CREATE TABLE foo (id INTEGER,value INTEGER);

Doing a SELECT * FROM foo WHERE id=1 migth yield something like this:

id    value
1     56
1     95
1     34
1     12
1     116

After having messed about with these values in my script, I end up with this:

id    value
1     56
1     34
1     22

Now, what is best practice: deleting all the current rows and inserting the new id/value pairs anew, or looping over the current data again, deleting where id/value no longer exists and then inserting the new id/value pairs?