[Taikonerd]

The DB/ORM mismatch is worse than it has to be, because SQL queries always return flat rows. But in code, we don't usually want rows; we want objects linked to other objects: "get me the Account objects and their associated Wishlists."

If you use a query language that knows that we want objects linked to other objects, you can still have an ORM, but it doesn't have to do as much heavy lifting. The query can already specify the objects and properties you want explicitly, so you don't need to worry about, "which properties of which objects do I flesh? Am I over-fetching?"

That's why I've become a booster of EdgeDB: https://www.edgedb.com/ It's a sort of "midpoint" between regular SQL and an ORM. And it's language-agnostic, unlike an ORM.

[1st1]

Yeah, that's the exact goal of EdgeDB -- unlock proper hierarchical data retrieval and mutation as well as making composition possible both at the schema and query layers. Most other perks EdgeDB has are the direct result of that.