Intro to CoffeeScript

121 points by aseemk 12 years ago | 89 comments

zalew 12 years ago |

> What I find sad, though, is that most people who dislike CoffeeScript don’t truly know it. They haven’t taken the time to really learn it, or even try it. Their reactions are usually knee-jerk, or based on fallacies. [...] The explanation for most people’s dislike of CoffeeScript is probably our natural resistance to new things and our comfort in what we know.

Oh, wow. I've been using CoffeeScript for about a year for the usual front-end crap, not long ago I re-evaluated Coffee for the second time before starting a full-JS project and certainly won't use it. While I'm a fan of bracketless whitespace significant langs (Python, Sass, Haml) CS's is just bonkers and you never know when a single space will bite you by compiling to a completely different code you had in mind. The ecosystem is weak not to say non-existent comparing to JS, and every JS solution and example out there from minor jquery stuff through SPA frameworks to mobile UI builders is in, well, JS (duh!), so public code reusability becomes a ridiculous task of converting to CS, only to compile it to JS once again. I still use CoffeeScript for small front-end work, meanwhile for any bigger work I find CS an obstacle over JS, and an unnecessary step between my idea and working code; especially when the JS ecosystem has grown a lot of tools to help you deal with what pisses you off (underscore/lodash as an example come to mind).

And from the slide:

> And as of July 2013, it's also the tenth most popular language on GitHub! That's more than Objective-C, ActionScript, C#, Clojure, Go, Groovy, Scala

Yay github charts as evidence of industry value.

Knee-jerk, fallacies, resistance? Call it what you want and good luck. Meanwhile, I will use coffee to keep me awake, not to write apps with.

//edit: care to explain the downvote?

calibraxis 12 years ago | |

In my experience of using CoffeeScript, the author is absolutely correct. CoffeeScript is rejected for reasons "usually knee-jerk, or based on fallacies." I introduced it respectfully and fairly in a former team, and at first people are generally fainting and acting like the world will end.

(This is backend node.js work which emphasized tests and code review. I don't know what it's like outside that.)

Not all rejection is knee-jerk. "Usually" is the operative word here. I would have loved knowledgeable, reasoned critique. I personally assembled as many reasoned critiques as I could, to demonstrate my respect for reasoned, fair decisionmaking. I am continuously disappointed by the low standards of discourse, in programming culture.

(The first random misbegotten idea which pops into my lizard brain to confirm my prejudices, is not a sufficient "reason".)

Eventually, most realize CoffeeScript's sheer maintainability can far outweigh CoffeeScript's (real) flaws; some former detractors become downright zealots. Sure, some still dislike it; but they tend to be ignored because it's really not that bad. Nothing compared to the real problems that a team encounters.

tlrobinson 12 years ago | |

I don't understand the "ecosystem" complaint. Every JS lib is interoperable with CS, so the only things needed in the CS "ecosystem" is syntax specific tooling. We'be got source maps and syntax highlighting for most editors, and most other JS tooling should work fine.

brightsize 12 years ago | |

I love CS myself. To me, in terms of aesthetics, CS is to JS what Scala is to Java. I have been bitten by the whitespace issue that you mention though. I don't recall the circumstance but an indent that was off just a bit did not cause the compiler to complain, but did cause it to generate code that I did not intend. It took some debugging of the JS output to find that one, and it certainly gives one pause to think how similar bugs might slip into the code undetected. I've never encountered this kind of problem with Python, so I wonder if the CS compiler just needs to be a little fussier/smarter with its whitespace processing?

zalew 12 years ago | | |

Compilers/interpreters complain about code that's invalid. Since the code is valid it shouldn't interfer, the problem is in the language's syntax design that's prone to shooting yourself in the foot. The design lets you mix indentations, something that Python won't let you do.

A very simple scenario where hell is a keystroke away:

    hello =
      woot: 1
      yay: 2
      funfunfun: (x) -> x + 1
    
    hello2 =
        woot: 1
        yay: 2
       nowthatsfun: (x) -> x + 1

Other:

    y1 = (x) -> 
        x + 1

    z1: (x) -> 
        x + 1
      
      
    y2 = (x) -> 
        x + 1

    z2 = (x) -> 
        x + 1

http://js2coffee.org/#coffee2js

Those are fairly visible here, not so much on a huge codebase (not to mention with some callback hell attached). You can write perfectly legal code that just doesn't work as expected.

