mroonga: prefer deferring full text on the same node for better cache efficiency
In this specific case, the number of grouped messages is bound, and actually not greater than the number of grouped messages being processed for normal indexing.
Showing
Please register or sign in to comment