Видео

Elixir, clojure??

Debug.Trace is exactly what I needed, thanks!

How is it biased, when they are treated equally? :P

Ah, so rows are just like traits from Scala where you can mix them in.

Don't be a Heyter!

Now this is where it's at!

sloooooooow down 😅

You are apologizing and giving ground tucking yourself into you personal subjective opinion corner. I prefer to hear bolder and general statements from you.

I appreciate that for OCaml you mention errors as a usecase for polymorphic variants in OCaml. That is my exact primary use case for them also; as the entire range of possible errors doesn't need to be defined somewhere. Add conflict detection in the database layer in a REST service, and the HTTP layer is forced to deal with this, returning a 409 status code (if the code in between didnt through a retry). Oddly enough, the only other language I've used where the same type of benefit can be achieved is TypeScript using string constants and union types. The only problem is; people who know TypeScript can't read your code if they don't grok FP error handling. But OCaml have GADT that can hide a type parameter when inconvenient, and extensible variants that allow for variants; where the total set of possible values may not be known at compile time. Both of these had their uses in my OCaml unit test framework, speed. GADT to have an infinite hierarchy of scopes with their own type parameters, and Extensible variants to allow the user to attach custom data to test cases; and retrieve it in a type-safe way. The code to retrieve it is a little verbose, so PPX rewriters were written to convert the switch expression into a one-liner.

More videos on this topic would be welcome. Preferably more structured

FP, as every other paradigm, has its strengths and weaknesses. Using the right tool for the job is more important. Yes, if you need to multithread, going for a functional approach is better, but single threaded, complex operations that are to be written maintainably likely are better written in OOP. Don't pick up a hammer and make every problem a nail... Get a toolbox and use the right tool for each job.

Putting stuff in cabal.project is not very useful because Hackage ignores it. You must specify an upper bound for every package, and you must do so in the .cabal file.

As far as I can see nobody misses Erlang in the commets so here I am 😢

please don't stop make content like this

Can we please stop picking on the functional programming bros. They are very busy telling other people they are getting it all wrong, that there is only one size that fits all (hint: it's fp), and also trying to get their very first program to work (without all the other fundamental concepts that you actually need). Moral of the story, abstraction and theory won't get you a job, or a working program.

This video is such a huge work, big thanks to you. I'm surprised that here was no comments and mine is gonna be the first. I'm relatively new to Haskell coming from C# where everything is kind of monopolized and ready to use, so, I did not put much effort into picking tools, because I didn't have to. Haskell is different, thank you for helping people like me to dive into this beautiful language. This is the first video on this channel, that I've seen, so, I've checked out the other videos and I'm actually interested in your work. New subscriber right here. Thanks again!

The sealed trait sum type pattern is not the same thing as the enum pattern. They're two different sides of the expression problem and both are necessary depending on if you're looking for behavior extensibility or case extensibility

讨论形式语言就不能正经学点数理逻辑么，函数式编程的表达能力来自其与证明论的紧密关系。跟古代工匠一样发言，能说明白才见鬼了。

functional programmers aren't inherently superior compared to other programmers, but functional code is inherently superior to imperative code

The thing is FP makes people who do it think they are smart and for a good reason. But some people start to attribute that feeling of "smartness" and explain it by "superiority" which is just a bias in the end.

There is one thing where a subset of FP (total functions) is superior to everything else - writing *provable* code.

Well articulated. Good style and structure.

As Tony Hoare said in 1968, we need to know *what effects* a procedure / function has. Having *no effects* accomplishes this goal, but it's more extreme than necessary.

Blasfemy

I am a functional programmer myself, however, Functional has its weaknesses. The main weakness of Functional is that it is not so good at lower level operations like memory manipulation. The other thing is that because of this it is less performant. So currently we still need to have lower level procedural (C or Go) or functional inspired (Rust) languages. I think the future of software engineering is some functional language that respects the underlying computer architecture. So maybe a functional language that can compile to optimized procedural code... Similar to Roc or something of that nature.

I liked codecrafters, but it is quite expensive

I love this video. It is unafraid to introduce lots of great functional languages, even some you may not have known!

Very good chapter. I'm going through your course after reading Learn you a haskell. Great to see Applicative Functors and Monads finally in "real" practice !

Bro you don't get the concept of a tier list facepalm

Seems kind of annoying to have to write your own bindings for the source code. Is there no way to automate that process? I'd like to try PureScript, maybe with React, for UI apps but this feels like doing double the work. I get that writing business logic in a more tightly constrained language like PureScript is beneficial over TypeScript, but this seems like you have to maintain two implementations almost.

Dude chill :)

Thanks for posting some Purescript. I’m really trying to get Purescript to catch on.

This demo would not be here without github.com/rowtype-yoga/purescript-react-basic-dom-beta and the author of the library

4:08 do-notation is what?

On my last attempt to install Haskell, I was like "ghcup? What is this? People think this is worse than python???"

This is what the politicians call a "hearts and minds" issue. Right now, people want to program in whatever could net them a high paying job. Haskell could be the most beautiful and useful language in the world, and they would parrot these talking points just because there has to be SOME reason why it's not a top 10 language. But let someone write a library in Haskell that creates a gold rush (imagine: Haskell on Rails...lol), then they'd be interested, then they'd start making up new stories about why Haskell is the best language. TL;DR: the mob is fickle.

This talk is really confusing, since it does not give an impression that the speaker knows what monads are. Weird.

I liked the animations. Unless Scala has monad transformer stacks, it wouldn't be helping me learn what's actually hard about Haskell.

Funny. I went the other way around. Java -> Haskell -> Scala :) Both Haskell and Scala are good languages to know <3

3:10 do you plan on covering Roc anytime soon? i really like it's language design choices

3:08 i did learn Haskell via OCaml as I had some experience with Rust prior. And Rust and OCaml share a lot of common aspects while OCaml still introduces most of the functional programming topics. I would recommend going to Haskell through OCaml for Rust and Swift devs as they both were inspired (even indirectly) by OCaml

Scala is a good way to prepare you for Haskell compile times.

Java can also model ADTs through sealed inerfaces and records.

Oh lookie, a video about an opinion I already hold.

I wouldn't watch it in fullscreen

This one video made me subscribe!!! Love how u pointed out that Haskell was lazy and hence needs monad for io

ELIXIR wasn't there because it's in an S++ tier.

Honestly i don't get functional programming hype, i'm no hater, i try to learn srust which implements some of the functional patterns, so i'd like someone to sell me on it

isn't typelevel more for Tagless stuff than transformers?

Good call on not mentioning, that many code bases are still on Scala 2 and use things like play! or Akka. :) I actually don't know how to feel about things like cats-effect. Nor "direct-style". We'll see how that develops, maybe it's going to be a simpler solution.