identifier uniqueness scope

[kcranston]

For the _id fields in the data model, what are the rules about uniqueness?

CanDIG is assuming that the ids for each schema (Treatment, Donor, Specimen, etc) are unique within that schema for a given Program. For example:

• each submitter_treatment_id must be unique for that Program
• you could have 001 as an ID for submitter_sample_id and for submitter_specimen_id (because specimen and sample are different schemas)

The ICGC-ARGO docs for identifiers states that identifiers "must be unique within all the data submitted by that program", which seems too strict.

[kcranston]

Closing this, because no comments / objections. We have implemented uniqueness as described above.