Re:Haskell for a New Decade [pdf]
by kccqzy on Wednesday June 24, 2020 @12:57AM
(#23624262)
> Instead of Haskell we should be thinking of programming language research and how to get the best ideas into our mainstream languages.
I feel like that's what the Scala people are doing. Yet the end result is that this kind of Scala code is just as slow to compile as Haskell, and just as hard to learn for newcomers as Haskell.
The difficulty lies in the concepts enabled by the language, not the language itself. Go read the GHC user guide, especially the part about language extensions, and you'll finish in an afternoon. Yet people are always coming up with new ideas and abstractions that use these language features in new ways. Haskell is an especially rich platform for new ideas and abstractions to foster. Lens is somewhat difficult to learn, yet in its simplest form it can even be used without any language extensions. (You won't be able to name the Lens or Traversal type itself without RankNTypes, but you just need to expand the synonyms; in other words it's just sugar.)
I feel like that's what the Scala people are doing. Yet the end result is that this kind of Scala code is just as slow to compile as Haskell, and just as hard to learn for newcomers as Haskell.
The difficulty lies in the concepts enabled by the language, not the language itself. Go read the GHC user guide, especially the part about language extensions, and you'll finish in an afternoon. Yet people are always coming up with new ideas and abstractions that use these language features in new ways. Haskell is an especially rich platform for new ideas and abstractions to foster. Lens is somewhat difficult to learn, yet in its simplest form it can even be used without any language extensions. (You won't be able to name the Lens or Traversal type itself without RankNTypes, but you just need to expand the synonyms; in other words it's just sugar.)