Hacker News Reader

Comment by christophilus | Hacker News Reader

I wanted to like Haskell, but just never could get to the point where I enjoyed using it. It always felt messy and complicated to me. I think the language extensions were a contributor to these feelings. I also felt as if I spent more time wrangling with the type system than actually solving my business problems.

Yet I really do like Clojure, F#, and PureScript. There's an experimental C++ back-end to PureScript now [0]. I wonder if that will ever be a viable production target?

Anyway, one of the things I like about PureScript is the row-types. Does anyone know if there's a plan to get row-types into Haskell?

[0] https://github.com/andyarvanitis/purescript-native

View on HN

> There's an experimental C++ back-end to PureScript now [0]. I wonder if that will ever be a viable production target?

Obligatory reminder to anyone enjoying PureScript so much they want to compile it to executable binaries for their backend work (instead of Node or such) --- I'm still hacking along on my PureScript-to-Golang trans/compiler (GH to follow in profile if interested). Unlike most alternative backends (to date) it's not a parallel fork of the purs compiler but works off the official purs compiler's `--dump`ed intermediate-representation files. Seemed more tractable to me to do it that way.

I also really dislike this extension system where you can unlock some magical features if you could just know what magical keyword to put at the top of your file.

The language was designed as a testbed for future PL research. That was the primary goal.

I am unconvinced of the practical utility of what row types give you for the added complexity.

The proposal to remove the Eff type (going back to IO) from purescript is telling

Do row types add a lot of complexity?

I think they add more complexity if you have to track and spell out each effect that is being added or removed from the stack

It's not surprising though, since usefully typing control effects with delimited continuations would require at least answer type polymorphism, but even more usefully something like session types.

There are a number of row type libraries and a proposed extension: https://ghc.haskell.org/trac/ghc/wiki/Plugins/TypeChecker/Ro...

Reply on news.ycombinator.com