It looks like this has been fixed on the trunk and on the 3.8 branch. At least, there's no _p_deactivate call on line 511 of serialize and there is an explict assignment of _p_changed to None.
Could you please verify the fix or provide more information?
It looks like this has been fixed on the trunk and on the 3.8 branch. At least, there's no _p_deactivate call on line 511 of serialize and there is an explict assignment of _p_changed to None.
Could you please verify the fix or provide more information?