That's a salient point. At the same time, the engineering and domain knowledge expertise required to pull this off is undeniably impressive. This takes the OR

That's a salient point. At the same time, the engineering and domain knowledge expertise required to pull this off is undeniably impressive. This takes the ORM "compatibility promise" into new territory, maybe even a new stratosphere. Imagine if someone (or company) starts a comparable open-source translation extension layer targeting Postgres as the backend. That would be interesting! Maybe start with the easier case of SQLite->Pg.