//edit:

worth a read, with more trap examples http://ruoyusun.com/2013/03/17/my-take-on-coffeescript.html

k__ 12 years ago | | |

Comparing Coffeescript to Scala is an outrage! ;)

When I heared about Coffeescript I thought about Scala, too. Not because it reminded me about Scala, but who it took to make Scala and then out of the blue some guy comes along with some new lang, which should ease some problems.

CS is to JS what PHP is to Perl.

aseemk 12 years ago | |

(I didn't downvote you.)

It's great that you tried CoffeeScript and prefer JS. Your opinion is well-formed.

With my statement, I was referring to the many developers I've encountered who haven't tried it, yet actively dislike it and argue against it. Hence the (subjective) "most" in my first sentence that you quote.

Cheers.

zalew 12 years ago | | |

(I know, OPs can't downvote direct comments)

You're right that people should try out stuff before criticising, and in general I support the notion; yet I fully understand people who don't want to jump ship onto very new things because leaving a well-established environment for an immature/unstable one can (but doesn't have to) be a risk. At least Coffee is relatively easy to leave - just take your compiled code and get rid of the source.

I'd also like to add for the rest: what convinced me to try Coffee was @jsjabber podcast with the author http://javascriptjabber.com/017-jsj-coffeescript-with-jeremy... - I think it explains the concepts pretty well (I recommend the whole podcast btw, it's great) and I encourage to listen to it if somebody is curious and wants to try out.

Lambdanaut 12 years ago | |

I'm in the same boat. Used it for a big project and now it's just cruft on my Javascript files. I have to continuously mentally process Javascript code I want to use from other libraries into its CS equivalent, and for what? All I get are list comprehensions and white space.

I originally found it very appealing because I wasn't well acquainted with Javascript or its prototype class system, and so I didn't have that life-saving knee-jerk reaction about CS at all. I dove right in and used it extensively. After getting used to both JS and CS, I've realized that Coffeescript is at best a Javascript obfuscator.

I understand the use of some layers on web tech. SASS is great. It introduces constructs that you legitimately can't get in CSS, like functions and variables. Coffeescript on the other hand is just a toy for the Python devs that are too lazy to properly learn Javascript.

rodw 12 years ago | |

> The ecosystem is weak not to say non-existent comparing to JS

In what circumstance is it appropriate to consider the CoffeeScript ecosystem to be distinct from the JavaScript ecosystem?

> [...] and every JS solution and example out there

It seems impractical to regularly code in CoffeeScript without being fluent in JavaScript.

FWIW I agree with you about CoffeeScript being a little finicky about the syntax. There's the bit a seemingly unimportant (non -leading) space char generating dramatically different results, and also things like:

    if foo and bar is x

not grouping the way you'd (or at least I) expect.

twerquie 12 years ago | | |

> In what circumstance is it appropriate to consider the CoffeeScript ecosystem to be distinct from the JavaScript ecosystem?

Example code, debugging tools, message forums, developer teams.

sdevlin 12 years ago |

> The explanation for most people’s dislike of CoffeeScript is probably our natural resistance to new things and our comfort in what we know.

This is a mildly insulting, as is the invocation of Blub.

CoffeeScript doesn't have a ton of exotic features. String interpolation, list comprehensions, and class definitions (!) are all familiar to programmers from other languages.

I'm guessing that a lot of people have weighed CoffeeScript's feature set against its overhead and decided the trade just isn't worth it.

sehrope 12 years ago |

This is a fantastic slide deck. The per slide comments below are great too. Particularly on an iPhone as you read them simultaneously.

We've been using CoffeeScript for a bit over a year now and never looked back. All of our front end code is written in CS and I can't imagine writing JS again. It's just objectively better. Less code to write, easier to read, minimal cruft, and a joy to use.

The only hiccup was getting the compilation integrated into the build process. We handle it with direct loading of the CS files while working locally with coffescript.js loaded as well. That way code changes are available as soon as you refresh the page. Our build script for the app (when packaging it up for staging/prod) pre compiles the CS files so there's no compilation lag in prod.

I've been experimenting with node.js for a side project and its pretty easy to setup a project entirely in CS. There's a pretty good template with CoffeeScript/Express/Mongoose to use as a template here: https://github.com/olafurnielsen/form5-node-express-mongoose...

benatkin 12 years ago |

This sounds a bit like the bi-polar LISP programmer. Belief that he groks things much better than most programmers grok things? Check. Belief that most people are behaving irrationally? Check. Belief that it's foolish to make an obvious choice? Check.

http://www.lambdassociates.org/blog/bipolar.htm

JavaScript is now a moving target. It's what the examples for new "HTML5" features and libraries are written in. It's what an open source library should be written in if you want it to get traction, and to attract other developers. While it may be a good choice for quickly modifying code on a single project, it isn't as good of a choice for something to get great in as JavaScript is.

