Poor performance using ResultSet.TYPE_SCROLL_INSENSITIVE (3 replies)

Hi All,

I use to write my queries using ResultSet.TYPE_SCROLL_INSENSITIVE statements.
The reason is that it's easy to paginate the results and retrieve query length without retrieving all rows neither running two queries.
Also this method helps me in every database (or it was what i thank).

Here is an example of what I use to do:

----------------------------------------------------------
int start = 0; // start will be the first row to retrieve
int limit = 20; // limit is page size

String sql = "select * from table";
PreparedStatement stmt = con.prepareStatement(sql, ResultSet.TYPE_SCROLL_INSENSITIVE, ResultSet.CONCUR_READ_ONLY);
stmt.setFetchSize(limit);

rs = stmt.executeQuery();
if (rs.last()) {
// I can move to the last row and get query length
queryLength = rs.getRow();
rs.beforeFirst();
}
if (start != 0) {
// I can jump to whatever position i need without having to read all positions in the recordset
rs.absolute(start);
}

int i = 0;
// I can get only limit (pagesize) or go to the end of results
while (rs.next() && i++ < limit) {
// Here I initialize Transfer or Value Objects (the list of....)
}
----------------------------------------------------------

This peace of code performs the best on oracle, sqlserver and DB2 and is database independent (because I'm using jdbc standard methods).

The problem I have is that MySQL JConnector don't use fetchsize, instead it retrieves all rows allways
(unless you use FORWARD_ONLY, CONCUR_READ_ONLY and FetchSize(Integer.MIN_VALUE and i can't use them to achieve what I want)).
See ResultSet information at http://dev.mysql.com/doc/refman/5.0/en/connector-j-reference-implementation-notes.html.

then, there is any chance to get the submitted code to behave optimally with MySQL???
Any ideas (or I will have to change all my DAOs for MySQL using limit clause for pagination and count(*) for query length....)

thanks in advance,