TODO: xray objects support the pickle protocol

[shoyer]

I actually don't know if our current objects support pickle. But we should add tests and make sure this works, for a dead simple form of inter-process communication.

[ebrevdo]

In [73]: h = cPickle.dumps(f)
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-73-a875f963f7d4> in <module>()
----> 1 h = cPickle.dumps(f)

/System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/copy_reg.pyc in _reduce_ex(self, proto)
     68     else:
     69         if base is self.__class__:
---> 70             raise TypeError, "can't pickle %s objects" % base.__name__
     71         state = base(self)
     72     args = (self.__class__, base, state)

TypeError: can't pickle Variable objects

[ebrevdo]

http://docs.python.org/2/library/pickle.html#pickling-and-unpickling-normal-class-instances

[shoyer]

Something to consider is whether we should always store the data as an ndarray before pickling an XArray, or whether we should just store a reference to the original sources (e.g., make a pickle-able version of netCDF4.Variable). I am inclined to the former so pickle can be used a simple serialization format, but Iris has implemented the later approach.

[shoyer]

Some investigation reveals that this is only an issue when pickling datasets containing netCDF4.Variable objects (not xray.Variable objects). So the simple fix should be to make sure that all values are loaded into memory before pickling an xray.Variable.