[Jython-users] RE: Issue with None and Serialization

[brian z. <bz...@zi...>]

Chuck,

Why can't you assume cPickle is available?  It's standard in the Jython
dist as a Java-implemented module.  I have modified your example a bit
to use it and it seems to behave properly.  Add the following two
methods:

--- code ---

import cPickle

def load_with_pickle(path):
	fp = open(path, "rb")
	try:
		data = cPickle.load(fp)
	finally:
		fp.close()
	return data

def save_with_pickle(obj, path):
	fp = open(path, "wb")
	try:
		cPickle.dump(obj, fp)
	finally:
		fp.close()

 ... 

print "Testing initial object..."
a = Test()
a.test()
save_with_pickle(a, "test.out")
b = load_with_pickle("test.out")
print "Testing deserialized object..."
b.test()

--- results ---

$ jython cc.py
Testing initial object...
attr is None: 1
attr type is: org.python.core.PyNone
attr == None: 1
Testing deserialized object...
attr is None: 1
attr type is: org.python.core.PyNone
attr == None: 1

Nonetheless, I see the same problem you do with the code snippet.

brian

> -----Original Message-----
> From: ch...@er... [mailto:ch...@er...] On 
> Behalf Of chuck clark
> Sent: Tuesday, March 12, 2002 7:32 PM
> To: jyt...@li...
> Cc: bu...@cu...
> Subject: Issue with None and Serialization
> 
> 
> I'm using Jython for my project and today we tried adding the
> functionality where you can save out a Config object and then when you
> restart you can load a Config if you need it.  I'm using the java.io
> libraries to do this because I can't assume pickle or any of the other
> python libraries are on the machine.
> 
> However I ran into an issue where after reading the 
> serialized object in
> some statements which checked to see if an attribute on the object is
> None were failing.  I've recreated the problem in the code 
> snippet below
> as succinctly as I can.  Does anyone have any ideas on what I'm doing
> wrong?  I have a feeling it must be something really simple or could
> this be a bug in Jython?
> 
> TIA
> chuck
> 
> -------code-------
> from java.io import *
> from org.python.util import *
> 
> class Test(Serializable):
>     def __init__(self):
>         self.attr = None
>     def test(self):
>         print "attr is None:", self.attr is None
>         print "attr type is:", type(self.attr)
>         print "attr == None:", self.attr == None
> 
> def load(path):
>     file = File(path)
>     fileIn = FileInputStream(file)
>     pyIn = PythonObjectInputStream(fileIn)
>     pyObj = pyIn.readObject()
>     pyIn.close()
>     return pyObj
> 
> def save(obj, path):
>     fileOut = FileOutputStream(path)
>     objOut = ObjectOutputStream(fileOut)
>     objOut.writeObject(obj)
>     objOut.flush()
>     objOut.close()
> 
> print "Testing initial object..."
> a = Test()
> a.test()
> save(a, "test.out")
> b = load("test.out")
> print "Testing deserialized object..."
> b.test()
> 
> -----results----
> Testing initial object...
> attr is None: 1
> attr type is: org.python.core.PyNone
> attr == None: 1
> Testing deserialized object...
> attr is None: 0
> attr type is: org.python.core.PyNone
> attr == None: 0
> 
> I would expect that if the attribute is of Type PyNone then it should
> return 1 to
> attr is None
> and
> attr == None
> 
> I thought the problem might be that between 
> serializing/deserializing I
> might somehow be getting a different instance of PyNone but this isn't
> the case as evidenced below when I inspect the objects in the
> interpreter after running th code above
> >>> type(a.attr)
> <jclass org.python.core.PyNone at 334527779>
> >>> type(b.attr)
> <jclass org.python.core.PyNone at 334527779>
>