The (de)serialization currently uses Pickle. This breaks when upgrading the classes of the stored objects, since the deserializer can't find the newly added fields in the file.
We need a backward compatible deserialization process. This probably means writing a (de)serializer explicitly instead of relying on pickle. (This has an added security advantage; unpickle-ing isn't very safe to do if the source is not trusted)
The (de)serialization currently uses Pickle. This breaks when upgrading the classes of the stored objects, since the deserializer can't find the newly added fields in the file.
We need a backward compatible deserialization process. This probably means writing a (de)serializer explicitly instead of relying on pickle. (This has an added security advantage; unpickle-ing isn't very safe to do if the source is not trusted)
See also: #69