this #elixir + liveview stuff has made me never want to write another REST API again
@ellie but isn't live view anti-let-it-crash?
Reliability is expensive, this is why at https://doma.dev we push many computations to the client, use append-only data structures and stay conflict-free.
This allows us to have a very reliable, very expensive, cloud active load balancer, which balances a cluster of federated backends. (We use `caddy` by the way, it's absolutely amazing).
I didn't buy into the whole #elixir web-interactivity stack, nor into #ecto (I actually have a very strong opinion that ecto is harmful for project budgets), that said, I can't imagine having painless multiagent server-side rendering in presence of server failts.
If we're talking principled architecture on budget, however, I feel like this is the best we can do with the current networking stack and everything beyond #plugs (which are alright, but a bit ad hoc and lack streaming support), especially live view, gets in the way.
@ellie that said, it's miles better than its direct competitors like PHP or Node.js-based SSR solutions.