There are several inefficiencies in the current lift-mongodb-record code mostly associated with findAll, in which multiple iterations are invoked when only one is needed. This is causing unnecessary memory usage as well as performance degradation especially with larger datasets.

It is primarily caused by a Java object implementing two conflicting interfaces which caused an ambiguity in JavaConversions. The path chosen to fix it was to add some extra conversions & iterations. Instead, it proves significantly more performant to explicitly annotate the Java object with a specific type and let JavaConversions do its thing.