Hacker News new | comments | show | ask | jobs | submit login
Forest, a multi-syntax functional language that compiles to WebAssembly (github.com)
132 points by eterps 10 days ago | hide | past | web | favorite | 24 comments





Well, this seems like an interesting project. Although very ambitious, the author seems to know what he's talking about, and has been going at it for a while, so kudos for that.

I find the design principles interesting. We might talk a lot about syntax abstraction, but I feel like that's not even the most interesting idea (it's very likely that many people who thought about making languages thought about it).

There are a few trends I find interesting, though. Programming languages are pretty tricky. You need to think about style and readability, performance and optimization, scalability and maintainability, data types and abstractions, algorithms and data transformations, parallelism and concurrency, etc. It seems to me like we have been introducing systems that progressively try to modularize and isolate those issues, sometimes completely hidding them from us. Syntax abstraction shouldn't be so surprising if we follow this idea. We got garbage collection, most optimization is done by compilers, we got dynamic typing, endless discussion about models for data representation, even visual languages, etc. The idea that programming languages shouldn't be limited to text in a single source file is very important in my opinion. I often envision future programming as "powerful IDEs" (and I never use IDEs nowadays) that can represent different properties/aspects of programs/code in different ways/views, isolating them or showing relationships/interaction between them. The complete opposite of what we see in movies, as always happens in reality, hahaha. Well, went a bit off-topic so I'll stop here.


Someone already mentioned Luna Lang. Another similar project that is further along is isomorf. You get projections into 6 or so syntaxes, and deployment to AWS Lambda.

https://isomorf.io/#!/syntax-picker/tours-ideas


Just a quick update on isomorf.io, we are bootstrapping the projections on the platform itself and so soon you will be able to design and implement your own. Stay tuned.

The multi-syntax support is really interesting. https://www.luna-lang.org/ is another language with a similar concept, with both a visual and a textual representation of the same code.

Would love to read more about the immutable data structures it supports. Rich Hickey has some great talks on efficient immutability in Clojure.

I really believe we're going to see a lot more new languages targeting web assembly. I wish it supported a standard, pluggable garbage collection system that could be easily adopted.


also GraalVM

I don't know if I buy the justification for supporting multiple syntaxes. It's a slippery slope towards appeasing a million people's particular syntactic pet peeves, and it's not worth the effort of supporting them all.

With the exception of s-expressions (too many parentheses) and mandatory/significant whitespace, most people don't seem to care too much about syntax, IME. Semantics are much more important.


> most people don't seem to care too much about syntax

Speaking as an Erlang enthusiast, I must disagree. I love the syntax, but most people see "different" and think "bad".


Yeah, I generally agree. But I think our understanding of like, language popularity is a little colored by "the era of language fads". C++ is probably first, then Java, PHP, Ruby, and now JavaScript. If a language doesn't build a huge mainstream base (Ruby maybe failed at this but maybe only for lack of a big corporate backer) or 100% occupy the dev zeitgeist for an extended period of time, we think it hasn't caught on. But like, Erlang powers a lot of web infrastructure. Whatsapp is most famous, but there are a lot of services I'm personally aware of that are written in Erlang, and often these were situations where a mature dev team wanted to see if Erlang solved the problems they'd experienced writing backend services in Java or Python.

So yeah while syntax probably prevents you from achieving rockstar status or world domination, I think those days might be over, and I think the true measure of success is whether or not experienced dev teams are using the tech to solve real problems.


I've been trying to get some feedback on a related idea for making 'view-independent program models', which could be rendered with alternate syntaxes (or even richer visual augmentation than strictly text).

The main idea is to represent languages as a graph of 'language constructs' and individual programs as paths through some language graph. That path is the 'model' which an editor or compiler/interpreter would operate on—but in the case of the editor, since the program model has no connection to syntax, you can configure how it's rendered on a per-user basis (or use it to experiment with new language UI concepts, etc.).

Here's a write-up with illustrations: http://westoncb.blogspot.com/2015/06/how-to-make-view-indepe...

I've had this on the back burner for a while now, waiting for the write moment to try building out a proof-of-concept version—but if someone could point out some kind of critical flaw in it and prevent me from wasting so much time, that'd be awesome :)


Hi, I am working on something similar, it's nowhere near usable but I hope that a proof of concept version could be ready this year. My variant is focused on visual programming with various representations. I will definitely read your docs. I am very interested in the models you have come up with.

That's great, definitely send me a message if you want to talk about it more at some point. You can find my website with contact info on my profile here.

"Syntax is polarizing, so we'll move the decision to the team level where they can argue endlessly about which to prefer."

Interesting project, I like the emphasis on immutable data and pattern matching.


It does raise a question of cross-team communication. If one team proposes a change in their dialect, can another team review it in a different dialect?

At that point, developing semantic diff capabilities seems to become vastly more important. At the same time, syntax abstraction might already give you the tools to start from as well.


> How do I edit the code in my preferred syntax? Do I need an editor plugin?

> When working on the project, each developer runs forest dev, which mounts a virtual filesystem in the local directory using FUSE, called dev/.

> dev/ contains all of the source files, projected into the developer's syntax of choice.


Right, the projection part is clear. The question was one of diffs, reviews, etc. What does a diff look like when projected between languages? Can you still feasibly use a line-based diff format, or do you have to account for transforming syntax structure? Do you just try to diff the projections instead? Lots of interesting things to try out.

1. Holy shit they use FUSE, that's pretty bananas.

2. We're definitely getting closer to the era of language swappability, where language X transpiles to language Y, language Y compiles to language X, and strict code formatters clamp down on corner cases. I think it's great and welcome, but I would expect a side effect of this to be that more "free" languages pop up that allow programmers to be a little looser in their implementations.


Why does anyone think multi-syntax is a good idea?

They think it will help adoption? But surely it also hinders it by fragmenting code examples in the wild (Stack Overflow, docs, blogs etc) into different syntax populations?


Lots of nice ideas which I don't think will work together. Nevertheless, it's always a good idea to give it a go, and the next time around maybe it will work out. For example, I don't think there is any sense whatsoever to have different syntaxes for exactly the same structures and semantics. You want different syntaxes to make different semantics more accessible. Also, reference counting and functional programming doesn't seem like a good fit, as much as I get the performance aspect.

Doesn't Reason leverage syntax abstraction in a similar way?

https://reasonml.github.io/


Yep, though we don't put much emphasis on it for now: https://twitter.com/jaredforsyth/status/1026706210043490304

This is a lisp syntax for Reason, that works already with our language-server integration, etc.


I like the idea of abstracting the syntax so that if you want to program in something Haskell like you can or program in a JavaScript like language. (I had a similar idea). I hope that the developer ergonomics that this provides is enough to balance the trouble of multiple people reading different syntax .

> For the sake of collaboration, we agree on structure and semantics, and agree to disagree on syntax.

`FooError on line 111` suddenly becomes a lot less meaningful... I suppose that's what the canonical representation is for, but having to map between the two seems like it would be at least somewhat annoying.


"Projectional Editing" is coming to my mind, in case you are looking for more input on the topic.

The Loyc project has similar goals: http://loyc.net/



Applications are open for YC Winter 2019

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: