I think the biggest weak points of firebase are a. a weak query engine (as soon as you reach non-trivial complexity you end up having to de-normalize data) and b. a weak rule engine (relying on a bunch of boolean logic is not going to scale as your app grows. you need more abstraction power)
Looks like your library addresses both -- crux has datalog queryability, and the rule dsl you wrote seems pretty powerful.
Good work Jacob!
1. Would love to learn a bit more about how the rule engine works.
From what I understand crux in essence only gives you key->blob semantics. How do you know that one `doc` is a `user`, and another doc is an `game`, etc
2. re: datalog subscriptions -- do you know if this is something inherently very difficult, or is it that crux hasn't implemented this yet?
2. It is inherently difficult. The closest thing like this that exists for datalog is I believe clj-3df[1]. Hence I was very excited when Materialize was launched publicly a few months ago. A version for datalog would be cool too, but SQL is good enough I think.
1) Crux is schemaless at its core, but you can build your own ~class hierarchy model using regular attributes and Datalog rules to differentiate between types of entities.
2) Subscribing to arbitrary Datalog without any form of polling requires a fundamentally different form of query algorithm (i.e. incremental view maintenance, as seen in Materialize), however there's a whole spectrum of possibilities available if you are willing to accept polling at some level. Hasura has quite an inspiring take on subscriptions that we may yet try to recreate on top of Crux: https://hasura.io/blog/1-million-active-graphql-subscription...