PPL Variables cannot be saved

[yuan-zhou]

It seems impossible to save an object of the Parma Polyhedra Library's Variable class and then load it.

sage: from sage.libs.ppl import Variable
sage: x = Variable(0)
sage: x.save(os.path.join(SAGE_TMP, 'pplvariable'))
---------------------------------------------------------------------------
AttributeError: 'sage.libs.ppl.Variable' object has no attribute 'save'

sage: save(x, os.path.join(SAGE_TMP, 'pplvariable'))
sage: load(os.path.join(SAGE_TMP, 'pplvariable.sobj'))
---------------------------------------------------------------------------
TypeError: __cinit__() takes exactly 1 positional argument (0 given)

CC: @mkoeppe

Component: geometry

Issue created by migration from https://trac.sagemath.org/ticket/19592

[yuan-zhou]

Description changed:

--- 
+++ 
@@ -1,4 +1,4 @@
-It seems impossible to save an object of the Parma Polyhedra Library's `Variable` class.
+It seems impossible to save an object of the Parma Polyhedra Library's `Variable` class and then load it.
 
 ```
 sage: from sage.libs.ppl import Variable
@@ -6,4 +6,9 @@
 sage: x.save(os.path.join(SAGE_TMP, 'pplvariable'))
 ---------------------------------------------------------------------------
 AttributeError: 'sage.libs.ppl.Variable' object has no attribute 'save'
+
+sage: save(x, os.path.join(SAGE_TMP, 'pplvariable'))
+sage: load(os.path.join(SAGE_TMP, 'pplvariable.sobj'))
+---------------------------------------------------------------------------
+TypeError: __cinit__() takes exactly 1 positional argument (0 given)
 ```

[mkoeppe]

comment:2

After the switch to pplpy, already save fails:

sage: from ppl import Variable                                                                                                                                
sage: x = Variable(0)                                                                                                                                         
sage: save(x, os.path.join(SAGE_TMP, 'pplvariable'))                                                                                                          
TypeError: no default __reduce__ due to non-trivial __cinit__