CoderSupremeWhat programming languages are considered the best in terms of productivity, design, and conciseness, rather than just popularity or longevity?
u_tamtamIMO Scala is one of the best programming languages out there. I know it might sound like zealotry because Scala is already way past its hype curve, and the “Through of disillusionment” already caught a fair bunch in ways that more recent and hyped languages haven’t yet, but it’s not only still very relevant today, but more and more so (IMO).
So, what’s to like about Scala? Like most of things, those are two-edged swords:
1- multi paradigm
To my knowledge Scala is the only language that unifies object oriented programming and functional programming so seamlessly. You can pick the right tool for the job, opting for imperative-style where it’s fit and choosing elegant composable/curried when appropriate, without having to bend your mind as much as you would with Haskell/clojure/OCaml/F#/… where things are more one-sided. The downside is that different programmers will have different takes and preferences as to what’s the most adequate style might be, and a same codebase might look very different from one place to the other.
2- type system
Scala has one of the most advanced type system. Nothing Rust or Kotlin might match any time soon, or ever. Scala’s implementation of GADTs, combined with its powerful pattern matching enables concise and idiomatic abstractions. Many of which are zero-cost thanks to things like opaque types, inlining, tail recursion, … There is a whole area of the Scala community striving to make invalid states irrepresentable (your code won’t compile if your instance of a pizza is missing a topping), which makes such libraries self-documenting and easy to use. The downside is that nothing prevents you from climbing the abstraction ladder and encoding everything in the type system when all you need is a simple trait/generic, and that’s a human/complexity management problem tooling and the language can hardly mitigate.
3- scalable
The author of Scala (who was a long-time Java compiler architect) wanted Scala to scale from shells one liners to complex multi-cluster distributed systems, and delivered on that. You can start small with a scala-cli proof of concept, transition to a mid-scale “python with types” kind of project, and grow up to very large and complex projects. Beyond the JVM, you can target the browser with scala-js and share models and validation logic between the front and back ends. You can target native binaries for instant startup/low footprint executables that are cheap to spin-up as microservices.
4- has a foothold in academics
A whole team at the EPFL is pushing boundaries of programming languages and using Scala and its compiler as a ground for experimentations. Scala 3 has a proven sound type system thanks to its foundations on the DOT calculus. Effects and Capabilities are being researched as part of the project Caprese to offer a solution to “what color is your function” (mixing sync and async), of memory management/lifecycles (more generic than rustc’s), of pure/side-effectful code, etc. The downside is that this gives an impression that Scala’s development lacks focus, but arguably those happen in distinct development branches and by different people.
Anyway, feel free to continue the discussion on: programming.dev/c/scala
I’ve only written a “hello, world” in Scala a long time ago, but it seems to be used a lot in the banking or fintech industry mostly. Is there a reason for this?
v_krishnaIt got this foothold pre-spark, largely due to the akka and typesafe/lightbend ecosystem. Then spark resulted in a lot of data engineers picking up scala (this was my entrance, from the Hadoop map/reduce world). And now cats/zio and effect systems have rounded it out.
Personally I love scala, my teams use it heavily (mix of styles but mostly zio-http, and a lot of spark). But scala3/dotty has been problematic enough for us that we decreed people stop updating their apps. The learning curve is HIGH (esp for functional/effect system scala). The candidate pool is small. I don’t know that if I were to start a greenfield project (without all the rest of the platform already using it) I would suggest we use scala. But the rest of our platform does, and we have tooling, and L&D tracks, etc. So onwards we go.
The learning curve is an interesting one. We’ve had new devs contribute within a sprint from zero knowledge. When they are contributing to an established platform.
Once they start looking beyond the walled garden things get complex. Scala supports a lot of variety in approaches. Which do they choose? Which is ”better"? Those questions can be hard to answer.
For me the variety of choice is great. For a new dev… Not so much.
C++ has the same issue, and that doesn’t stop everyone from using it.
Seconding what the below poster says. Spark and akka. Akka streams, specifically, was the only real distributed streaming system for a very long time. Which is very, very nice for low latency. Even now it’s tough to beat.