Out of curiosity I ran this over some JS code I've written before. I noticed that JS converted to coffee looks absolutely shitty compared to code that was written in coffee to begin with.

This leads me to think how much of your code style can be influenced by the syntax of your language.

coffee's syntax encourages your to create many small functions which gives coffee code its uinque look. In JS, creating functions is just cumbersome enough to type so you don't do it as often, leading to completely different code.

(my https://github.com/pilif/tempalias/blob/master/lib/tempalias... is a very good example of quite-acceptably looking JS that gets butchered looks-wise when translated into coffee)

awesome, especially since I am planning to move couple of projects to rails 3.1

most popular libraries easily got converted to coffee

but faced some problems with hand coded javascript files from a project. "Label not supported" error. going to test those with jslint

you know, there's no need what so ever to convert your existing JavaScript or any third-party libraries to coffee script when you are moving to rails 3.1.

You can just keep your existing JS code, or, if you want to port that, call directly into JS libraries from your new coffescript code.

Why make another converter in python when we have one in JavaScript already?

https://github.com/mindynamics/js2cs

For one thing, notice that there haven't been any updates to that project since July 2010. The CoffeeScript language has undergone serious changes in that time, so output from that converter is largely unusable.

I made a quick gist showing jquery's source getting converted to cs then back to js

https://gist.github.com/1000151

curious if you benchmarked it, if it would be any better.

I hacked together a quick comparison using http://mootools.net/slickspeed/ as a base. I'll redo it publicly if anyone has interest.

The coffeescript version is most certainly worse, but negligibly so. It adds about 1-2 ms to half of the functions.

That being said, thats not surprising. Just like when you make a copy of a copy you only loose. You'll never gain by compiling back and forth.

It was one of my complain about CoffeeScript[^1].

This with the existence of CoffeeScript.eval and now, I believe I will use CoffeeScript.

[^1]: http://yannesposito.com/Scratch/en/blog/2011-01-03-Why-I-sad...

Yannes, re your last point, look into the => function glyph, which handles the function binding issue you describe.

Yes, thank you.

There's probably great uses for this, such as learning and adding on to other CoffeeScript scripts. I think it's hilarious that Javascript converts to Coffeescript, which then converts to Javascript.

I guess this fits the definition of a disassembler.

I'm sorry, surely this is taking it too far?

It seems like it could give people a nice headstart if they want to port their javascript codebase to coffeescript. At the very least, it's an interesting side project for the author.

It's a nice side project, but I can't see the functional usage of coffeescript in the long term. The whole concept of "compiling" to get valid javascript seems odd. It's not as if JS is a hard language or a particularly verbose language.

It's an ugly language.