We work with in-memory database and I was very pleased to see that iBatis adds almost no overhead compared to pure JDBC except when we use column="someColumn" in our resultmaps. So we have to use columnIndex which, of course, is a bit too inconvenient.
Simple profiling hints that the difference arises because of multiple calls to Character.toUpperCase. I suspect those calls are to match database column by name. If it's correct, may ask for a feature that will accelerate using column names, say, by translating all column names into indexes and caching them at first execution of queryForXXX, so that next executions will not need any string comparisons? TYA
