Howto get RowCount of ResultSet

Question

Hello,
is there an intelligent way of getting the rowcount of an resultset?
For now I execute 2 SQLQueries, where the second is just executed to get the amount of rows in this selection, which is a lot of sumb overhead:
ResultSet rs = db.getStatement().executeQuery("SELECT * FROM BOOKS WHERE LOWER(titel) like '%" + searchStr + "%' OR LOWER(autor) like '%" + searchStr + "%' OR LOWER(instrument) like '%" + searchStr + "%' ORDER BY " + orderTableName + ";");
               ResultSet countSet = db.st2.executeQuery("SELECT COUNT(*) FROM BOOKS WHERE LOWER(titel) like '%" + searchStr + "%' OR LOWER(autor) like '%" + searchStr + "%' OR LOWER(instrument) like '%" + searchStr + "%';");
Any ideas howto do this in a better way?
Thank you in advance, lg Clemens

Answer 1

[Standard pre-packaged copy&pasted rant follows]
The really really best way to know the number of rows in a ResultSet: write your application so that you don't need to know it.
To get the number of rows in a ResultSet, loop over the result set with while(res.next()). Read each row into an object. Add each object into e.g. a LinkedList. At the end, you will have the rows nicely converted to objects (which you usually should do anyway). And list.size() will give the number of objects.
There are other ways to get the number of objects, but list.size() is the easy and reliable way. If you don't need the rows converted to objects ...why did you select them in the first place?
Tricks with last()/getRow() sort of work. They read the entire result set into memory (in a memory-inefficient way: the storage that scrollable result sets take is almost certainly more than "real" objects). But you'll need to write the while(res.next()) loop anyway. No point in having the computer do the same thing twice. So last()/getRow() is inefficient and just plain extra silly work.
Still thinking of using last()/getRow()? Bad idea. When you execute a select, the database server doesn't necessarily even know the number of rows you selected (the cursor "materializes" as it is being read). The server will hand rows to the JDBC driver, without being able to tell in advance how many there will be. So the only way getRow() can know how many rows there are is by reading in all of the rows. This consumes time and memory. And you STILL have to write the while(res.next()) loop, so the computer is doing the same work twice and you are writing extra code.
"select count(*) from ..." is another way, but it has a few problems: (1) The query gets executed twice, so it is almost twice as slow. (2) The number of rows can change between the two queries. (3) You'll need to write the while(res.next()) loop anyway, so you'll be doing silly useless extra coding work.
Best Practice: stop needing the number of rows.
Second Best Practice: while(res.next()), create objects, put into a LinkedList, use list.size().
Silly Extra Work And Slowness: last()/getRow() or select count(*).
There may be situations where last()/lastRow() may be reasonable. You need to know how your database implements them and understand the performance implications. Don't even try if you can at all use the better ways.
There is a related question: how do I page database results. That may require "select count(*)" or something, depending on how you want to do it. There is no really perfect universal solution for that question using plain SQL.