There's the saying that you need to know JavaScript in order to write CoffeeScript, but you really only need to be able to read JavaScript in order to write CoffeeScript. My guess is that a lot of developers who mostly write CoffeeScript would trip over simple things for a while if they tried switching to JavaScript. This might keep them from seizing opportunities for JavaScript developers such as contributing to an existing JavaScript project or an open source library that's implemented in JavaScript.

jashkenas 12 years ago |

It's funny to see some of the comments on this thread (or the comments on any thread about CoffeeScript), demonstrating exactly what Aseem is writing in his post. Delicious irony, that.

To make a different point, and one which can easily get missed -- the other unsatisfactory argument I hear somewhat often is: "I love the idea of CoffeeScript, except for X feature, which is terrible, so I won't try it." Where X can stand for scoping, classes, optional parentheses, and the like...

One of the big goals with the CoffeeScript compiler -- despite all of it's flaws and imperfections -- was to annotate and make accessible the source code, and to keep it tiny (I think still somewhere ~< 2,000 LOC), so that if there's one particular thing you don't like, you're free to change it. So if you're intrigued, but dislike X, feel free to fix X before using it, and send your patch along as a pull request. Open source, right?

http://coffeescript.org/documentation/docs/grammar.html#sect...

nsmartt 12 years ago |

I mostly enjoyed the slides, and I'm interested in CoffeeScript, especially since I'm going to be working a little in Node in the near future. This post is a critique on the blog post, rather than on the slide content, which was more or less fine.

Too much focus placed on how those who dislike it haven't given it a proper shot. This could be true, but six paragraphs is overkill. You make a few points: A) they aren't being open-minded enough, B) they are the typical blub programmer, and C) they shouldn't feel bad because it's perfectly understandable to do something so mundane. Had I been someone who had spent a few hours toying with CS and left unconvinced, you'd have lost me pretty quickly.

You mention that you hated it too, at first, but not early enough. I'd submit that you're far more likely to lose readership than to open eyes with this method.

Also, as someone who isn't particularly set against CS, I didn't enjoy reading so much about how I shouldn't feel bad for hating CS at first sight. I didn't.

ww520 12 years ago | |

Sounds like the OP has to justify his large mental investment he made in CS by bashing others who have different opinions. "I'm not a blub like them. I'm not close-minded like them. I'm not doing mundane things like them."

It's a telling sign about a language when the language advocates have to employed negative psychological ploys to convince people (or convince themselves) to use the language.

nsmartt 12 years ago | | |

On the contrary, it seems that OP was going out of his way to explain that it's acceptable to have initially hated CS.

If CS is actually terrible, why don't we talk about what actually makes it terrible? I can't comprehend the notion that evaluating the advocates of X will demonstrate anything concrete about X.

aseemk 12 years ago | |

Thanks for the great feedback. I think you're totally right.

I didn't intend to come across as so preachy in my post. I was just writing freely without (too) much worry.

Live and learn. =)

nsmartt 12 years ago | | |

I can't fault you for that, and I could probably learn a thing or two from writing more freely.

bentlegen 12 years ago |

Why does everybody compare CoffeeScript against vanilla JavaScript? I find using underscore makes my code sufficiently elegant that I've never felt the need to switch to CoffeeScript.

tlrobinson 12 years ago | |

    var newArray = _.map(_.filter(array, function(item) {
        return item % 2;
    }), function(item) {
        return item + 1;
    });

    newArray = (item + 1 for item in array when item % 2)

Uh huh.

oftenwrong 12 years ago | | |

You could also do it with chain, which could make the first example a bit simpler to read and write. It is something like:

    var newArray = _.chain(array).filter(function(item) {
            return item % 2;
        }).map(function(item) {
            return item + 1;
        }).value();

I often use CoffeeScript + Underscore

ceronman 12 years ago |

> Indeed, I, too, dismissed CoffeeScript when I first saw it. But now that I’ve actually learned it and used it, I see how wrong I was: CoffeeScript truly is more powerful than JavaScript.

