The link validation is a bit bizarre and doesn't seem correct, or it's at least very provisional and not rigorous enough. For example, you could have two models with "collections" of each other - each expects the other to hold a foreign key, but neither does because there is no join model - but they should check out as reciprocally linked. Also, a self-referential link (which is in use, e.g., in the ipi::patient model for setting a reference_patient) checks out - but only because it proves to be its own reciprocal! Bizarre and strange. These behaviors should be clarified before we start trying to figure out how to make mutations, since in the future we'll have to validate graph integrity at every step.
The link validation is a bit bizarre and doesn't seem correct, or it's at least very provisional and not rigorous enough. For example, you could have two models with "collections" of each other - each expects the other to hold a foreign key, but neither does because there is no join model - but they should check out as reciprocally linked. Also, a self-referential link (which is in use, e.g., in the
ipi::patientmodel for setting areference_patient) checks out - but only because it proves to be its own reciprocal! Bizarre and strange. These behaviors should be clarified before we start trying to figure out how to make mutations, since in the future we'll have to validate graph integrity at every step.