…sions and stats.
…is all pretty boring since I simplified the reader code.
… like delete at all. Updated my test to reflect my newfound understanding.
…es this found. I really don't like having unit tests rely on network resources, but a dedicated memcached on localhost is obviously better than not having one at all (based on my unit tests). I still don't have a way to model the various failure modes I have code to support, though.
receive. This revealed a strange condition where a server gave us an error while we were transmitting data. I don't know exactly what caused this error (SERVER_ERROR object too large for cache). I'm feeding it what I think is my largest object, but it doesn't break. Will have to experiment more.