44 comments

[ 3.1 ms ] story [ 98.3 ms ] thread
Initiatives like this make me really excited, but the biggest problem is always long term maintainability and interoperability with the rest of the browser ecosystem. Without a huge company backing the effort, it becomes really risky to use this long term because development could drop off or never gain traction.

If you don't have a huge open source community, you need one to push the ball forward and keep things going. That's really hard to get without users, which is a great catch 22. I'd almost suggest building a "killer web app" on Fengari just to demonstrate it, or at least to indicate that its development effort won't go the way of Moonshine.

I love Lua, and I would love to use this. The risk, however, is pretty high in anything but an experimental context -- at least for me.

Sounds like you're yearning for WebAssembly. With WebAssembly you can write your solution in any language it just needs to be compiled to WebAssembly. Every major browser is backing WebAssembly.

As a result (and for example) Microsoft is making it possible to write front-end code in C# which is compiled with a small .NET runtime into the browser with WebAssembly. In theory efforts could be (and likely have been) made to compile Lua to WebAssembly.

Quick search reveals at least two results:

https://scriptinghelpers.org/lua

https://github.com/vvanders/wasm_lua

WebAssembly is not a silver bullet, and is intended for lower level code than I think you expect.

There's little to no interop with the front-end world as we know it (aka the DOM), and any briding to it is still quite expensive. You'd have to re-implement lua's garbage collector in wasm, etc.

wasm is great for high performance data crunching—aka why it's so attractive for 3d engines, games, and the like.

It's less great for developing front end apps; unless you feel like building much of a browser within it.

---

It _will_ get to that point, but we've got a while before interop is a realistic thing

I think the golden question is if browser vendors are going to follow along community wishes and eventually turn the browser into a general purpose VM, or just put a wall and state "this is all you will ever get".
I know currently it's nowhere near a silver bullet. As for not being able to use it now I would say Microsoft has to disagree:

https://github.com/aspnet/blazor

There's plenty of things to work on to get WebAssembly to the state where it could be used as a JavaScript replacement, but none of it impossible. Blazor shows you can build WebAssembly based web front-end applications today. It definitely takes some resources but I'm glad Microsoft is there to work on it to once again leave an example for the rest of the developers who may want to follow a similar pattern for new language implementations.

To be fair, I'm not sure how much overhead their approach may have pertaining to the DOM with the way they're handling things.

I really had to double-check the date to see that it wasn't an April Fools joke :)

Historically, programming languages claiming rewrites in some other 'competing' language have been April Fools fodder e.g. Emacs in Perl https://www.gnu.org/fun/jokes/gnuemacs.en.html

Perl 6 on Javascript is real, though. Rakudo.js has come a long way, but it's not passing all of the tests yet. It's a quite a bit bigger task to convert a kitchen sink language than a smaller language like Lua.

http://blogs.perl.org/users/pawel_murias/

FSF's pedantry in the "Emacs in Perl" joke page is fascinating:

"The Free Software Foundation does not claim copyright for this joke."

Summary: we like it better than JS.
I really hope they eventually push their efforts towards making Fengari a Lua -> WebAssembly runtime. Every modern major web browser has WebAssembly support, it might still be young in development, but it's a start.
WebAssembly still doesn't allow gc cooperation. It's still a long way away
The list of niceties of lua from the article:

> First class functions and closures

> A versatile data structure: the table

> Vararg expressions

> Lexical scoping

> Iterators

> Coroutines (see below)

Well, looking at modern day js:

* First class functions and closures, check

* the table: a combination of `Array`, `Map` and `Map<Number, T>` can prettify much cover typical usecases of lua tables, so js is close in this regard

* Vararg expressions, check

* Lexical scoping, check

* Iterators, check

* Coroutines, with libraries and abstractions on top of generator functions

In addition, in js land, other paradigms of concurrency exists, such as CSP (goroutines), FRP (Rxjs and co). Of course, these can be implemented in Lua as well, but the libraries may not quite as battle-tested.

The author also mentions a sane coercion system; this still true, but ever since "use strict", unexpected coercions can mostly be avoided in js.

On the other hand, js has:

1. Typescript and static typing (a bit unfair to consider this js, but ts gets access to a large portion of high quality npm packages), flow is also great

2. Multiple mature echosystems of UI frameworks (React) - for instance if you need a date picker or an autocompleter, it's just `yarn add` away; I doubt enough such libraries exist for browser side lua yet

3. Tooling support - webpack, prettier, postcss and so on

I like lua, but I feel it does not offer much in addition to what modern js already provides. Of course bring more languages to the browser is always a good thing, on the other hand I do feel this is a tough sell to the larger community.

