The other major neat thing about functional languages is how they let you treat functions like objects, but most modern programming languages have adopted that as well, so it's not a big differentiator anymore.
In theory, writing in a functional language that allows only "pure" functions (aka. functions w.o. side effects), makes it easier to control state.
In practice, side effects exists and are required for programs to do anything useful.
In my opinion, one mistake of many purely functional languages was to be so focused on this purity, that it made it needlessly hard to write useful code in them, especially for people coming from an imperative/procedural/oop style of doing things. And you need these people if you want your method to gain traction, because the vast majority of code written, is imperative.
The irony is, that FP could probably have had a lot more success if it didn't clamour on about pure functions so much, and was less focused on implementations (aka. languages) than on methodology (aka. coding style) Because it is perfectly possible to write pure functions in most languages, including OOP language, even if those functions are not "pure" internally, or are not "pure" all the time and under all circumstances.
And yes, doing so has really nice advantages. I have refactored quite alot of codebases into using a more functional approach, and what I found was that this makes it harder to introduce bugs, makes it easier to track bugs, and makes it easier to reason about my code.
So yeah, functional programming, used if and where it makes sense, does work, and is useful.
Except functional languages like Lisps and the MLs never were pure, the only (used by a significant number of people) has been Miranda/Haskell (ignoring Coq and the other proof assistents). Or, to put it in other words: ML (no, not that ML) turned 50 this year, Scheme is oder than 50 too and Miranda/Haskell ~36. There never had been a shortage of "impure" functional languages since OOP existed.
So maybe it's time for FP as a whole to accept the fact, that there seems to be something fundamental about the way it's paraded implementations look and feel like, that puts off a lot of programmers.
Maybe it's time for FP to accept that the paradigm as a whole has a lot to contribute that is useful to everyone, but doesn't need a new language with largely different syntactic constructs to do so.
Note that there are many functional languages which don’t care about purity at all: for instance, Scheme, OCaml, SML and Elixir. I get annoyed when people confuse Haskell for the broader paradigm of functional programming. (Even though Haskell is my main programming language!)
But people tend to avoid the name "procedural" at all cost - which is bad because procedural programming really has it's advantages and should be clearly separated from FP which also has certain advantages.
If you declare a function effect-free, and try to perform effects inside it, then you'll get a compiler error.
You are still allowed (and it's common practice!) to declare your functions as effectful.
Unfortunately the 'signal' of this message is often lost in the noise: For every Haskeller happily writing effects, there's 9 non-Haskellers writing that 'Haskell's big mistake was being pure and not allowing effects - you should use X instead'.
It's definitely a different way of thinking and it has a lot of benefits. However, some of the downsides (which Flix seems to fix with region-based local mutation) might be implementing a sort that keeps two copies of a list in memory just to return the second list and discard the first. Depending on your list, this may not be an issue.
