Hacker News Reader

Comment by thangngoc89 | Hacker News Reader

OCaml is a very powerful language. I'm still discovering new features/patterns everyday. But coming from a background with C/Javascript, the syntax is definitely hard to understand. For this particular problem, I think ReasonML has done a very good job of introducing OCaml platform to a wider audience.

Shamelessly plug: I'm working sketch.sh [1] as a learning project for ReasonML. It's a interactive playground (like IPython/Jupyter) for OCaml and ReasonML. It's provide inline types and evaluated values. Give it a try if you're learning OCaml/ReasonML.

[1]: https://sketch.sh

View on HN

> But coming from a background with C/Javascript, the syntax is definitely hard to understand.

I work with Haskell fairly regularly, and have experience with a few other languages (Python, Racket, C/C++, Java, etc.), but I definitely agree with this sentiment. I am not a fan of OCaml's syntax at all.

I haven't had a chance to check out Reason yet though, so maybe I'll spend some time playing with that soon!

I don't think its syntax is that bad. A bit daunting at first, but it isn't like you need to learn a ton to understand it, so once you have the basics it is pretty readable.

I don't get it. I work day to day with Algol-syntax languages like JavaScript and Java, but have dabbled in ML languages like Elm and Haskell. One of the things I really like about those languages is the syntax -- succinct and clear. What takes time and is hard for me is grasping the semantics and patterns.

I find it pretty clear in simple cases but ambiguities and associativity rules can make interpreting some code a real pain.

OCaml's syntax is way more verbose and noisy than Haskell's or Elm's. It tends to have longer keywords, more sigils, … and some of the defaults are inconvenient (non-recursive lets)

https://wiki.haskell.org/OCaml has some trivial examples

As a contrasting opinion I started reading the Ocaml book about something real world and I felt the syntax finally something fresh, beatiful and clean. But I also suffer from having to use scala everyday at work.

As a big Ocaml fan, the syntax is one of my least favourite aspects of the language.

Two things that I find particularly annoying are constructors (which don't have the same syntax as funciton calls) and the multitude of shift-reduce ambiguities that don't do the thing that you want. For example, in the following code the W gets parsed as part of the inner match:

    match foo with
      | X -> match bar with
               | Y -> 1
               | Z -> 2
      | W -> 3

A lot of this could be remedied by having more expression terminators. ReasonML uses C-like curly braces but even a simple "end" token to close the match expressoin (similar to what Lua does) would already be enough.

This tripped me up when I was first exploring OCaml. FWIW i've taken to wrapping the nested matches with parenthesis or a begin/end

    match foo with
      | X -> (match bar with
               | Y -> 1
               | Z -> 2)
      | W -> 3

    match foo with
      | X -> begin match bar with
               | Y -> 1
               | Z -> 2
             end
      | W -> 3

The indentation sensitive syntax of F# was the solution to me, I would love to see it being integrated into OCaml.

Ohh interesting. Didn't get that far yet. But good to know. Is there somekind of place listing these gotchas with the syntax? Seems something good to be aware.

Check out the point "comparison to ocaml" in the reasonml docs

Reply on news.ycombinator.com