I've been looking at the use of pickle on the memcache side only, and it looks to me that moving to JSON there shouldn't be an issue -- I don't see any of the user calling it with really complex object types.
The only issue I can see with the default encoder/decoder is that some types will be changed after serialization/deserialization (for instance, tuples become lists, str becomes unicode, etc.). But I don't think it's something that will be a problem in the code.
I've been looking at the use of pickle on the memcache side only, and it looks to me that moving to JSON there shouldn't be an issue -- I don't see any of the user calling it with really complex object types.
The only issue I can see with the default encoder/decoder is that some types will be changed after serialization/ deserialization (for instance, tuples become lists, str becomes unicode, etc.). But I don't think it's something that will be a problem in the code.