This is something that I've been pondering since reviewing the number of languages and frameworks I sampled in 2018. Time to pick winners.

Back-end:

  - backendless if possible (e.g. use GraphQL, Firebase, etc)
  - minimal (1:1 with document store):
    PostgreSQL/JSONB, FoundationDB, MongoDB, RethinkDB, CouchDB(mobile-sync)
  - relational: PostgreSQL(master/replica), MySQL(multi-master), CockroachDB/TiDB (sharded)
    languages: Elixir(Phoenix), Kotlin/Java(Javalin), Clojure(Liberator), Go, Kemal(Crystal)

Front-end:

  - HTML+JS using Phoenix/LiveView (or Vaadin?)
  - Vue.js (possibly Elm) for SPA

Runners-up:

  Spring - slow startup, JPA/Hibernate quirks, latency spikes (gc? of JPA/Hibernate)
  Micronaut - could be the next big thing (too much like Spring, learning curve)
  Ktor - coroutine support potential interesting, prefer a Kotlin+Java ecosystem
  SparkJava - too bare (used with Sql2o), Javalin is a spiritual successor
  DropWizard - older, poor documentation. But JDBI is sweet to use with other microframeworks
  Amber (Crystal) - Kemal has a much faster edit/compile/run cycle
  Rails/Sinatra/Web2py/Django/Flask/etc - prefer static typing and faster/smaller runtime

This is obviously still too long a list. Continue testing and culling.