As I mentioned in other entries, I like to use UUID values. The current version of RIFE doesn't handle UUID properties correctly. However, a few minutes digging into the source code showed me that the changes needed for support where trivial.

I created a com.uwyn.rife.database.types.databasedrivers package in my own project and then copied the org_hsqldb_jdbcDriver.java into it.Then I updated the getSqlType() method by added the highlighted code shown below immediately after the test for the character class.

  else if (type == java.util.UUID.class) {
    return "VARCHAR(36)";
  }

I'd love for this change to be propagated to the rest of the driver files and slipped into an upcoming release.

The RIFE framework has functionality which abstracts the Sql Create Table statement into Java classes. The fundamental reason to use an abstraction instead of writing the Sql directly is that the underlying framework generates Sql specific to the targeted database. Why is this important? Frequently I use an open-source database like Hypersonic for development but a commercial product like Oracle for production.

The following code demonstrates the technique:

  String driverClassname = "org.hsqldb.jdbcDriver";
  String url = "jdbc:hsqldb:hsql://localhost:9101/test";
  String username = "sa";
  String password = "";
  String poolSize = 5;

  Datasource ds = new Datasource(driverClassname, url, username, password, poolSize);

  CreateTable create = new CreateTable(ds);

  create.table("beer")
    .columns(Beer.class)
    .primaryKey("id")
    .precision("brand", 50)
    .nullable("brand", CreateTable.NOTNULL);

  String createSql = create.getSql();

When executed, the generated SQL looks like this:

CREATE TABLE beer (brand VARCHAR(50) NOT NULL, id INTEGER NOT NULL, price NUMERIC, PRIMARY KEY (id))

For completeness, here are the relevant parts of Beer.java

public class Beer {

    private String brand = null;

    private BigDecimal price = null;

    private int id = 0;

    // ... snipped out getters and setters.
}