1. largely avoiding huge tightly-coupled frameworks like Django
2. writing a little more boilerplate with the payoff of making code a lot clearer to reason about without having to step through tons of layers of overly-dynamic libraries
I think Go will really start to eat up the backend web services space in the coming years.
scotty (-> net/http), aeson (-> encoding/json), and postgresql-simple (-> database/sql) are pretty close, although aeson & postgresql-simple need some tooling on top to get them on par w/Go struct tag reflection imo.
I still love to use more advanced Haskell libraries, but there's something to be said for stitching together functions in IO with simple types.
That's absolutely NOT true of Haskell. It's a much more nuanced language and while I'm sure it's great once you learn it, the learning curve is steep. Because of that I think it's much easier to hire Go developers than Haskell developers too
There are economics at play here too. If your turnover rate is high (1-2 years even), the company can get the short end of the stick, while the employees get to pay the fixed cost of learning Haskell on company time.
My experience though is FP novices can get to doing new-hire-level tasks in Haskell within a couple weeks of mentorship. I've seen 0-FP-experience interns dive into applicative functor code in _Scala_ and have no trouble so long as we sat down and explained things from first principles (and motivated the value of the abstractions we use.)
I have actually found the opposite to be true for hiring. Every place I've worked has had more Haskell applicants than they knew what to do with. Largely because Haskell can make your company uniquely desirable in a sea of options.
The only time a Haskell company I worked at "had trouble hiring" was when we turned away countless Haskell-experienced developers due to the sole veto of a VPE over and over again. After months of observation, it was clear that he was actually starving the team of Haskell resources so he could build momentum to move away from it (despite the team being fine with the language)
Not just Haskell, but any mature OO language/runtime as well (JVM, .NET). The fact remains that golang is popular because of Google's name behind it. While its concurrency model is somewhat decent, its warts get in the way very quickly for any non-trivial project.
I think Java/C# are verbose and ineffectual languages and people are seeking alternatives that make a lot more sense and deliver more power and speed.
* ResourceT is composable defer
* Streaming libraries (I use conduit due to familiarity) mean I don't have to write wonky loops to process data in constant memory
* ExceptT & Either is more composable error handling
* Parametric Polymorphism all around, but especially when it comes to concurrency (no need to hand-roll goroutine rube goldberg machines!) and collections (same but for hand-rolled loops)
* Async exceptions mean you can't have runaway threads without added implementation burden (I'm looking at you select + Context!)
* Sum types for a variety of use-cases :)
I've been using Rust as "fast Python". It's super expressive and you can get things done quick and dirty if you want.
IMO, the one advantage Go has over Rust is the fast compile time. Rust's type system makes it worth the wait (which isn't even really that bad).