Closed ohuu closed 2 weeks ago
Not sure this makes sense because the only thing that needs an ID now is the run_id in the ScenarioIteration and CpuMetrics table. This cannot be an auto-generated-by-the-db value, so we're down to using either nanoid or uuid v7 and uuid is not supported by sqlite (it's just a text entry) so there's really no advantage to using uuids
Nanoid produces human memorable strings but they are not easily indexed which could cause performance issues in large dbs. We could replace these with sequential integer ids or with uuid v7.
Leaning more towards sequential ids as I'm not sure if sqlite supports uuid properly so we hit the indexing problem again.