Details
-
Question
-
Resolution: Done
-
Major - P3
-
None
-
None
-
None
-
None
Description
Currently we have different matching semantics for sparse vs non sparse indexes (since sparse indexes don't contain documents lacking the indexed field, those are not returned). With cursor interleaving in the query optimizer, if two indexes are used and one is sparse while the other is not, some documents may be loaded from each index. The resulting matching semantics will not be the same as the current sparse or non sparse matching semantics.