It would be great if it was possible to define indexes on an output collection of a map/reduce before the permanent output collection was swapped.

This would be very useful for scenario's where an application uses map/reduced data that's re-generated periodically to for example display pre-calculated data to users. There are use cases where you want to select data in a huge map/reduced output and need indexes to make this perform well. I could supply our use case if that's interesting to you.

If the collection would be swapped by the periodical updater after an index is generated performance would not be degraded by indexing because the index would already be in place.