> Since copy_reg lets you specify arbitrary code to serialize arbitrary > objects, you shouldn't run into any single object that you cannot > serialize to a pickle.
In http://www.effbot.org/librarybook/pickle.htm, it specifically mentions that code objects cannot be pickled and require the use of copy_reg, as follows: import copy_reg import pickle, marshal, types # # register a pickle handler for code objects def code_unpickler(data): return marshal.loads(data) def code_pickler(code): return code_unpickler, (marshal.dumps(code),) copy_reg.pickle(types.CodeType, code_pickler, code_unpickler) # # try it out CODE = """ print "suppose he's got a pointed stick" """ code = compile(CODE, "<string>", "exec") exec code exec pickle.loads(pickle.dumps(code)) I cannot understand 2 things, which I seek assistance for: 1. Is code object the only thing can cannot be pickled (less facing recursion limits)? 2. In the above example, how copy_reg works with pickle? Thanks and Cheers Maurice > > However, both pickle implementations are recursive, so you will be > limited by the amount of memory you can allocate for your stack. By > default, this will limit you to something like object graphs 333 edges > deep or so (if I'm counting stack frames correctly). Note that this > does not mean you cannot serialize more than 333 objects at a time, > merely that if it takes 333 or more steps to go from the first object to > any other object in the graph (using the traversal order pickle uses), > the pickling will fail. You can raise this limit, to a point, with > sys.setrecursionlimit(). > > Jean-Paul -- http://mail.python.org/mailman/listinfo/python-list
