Issue454586

Unpickling of large lists or multiple loads give the 
wrong results.

Attached is a example.
Environment JAVA JDK 1.3, Windows 2000, jython 21a3

Logged In: NO 

Forgot the example:
------------------

import pickle

pfile=open("ptest.pi","wb")
p=pickle.Pickler(pfile)
for l in range (1,10000):
    row=[str(l),str(l)]
    p.dump(row)
pfile.close()

print "reading"
n=1
try:
    pfile=open("ptest.pi","rb")
    l=pickle.load(pfile)
    while l:
        comp = [str(n),str(n)]
        if l != comp:
            print "Pickle error"
            print str(l) + " should be " + str(comp)
        n=n+1
        l=pickle.load(pfile)
    pfile.close()

except EOFError:
    print "End reached, well done"
    pfile.close()

Logged In: YES 
user_id=4201

To be correct, it is actually pickle.dump() that writes an 
incorrent pickle to the file. The problem is caused by bug 
#222789 where the id() of jython objects is not unique and 
pickle.py expects all id() values to be a unique integer.

The bug does not exists in cPickle, so the workaround is to 
use that instead of pickle.py. A real fix might be possible 
but the solutions we have come up with so far, would 
increase memory use for all objects. It migth never be 
fixed for real.

I'm going to close this report because the problem is 
already described in #222789.