My experience was exactly the opposite. When I first saw CoffeeScript, I loved it, I started using it in my projects, it was really nice. But once I learned to use it, I stepped into all the annoyances caused by a bad use of significant whitespace, the many ways of doing the same thing, the tooling overhead, and the lack of readability.

In the end, I realised that there was a negative balance of good and bad things introduced by CoffeeScipt. It wasn't worth it. JS wasn't a bad language after all, and the improvements of CS were minimal if you know JS well.

virtualwhys 12 years ago |

I'm all for CS, use it daily, love it. It's absurd how much less code you have to write compared to the generated JavaScript. Huge win for this alone.

However, TypeScript is perhaps the better long run alternative to plain, boilerplate laden JavaScript. Bringing type safety into the mix is pretty compelling, going to check out TS at some point.

chadhietala 12 years ago |

I stopped using CS because it's going to have some issues when ES6 becomes more of a real thing. For instance, how is CS going to handle for/of loops when ES6 has it's own for/of loops that is slightly different. The same goes for class and super.

There will have to be CS2 that introduces new semantics and old apps will have to be migrated.

shardling 12 years ago | |

>There will have to be CS2 that introduces new semantics and old apps will have to be migrated.

I'm sure there will be a new version of CS when ES6 becomes widespread, but why would old apps need migrating? CS produces regular JS, and thus will be 100% compatible with ES6.

zalew 12 years ago | | |

Actually coffee's js is currently compatible down to IE6, which the author said is what he wants to support at this point. So unless MS were planning to backport ES6 compatible JavaScript to the dead IE6, CoffeeScript supporting ES6 will either change/break at some point or become a unreasonable backward-compatibility clusterfuck (probably the former). ES6 becomes official in late 2014 so there is probably enough time for a lot of changes to happen.

thejsjunky 12 years ago | |

This issue was actually raised a couple weeks ago here on HN and Jeremy responded thusly: https://news.ycombinator.com/item?id=6079709

fordi 12 years ago |

It would have been great if he had mentioned IcedCoffeeScript at some point too. Coffeescript by itself is 'ok' but IcedCoffeScript - at least on the serverside - is fantastic.

aseemk 12 years ago | |

Indeed! I have much respect for Max and company. We use and love Streamline.js ourselves -- works great with CoffeeScript.

https://github.com/Sage/streamlinejs

I'd love to talk on or write about Streamline at some point too.

sergiotapia 12 years ago |

Is CoffeeScript used much outside of Rails? If not, wouldn't it be a better idea to focus on mastering javascript so it applies globally?

agilebyte 12 years ago |

I would just say that there are big differences between a JS `for` loop (`for in` in CS) and an ES5 `forEach`. The fact that you are creating a new function scope and are not returning all elements as you might expect:

  arr = new Array 3
  arr.push 4

  ( alert el for el in arr )

  arr.forEach alert

pspeter3 12 years ago |

My biggest issue is that the language can be ambiguous and I'm not sure how to check what I'm compiling to without running the short cut all the time in Sublime Text 2

tracker1 12 years ago | |

the coffee console app has a watch parameter iirc.. as do many build tools...

pspeter3 12 years ago | | |

I'm not worried about the actual build step. I'm talking about that sometimes it's hard to figure out how the CoffeeScript compiler is parsing and generating the JavaScript

itsbits 12 years ago |

for me its upto the developer to choose between JS or CS. I used CS for some time and i liked it. But still i preferred to switch back to JS only to not to waste my time in correcting CS syntax issues and play with JS.

consonants 12 years ago |

Yes, let me learn this domain specific language for a domain specific language so that I can learn another domain specific language whenever this one stops trending.

pjscott 12 years ago | |

If you know Javascript, learning Coffeescript is easy. I became proficient in Coffeescript in a few hours, and I sure don't feel like a genius, so other people can probably do the same. It's just a more convenient way of writing Javascript, with a nicer syntax and some really pleasant features.

consonants 12 years ago | | |

The issue at hand isn't whether or not it is easy or hard to learn, it's that the rabbithole can go on forever as all language implementations will manifest inadequacies which you will have to 1) be familiar with 2) know how to work around.

It's an inherent phenomenon that DSLs will change and come and go at an incredibly fast rate. I just don't find it worthwhile to write code that is dependant on the DSL du jour and pigeonhole myself and whatever unlucky sole has to maintain the legacy code in the future. Imagine being dumped on a mountain of ColdFusion legacy code.

I just find it kind of absurd is all.