Question

We have a somewhat complex hierarchy of classes in our entity model. We have a base class BaseEntity for all entities, then some “level 2” classes derived from BaseEntity, some “level 3” classes derived from any “level 2” class, and a lot of “level 4” classes, derived from any “level 3” class. That means, we have something like this:

• BaseEntity
  • DerivedEntity_1
    • DerivedEntity_1_1
      • DerivedEntity_1_1_1
      • DerivedEntity_1_1_2
      • DerivedEntity_1_1_3
    • DerivedEntity_1_2
      • DerivedEntity_1_2_1
      • DerivedEntity_1_2_2
  • DerivedEntity_2
    • DerivedEntity_2_1
      • DerivedEntity_2_1_1
    • DerivedEntity_2_2
      • DerivedEntity_2_2_1
      • DerivedEntity_2_2_2
      • DerivedEntity_2_2_3
      • DerivedEntity_2_2_4

And so on and so forth…

Now, when transferring instances of these types via WCF using protobuf-net, only the properties of the “level 4” classes are transferred – all properties from the base classes are lost!

From several other StackOverflow threads (here, here, and here), I read that this is the default behavior and that you have two options to work around it:

1. Use the [ProtoInclude] attribute on the base class.
2. Configure the inheritance at runtime using a TypeModel and AddSubType().

However, none of these options are feasible in our scenario since we have a lot of derived classes. Option 1 would mean hundreds of attributes on the level 1 base class. Option 2 would mean a lot of calls to AddSubType().

Are there any other options?

Answer 1

Ultimately, it needs to understand how it should store the data, so that you can get it back. It can’t just guess… So: the only robust way to do that is to define the model somewhere. That doesn’t have to be in raw code, though: for example, maybe it could be stored in an external file that you include and process (via AddSubType) at runtime.