We have been running into some issues where by links can get bad data into them various ways in the ontology. We have observed for example actions may accept invalid FK values for links. For example consider the following:
Create two objects:
- Patient - > PK PatientID
- Patient Record → PK PatientRecordID
- Link → Patient.PatientID → PatientRecord.PatientId
Then create an OSDK app and try putting a bad PatientRecord.PatientId value (ie the value does not exist in Patient) into calls to Create Patient Record Action, Upsert Patient Record Action, and Modify Patient Record Action.
This is one scenario we tested and it allows and indexing succeeds. We have read that Foundry allows this but we are wondering if there are plans to support the same kind of referential integrity checks transactional databases support. We understand the Ontology should not be normalized like relational databases. We are doing top down domain modeling. But without these contraints we are relying 100% on the application layer to protect the data model.