I think you misunderstand the aim of the article. It’s not a case and point against js but for Lua in the browser. When I list positive points of Lua, I’m not implying they don’t exist in JS.
Javascript generators make for painful coroutines, because a function called by the generator can't yield all the way out.
Yield * fn()?
You can use that sometimes, but it really complicates return values in a way Lua just doesn't have to deal with.
Earth’s Moon in Greek is actually “Feggari”, from the verb “feggo” which means to radiate light. Moon in general in Greek is “Selini”, a natural satellite orbiting a planet.
Yes it’s feggari but I think that with two gammas, it’s pronounced fengari. Could be wrong I’m not greek. But it’s too late to change :)
I am Greek and no it’s not , gg , is pronounce as g in gamma, and one g is pronounced as y in yellow
Reading the sources I used to search for the correct term, I think I may have been fooled by ancient greek allophones. Ex. : ὁ ἄγγελος /ho áŋgelos/ Anyway like I said, too late to change now!
IRC logs from last time it was discussed:

    2018-03-05 23:47:38     assos   in greek it pronounced fegari instead of fengari
    2018-03-06 00:12:41     RhodiumToad     assos: several modern greek dictionaries I checked give the "fengari" prononciation either alone or alongside "fegari" - are they wrong?
    2018-03-06 00:31:44     assos   RhodiumToad: after I checked you are right, the pronunciation depends based on the location in greek, the correct one is fengari. Btw, the word is a derivation of φεγγος which translates to light/brightness.
    2018-03-06 00:31:53     assos   in greece*
(comment deleted)
Is Lua 5.x stable and reliable? Maybe it is, or promises to be now?

On the whole, Lua seems prone to redesign itself and make relatively big changes, like the env changes from Lua 5.1 to 5.2.

I think that works well, since Lua is often embedded and version locked by the application that uses it, but it is the opposite of a cautious and stable evolution of the language.

It’s stable relative to JS which is a nightmare in this regard.
How so? JS is backwards compatible in the extreme, because browsers. New features does not equal unstable.
I’m talking about the cognitive load of the language. Learning JS is a real complex task whereas learning Lua is not. Learning JS as is is hard since its moving so fast. I think nobody can deny the confusion surrounding the language. Lua is stable is the sense that 5 years from now you will still know how to use it even if you don’t keep up with the latest updates.

Also if you have ever dealt with any babel/webpack craziness recently you know what I’m referring to :)

So JavaScript _the language_ is "unstable" because you (for no discernable reason) switch to every bleeding edge JS framework that gets posted to HN?
I’m not talking about frameworks. I’m talking about ES5/6/2016/2017/7/8 ... which all add new concepts and complexity.
You know those new concepts and complexities are optional? And backwards compatible? None of my jQueries broke when generators came out. And that's not even very fast at all. And people have been complaining for the longest time that JS didn't have those features.
This really is not a good argument. You can still write php like in 98, that doesn’t mean you shouldn’t learn modern php if you want to work ever.
But isn't modern PHP moving too fast though? They also break BC with a proper deprecation process. I don't know what you mean it's not a good argument, I've just been casually pointing out that your points are completely invalid. There's actually pretty good reasons to use lua over JS, you don't have to make up reasons.
I think I list a good amount of reasons to use Lua. The focus of the article is far from being fixated on JS being bad.
Maybe "relative to the JS ecosystem" instead?
I remember seeing the talk on this at Lua Workshop 2017. daurnimator gave a great demo. Glad to see it get more publicity and an actual release - looks really promising
I wonder how Fengari does on code size and performance? I don't see any benchmarks offhand.
It’s interesting to me that people are pumped about async/await.

If we time travel back to when Node caqme out, the single most common reason I saw cited for its success was something like: “JavaScript’s simple event-based programming model is a natural fit for a server coordinating I/O from different places”.

Back then we had callbacks and closures and it was just powerful enough you could do almost anything, but simple enough that you were never truly lost.

Now I see developers pumped about a world where callbacks, promises, and async/await are all mixed together in an NPM soup.

I wonder if that mixture might be the thing that finally pushes JavaScript out of reach for beginners (if it’s not there already).

And maybe that opens a door for something like Lua, if it really can embrace async functions without the baggage of Promises and callbacks spread throughout public repos.

Even if that works out though, I do wonder about that original reaction to Node. Maybe Ryan Dahl really hit on a good control flow model that the JS community then threw away. I do think that platform devs sometimes try too hard to make async “disappear” instead of trying to make it inspectable.

Anybody who has used node.js for more than 10 minutes knows that the callback style makes code completely unreadable. Unreadable code is a breeding ground for bugs.

Back in 2009, that was an acceptable trade-off, because there were advantages to JavaScript, such as sharing code between client and server.

Now we have promises and we don't have to put up with callbacks anymore (eventually!). FWIW async/await is the syntax, and promises are the runtime, they are not competing features.

That’s a really insulting way to argue your point. I’ve been writing JavaScript for 15 years. Full time for probably six years, when I dropped Rails for Node and Ember.

Right now I am programming full time in a promise shop for work, so I am very familiar with that. At home I write 100% callback style and find it superior.

The worst is when you have to mix the two, which is why NPM is such a shit show, and why I think async/await will only make things worse.

If you’re willing to spend a few minutes considering an alternative perspective to your own, you might read this: http://callbackhell.com

You say promises are more readable. I do agree promises make it easier to read and write awful code with gnarled architectures. Callbacks force you to name things and separate concerns, which is helpful if your goal is to write good code. If you just want to hack layer upon layer of garbage into the system, as most employers would prefer you do, then promises will help you hold on for that ride longer. Callbacks will break down faster.

I think that’s a feature, as it alerts me to architecture problems sooner. But you may prefer less refactoring and a more additive-only programming model, in which case promises will be a useful tool for you.

If your coworkers are so bad, wouldn't you rather get a nice message in your console when they forget to handle an error? With callbacks, error handling is entirely on the honor system.

There may well be some situations where callbacks are tolerable, but any time you have to converse with an external synchronous resource like a filesystem, database or REST API, it's just painful. Making a new function (named or otherwise) for each line of code is not good style, in any language.