RequestFactory theory: Why is Locator<>.find() being called so often?

Question

I'm new to RequestFactory but with generous help of Thomas Broyer and after reviewing documents below it's getting much better :)

• Getting Started with RequestFactory
• Request Factory Moving Parts
• RequestFactory changes in GWT 2.4

But could you please explain why Locator<>.find() is being called so unnecessarily (in my opinion) often ?

In my sample project I have two entities Organization and Person that maintain parent-child relationship. When I fetch Organization Objectify automatically fetches child Person. Also I created two methods in my service layer findOrganizationById and saveOrganization that load and persist objects.

Now consider two scenarios:

When I call findOrganizationById in the client following calls occur on server side:

OrderDao.findOrganizationById(1)
PojoLocator.getId(Key<?>(Organization(1)))
PojoLocator.getId(Key<?>(Organization(1)/Person(2)))
PojoLocator.getId(Key<?>(Organization(1)))
PojoLocator.find(Key<?>(Organization(1)))
PojoLocator.getId(Key<?>(Organization(1)/Person(2)))
PojoLocator.find(Key<?>(Organization(1)/Person(2)))

By calling OrderDao.findOrganizationById I've already received full graph of objects. Why call .find twice in addition to that? It's extra load on Datastore that cost me money. Of course I cache it but it would be neat to fix it. How can I avoid these additional calls ?

Similar thing happens when I save object(s) by calling saveOrganization in the client. Following calls occur on server side:

PojoLocator.find(Key<?>(Organization(1)))
PojoLocator.find(Key<?>(Organization(1)/Person(2)))
OrderDao.saveOrganization(1)
PojoLocator.getId(Key<?>(Organization(1)))
PojoLocator.find(Key<?>(Organization(1)))
PojoLocator.getId(Key<?>(Organization(1)/Person(2)))
PojoLocator.find(Key<?>(Organization(1)/Person(2)))

I can understand need for fetching two objects from DataStore before updating it. RequestFactory sends deltas to the server so it needs to have entire object before persisting it. Still since I load full graph at once it would be nice not to have second call which is PojoLocator.find(Key<?>(Organization(1)/Person(2))). And I truly can't understand need for .find() calls after persisting.

Thoughts ?

My proxies

@ProxyFor(value = Organization.class, locator = PojoLocator.class)
public interface OrganizationProxy extends EntityProxy
{
    public String getName();
    public void setName(String name);
    public String getAddress();
    public void setAddress(String address);
    public PersonProxy getContactPerson();
    public void setContactPerson(PersonProxy contactPerson);
    public EntityProxyId<OrganizationProxy> stableId();
}

@ProxyFor(value = Person.class, locator = PojoLocator.class)
public interface PersonProxy extends EntityProxy
{
    public String getName();
    public void setName(String name);
    public String getPhoneNumber();
    public void setPhoneNumber(String phoneNumber);
    public String getEmail();
    public void setEmail(String email);
    public OrganizationProxy getOrganization();
    public void setOrganization(OrganizationProxy organization);
}

My service

public interface AdminRequestFactory extends RequestFactory
{
    @Service(value = OrderDao.class, locator = InjectingServiceLocator.class)
    public interface OrderRequestContext extends RequestContext
    {
        Request<Void> saveOrganization(OrganizationProxy organization);
        Request<OrganizationProxy> findOrganizationById(long id);
    }

    OrderRequestContext contextOrder();
}

and finally my Locator<>

public class PojoLocator extends Locator<DatastoreObject, String>
{
    @Inject Ofy ofy;

    @Override
    public DatastoreObject create(Class<? extends DatastoreObject> clazz)
    {
        try
        {
            return clazz.newInstance();
        } catch (InstantiationException e)
        {
            throw new RuntimeException(e);
        } catch (IllegalAccessException e)
        {
            throw new RuntimeException(e);
        }
    }

    @Override
    public DatastoreObject find(Class<? extends DatastoreObject> clazz, String id)
    {
        Key<DatastoreObject> key = Key.create(id);
        DatastoreObject load = ofy.load(key);
        return load;
    }

    @Override
    public Class<DatastoreObject> getDomainType()
    {
        return null;    // Never called
    }

    @Override
    public String getId(DatastoreObject domainObject)
    {
        Key<DatastoreObject> key = ofy.fact().getKey(domainObject);
        return key.getString();
    }

    @Override
    public Class<String> getIdType()
    {
        return String.class;
    }

    @Override
    public Object getVersion(DatastoreObject domainObject)
    {
        return domainObject.getVersion();
    }
}

Answer 1

The pairs of getId and find at the end are the default implementation of Locator#isLive: it assumes an object is live (i.e. still exists in the data store) if finding it by its ID returns a non-null value.

RF checks each EntityProxy it ever seen during the request/response for their liveness when constructing the response, to tell the client when an entity has been deleted (on the client side, it'd then fire an EntityProxyChange event with a DELETE write operation.

You can of course override isLive in your Locator with a more optimized implementation if you can provide one.