- Major refactor of how data is stored, which fixes a number of bugs which were creating duplicate entries
- _id field is now generated by cache to guarantee no dup entries, instead of using objectID. See the docs for details.
- Moved to official MongoDB URI format for connections, allows slaveok parameter
- Fixed more outstanding issues; bad code was making lookups fail.
- Added support for Sparse collections, in which each key is a row.
- For Beaker Sessions, this already happens automatically, but in say Pylons Caching of a specific template, it puts it all in one place. u
- This forces it to split, and should help performance.