Yes and Yes!  There is absolutely a lot of potential there -- it's also one
of the reasons I'd really like to see RowHandler become a first class part
of the dbutils library -- ProxyGeneratingRowHandlers could do all sorts of
interesting things.



On Tue, Feb 10, 2009 at 5:28 PM, Dan Fabulich <d...@fabulich.com> wrote:

>
> Have you seen ActiveObjects?  It's an incubating ORM layer over at
> dev.java.net; it's got some neat stuff in it.
>
> https://activeobjects.dev.java.net/basic-concepts.html
>
> I'm especially drawn to the idea that a bean could be defined as an
> interface, rather than as a concrete implementation.  It saves me from
> having to actually define the implementation of getters and setters, while
> still preserving type-safety (and auto-completion in my IDE).
>
>  interface Employee {
>    public String getName();
>    public void setName(String name);
>  }
>
> It's easy to use java.lang.reflect.Proxy to automatically generate simple
> implementations for bean-like interfaces.  The InvocationHandler could just
> keep a DynaBean internally that maps properties to values; when a getter is
> called, we'd just call an appropriate getter on the DynaBean; when a setter
> is called, we'd just call an appropriate setter on the DynaBean.
>
>  Employee e = instantiateBeanInterface(Employee.class);
>
> The result would be something like a type-safe DynaBean, which intrigues
> me.
>
> We could go even further.  Perhaps if the interface bean (IBean?)
> implemented other tagging interfaces, we could add in additional
> functionality.
>
> For example, BeanUtils could define an interface like this:
>
>  interface DirtyManagedBean {
>    public PropertyDescriptor[] dirtyProperties();
>    public boolean isDirty(String propertyName);
>    public void clean(String propertyName);
>    public void cleanAll();
>  }
>
> Then, I could make my Employee interface implement DirtyManaged, and have
> the automatically generated bean keep track of which properties were dirty
> (had been modified) and which were clean.
>
> We could also have an interface like this:
>
>  interface SupportsPropertyChange {
>    public java.beans.PropertyChangeSupport propertyChangeSupport();
>  }
>
> If my Employee implemented SupportsPropertyChange, I could retrieve its
> associated PropertyChangeSupport object and attach change listeners; we'd
> guarantee that the generated bean would fire change events via the support
> object.
>
> FOR DATABASES
>
> Specifically, I'm imagining IBeans would be helpful in the world of SQL
> databases, outside of any ORM layer.
>
> 1) You could automatically INSERT/UPDATE a DirtyManaged bean, adding only
> explicitly dirty columns to the SQL statement.
>
> 2) In general, an IBean could wrap around a PreparedStatement:
>
>  PreparedStatement ps = prepareStatement();
>  Employee e = proxyPreparedStatement(Employee.class, ps, columnList);
>  employee.setFirstName("Dan");
>  employee.setLastName("Fabulich");
>  ps.addBatch();
>  employee.setFirstName("Somebody");
>  employee.setLastName("Else");
>  ps.addBatch();
>  ps.executeBatch();
>
> 3) An IBean could wrap around a ResultSet; when you call .next() on the
> result set, the IBeans getters/setters could return the new values,
> eliminating the need to create N objects for N rows.
>
>  Employee employee = proxyResultSet(Employee.class, resultSet);
>  while (resultSet.next()) {
>    System.out.println("Name: " + employee.getName());
>    System.out.println("Address: " + employee.getAddress());
>  }
>
> 3) If you're doing a join, you could define a new bean that wrapped around
> the joined query via multiple inheritance.
>
>  interface EmployeeAddressQuery implements Employee, Address;
>
>  Statement s = c.createStatement("SELECT * from employees, addresses");
>  resultSet = s.executeQuery();
>  Employee employee = proxyResultSet(Employee.class, resultSet);
>  while (resultSet.next()) {
>    System.out.println("Name: " + employee.getName());
>    System.out.println("Street Name: " + employee.getStreet());
>  }
>
> Thoughts?
>
> -Dan
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org
> For additional commands, e-mail: dev-h...@commons.apache.org
>
>

Reply via email to