An informative post that attempts to compare accessor speed for different object systems (and some non-object systems). The benchmarking is hardly perfect (read the comments for complaints) but I think it may spark a good discussion.