What you do is for each object you use findentitygraph to return a list of related objects to the depth you require.

I'm surprised this is not already provided by DevForce.  I realize the complexity of deep-cloning the graph.  But as Ward suggested, this indeed can be handled via metadata (ala the buddy classes that can be created to implement verification), to indicate what is clonable and what is not.

Alternately, why not provide a deep-clone method, with a depth parameter?

 

In lieu of using something from IB, I would like to use hueha's suggested implementation, but the code posted here is missing definition of IMyCloningMethods, and I'm at a loss when it comes to inferring its implementation.

Any chance of posting it, hueha?

I'd be much obliged.

 

Thanks,

public static void DeepClone(Entity objectToClone, JTS2Entities mgr, Action<Entity> callBack)

        {

            Dictionary<EntityKey, Entity> clonedItems = new Dictionary<EntityKey, Entity>();

            DeepClone( objectToClone, mgr, clonedItems, callBack);

        }

        private static void DeepClone(Entity objectToClone, JTS2Entities mgr, Dictionary<EntityKey, Entity> clonedItems, Action<Entity> callBack)

        {

            ((IMyCloningMethods) objectToClone).GetEntityGraphAsync((list) => 

            {

              //set up clones

              foreach (Entity relatedItem in list)

              {

                  Entity relatedClone=(Entity) ((ICloneable) relatedItem).Clone();

                  MakeUnique((IMyCloningMethods) relatedClone);

                  mgr.AddEntity(relatedClone);

                  clonedItems.Add(relatedItem.EntityAspect.EntityKey, relatedClone);

              }

              //fix up IDs

              foreach (Entity relatedItem in list)

              {

                  FixForeignKeys(clonedItems[relatedItem.EntityAspect.EntityKey], clonedItems);

              }

              foreach (Entity relatedItem in list)

              {

                  RelinkRelations(relatedItem, clonedItems[relatedItem.EntityAspect.EntityKey], clonedItems);

              }

              callBack(clonedItems[objectToClone.EntityAspect.EntityKey]);

            });

        }

Yeah I eventually came to the same conclusion.  To get the list of objects to clone I now use have a method on all my entities that uses EntityAspect.EntityManager.FindEntityGraph to return the objects I want to clone.  Then I clone all the items and keep a mapping table to map clones to the original object.  I need to do this in order to remap all the related objects to the corresponding clones.  I used reflection to get to your property attributes.  I'll post my code in the next post.  Do you think I should rewrite using the GetEntityMetadata?

 

DevForce maintains a MetadataStore with a wealth of information about entities and their relationships. You can get the metadata for an entity type via

2) The problem is I'm trying to do it generically using reflection so there could be multiple steps needing to be resolved.  I suppose I could jump into the clone method recursively and asynchronously but the concept is starting to cause a stack overflow in my mind :D.  I'll give it a go

 

1) Unfortunately you will not be able to know if a field is a unique index (if not part of the PK)

 

2) since it's async, you must wait until it's resolved:

 

 

You can use

 

 

to find out the entity key.

 

You can also find if an entity is a pending entity with

 

 

 

1) I have a unique index on a field (companyName) and when I clone the entity and save it I get a violation of unique constraint from the database.  Is there a way of knowing what unique indices/constraints are in a model?  If so I could append a string to ensure it is unique before saving the clone.

1) Create clone

2) Reflect off properties and find any relatedentitylists

3) For each item in each RelatedEntityList

          recursively call deep clone with item

          set the property that is the same name as this class name to the clone

there are a couple of issues I can think of

1) I have a many to many relationship which could cause the recursion to loop infinitely.  Maybe I could keep a track of original entity objects I have already cloned.

2) I'm assuming that any relationships are cleared from the clone.  

For example lets say I have 

a company table with a companyid

        a location table with locationid, foreign key companyid

        a contact table with contactid, foreign keys companyid, locationid

If I clone a contact and I set the locationid to the cloned location then the companyid should also point to the cloned company not the original one

i.e. cloning an employee and his orders:

 

 

 

That's good news.  So any ideas on doing a deep clone?