitting the tree shortly.
Thus, is this bug fixed now?
Yes it is!