Researching. https://clojurians.slack.com/archives/C019ZQSPYG6/p1693449058790539. Got more than I could possibly deserve.

Some food for thought: • https://pathom3.wsscode.com/ • https://www.youtube.com/watch?v=IS3i3DTUnAI I think EQL and Pathom is a "killer app" in Clojure for building bottom-up APIs. One of the things that helped me a lot with domain modeling is to think about how one can "flatten" relationships/entities as much as possible. That way, the user of the API is free to build up custom trees during query time.

Note: I'm not suggesting you must even expose the EQL / EDN as the primary API to the end-user/browser. I'm in favor of even modeling the domain via pathom/eql and then wrapping it in a "RESTful" http API if that's what the outside world wants. :)

While Life will have ancillary, sub-domain facilities down the road, it’s core interface is a chatbot fronting an evolving knowledge graph. After I gnaw on https://clojurians.slack.com/archives/C05P6UP4G2K/p1693484420616949?thread_ts=1693440821.119119&cid=C05P6UP4G2K, I’m going to ask #datomic and #ions-aws if anyone has seen such a thing.

Following Sean’s link to https://jacobobryant.com/p/post-2019-ion/ (from https://jacobobryant.com/p/post-2019-ion/), I stumbled into tonsky: https://tonsky.me/blog/the-web-after-tomorrow/ To my current understanding, this is the stuff of Stage II.

There are some even more far-out ideas: • if you search for "Reactive Datalog" or "Differential Datalog", there are lots of interesting ideas • https://github.com/hyperfiddle/electric (and related https://github.com/leonoel/missionary) • the Rama architecture implemented/proposed by https://blog.redplanetlabs.com/

But I would not suggest that any of this is necessary at the moment, you will go down too many rabbit holes :)

Which reminds me of another book - if you haven't heard of it - https://dataintensive.net/

Thank you for trusting me with all this.