todo:

• for the case where keyspace == null, try session.getKeyspace()
• enabling schema validation through an annotation is a bit too static. It would probably be more flexible with a runtime parameter:

new InventoryMapperBuilder(session).setSchemaValidation(true).build()

• trying to match a table and then falling back to a UDT is a good default. But maybe we could also have an optional annotation to provide hints to the mapper (keeping the name generic because other hints might be useful in the future):

@Entity
@SchemaHint(targetElement = TABLE)
public class Product { ... }

• we could also validate the partition key and clustering columns
• propagate isSchemaValidationEnabled in InventoryMapperBuilder to prevent backward compatibility problems
• similarly, I think we have enough information to validate the types: we know the Java type from EntityDefinition, and the CQL type at runtime from the schema metadata. So we could call codecRegistry.codecFor at runtime, and fail if it returns no codec.
• we recently introduced an EntityHelperBase parent class. There's probably opportunities to move some code there instead of generating it, for example methods such as:
  void validateColumnExists(TableMetadata table, String columnName, GenericType<?> javaType)
• 👍 for reporting all missing columns instead of failing on the first one. We could extend this to all the entity helpers of a given DAO, this way we validate the DAO's complete schema in one go.
• docs