CS2 Discussion: Question: Survey for devs/teams with maintained Coffescript codebases: happy with CS, or considering migrating?

[coffeescriptbot]

From @greghuc on 2016-09-10 11:21

Coffeescript has been around since 2009. Since then, a lot of Coffeescript code has been written, and non-trivial projects developed with it.

Given we're discussing the future of Coffeescript, it would be great to get the views of developers/teams who have non-trivial, maintained CS codebases, which they plan to maintain and extend over the coming years.

Specifically, it would be good to know whether you/the team are:
a) happy to keep using Coffeescript for the project
b) are considering migrating the project from Coffeesript to ES6/other.

Here's a little markdown 'survey' template below to use, or feel free to comment free-style.

**CS project summary.**
Quick summary: n developers, lines of code, years established, etc

**Is the project considering a switch from Coffeescript to ES6/other?**
No, or yes with language being switched to.

**If considering switch from Coffeescript, please explain your reasons.**
Give your reasons.

**If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?**
Give your reasons.

[coffeescriptbot]

From @greghuc on 2016-09-10 11:21

CS project summary.
1.5 developers, 14K LOC, 3 years established.

Is the project considering a switch from Coffeescript to ES6/other?
Yes, to ES6, but not immediately.

If considering switch from Coffeescript, please explain your reasons.
I'm uncertain over the long-term substainability of Coffeescript. As Javascript develops at rapid speed in both browsers and nodejs, I want confidence that Coffeescript will still interoperate with latest Javascript+libraries in 3-5+ years. Recent incompatibilities with ES6 have raised uncertainty for me (broken tagged literals, and reports of ES6 inoperability. So I might de-risk the project by switching to ES6 Javascript.

If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?
I feel more productive and less likely to make bugs with Coffeescript. It's a small, clean language with a functional feel. I don't consider ES6 massively superior to Coffeescript. Yes it's introduced some Coffeescript-influenced features, but it's also added even more complexity to Javascript.

[coffeescriptbot]

From @nwinter on 2016-09-10 14:49

CS project summary.
Quick summary: CodeCombat, 7 core developers and hundreds of open source contributors, about 100K lines of code, 3.5 years old

Is the project considering a switch from Coffeescript to ES6/other?
Not actively considering, though a couple of us would like to. Maybe eventually. We'd go to ES6 or maybe Typescript.

If considering switch from Coffeescript, please explain your reasons.
It's a polarizing language that can make it harder to hire developers who think they won't like it. There's a perceived lack of momentum compared to ES6. There are a few cool things in ES6 we'd like to use. It's more future-proof.

If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?
Primarily the cleaner, more powerful syntax. There are some language features I can't live without. Also, no need to migrate our code. We haven't run into any interoperability problems yet.

[coffeescriptbot]

From @greghuc on 2016-09-10 21:37

Re existing CS projects, @devongovett justifies porting fontkit from CoffeeScript to ES6 here. He also plans to migrate pdfkit to ES6.

[coffeescriptbot]

From @alangpierce on 2016-09-11 01:53

CS project summary.
Benchling, 17 developers (about half of whom joined in the last 6 months), about 200K lines of CoffeeScript code, about 3-4 years old.

Is the project considering a switch from Coffeescript to ES6/other?
We've already decided to switch to JavaScript, and new code is written in JavaScript. I've been contributing a bunch to decaffeinate in my free time to make it easier to switch.

If considering switch from CoffeeScript, please explain your reasons.

• JS generally has better tooling, like linting, code coverage, etc.
• We were unhappy about various sharp edges in CoffeeScript, primarily implicit returns.
• We prefer JSX for React code, and we weren't comfortable with the risk and build complexity of CoffeeScript JSX solutions.
• We write fairly disciplined code and we're hoping to use flow at some point, which requires JavaScript.
• Once we decided to write new code in JS, moving off of CoffeeScript became important to reduce ramp-up cost for new people.

If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?

We're agreed on switching to JavaScript, but there are certainly some hesitations:

• Some people on the team still like CoffeeScript better, particularly range syntax and ?..
• Switching programming languages takes a lot of work and isn't particularly important for any company goals. By my super-rough estimates, converting the codebase to JS would take 4000 person-hours without decaffeinate and 400 person-hours with decaffeinate in its current form. That's a lot of time that could be spent building useful things that help people.

[coffeescriptbot]

From @greghuc on 2016-09-13 07:52

Feedback from Atom Editor at atom/atom#12656. Relevant quotes:

• "the current strategy appears to be that when new files are checked into the codebase, those should be written in ES6. No more Coffeescript files should be added. Existing coffeescript files may be converted to ES6 when they undergo a major rewrite."
• "Only suggestion I'll make is that you swap the freakin' of and in operators the other way around so devs won't keep confusing the two when flipping between CoffeeScript and ES6."

[coffeescriptbot]

From @honzajavorek on 2016-09-13 08:59

CS project summary.
Quick summary: Dredd & Apiary, ~15 developers at the moment, thousands of lines of code, mostly since ~2012

Is the project considering a switch from CoffeeScript to ES2015/other?
We decided to write new projects in ES2015 if possible. There's currently no strict policy about existing CoffeeScript codebases (which we have plenty of), but they're maintained and kept being developed. If we're tearing off some parts of larger projects into separate libraries or services, they're almost always started from scratch as ES2015 codebases. There's currently no plan to systematically decaffeinate any existing projects, but the codebases are considered as legacy.

If considering switch from CoffeeScript, please explain your reasons.
The company started with CoffeeScript in the past mostly because it was able to abstract away annoying limitations of standard ES5 JavaScript. Now that ES2015 offers new features and less annoying syntax sugars, there's no reason to use CoffeeScript anymore. While many of us are Python and/or Ruby fans, we don't see the benefit in using CoffeeScript just for the "clean" syntax - on the opposite, the syntax is sometimes way too benevolent (namely missing parentheses) and some of it's features are designed dangerously (variable scoping, implicit returns), so in the result the language produces mistakes and unreadable code rather than prevents them. We made our own strict style guide to prevent common mistakes. The language is now also now missing features which are already available in ES2015 and isn't able to keep up with growing JS ecosystem and tooling (e.g. eslint, coverage tools, ...). Basically, ES2015 seems like the future to pursue, while CoffeeScript turns out to be the (necessary) past to make the future happen.

(Fun fact: I agree with switching to ES2015 now, but if the company was using plain JS instead of CoffeeScript in the past, I probably wouldn't even consider to join them. I was very strongly Python-oriented and I completely hated plain JS. With CoffeeScript it was at least bearable.)

[coffeescriptbot]

From @davibe on 2016-09-13 19:32

CS project summary.
Quick summary: Yakyak (https://github.com/yakyak/yakyak), 2 devs + contributors, almost 6k lines. Also a private startup, 5 devs, 40k lines of code, 10 months.

Is the project considering a switch from Coffeescript to ES6/other?
We are resisting. We are waiting for a solution that allows us to support new features with no retarded syntax involved.
Yakyak is written in functional style therefore everything-is-an-expression of coffee fits better. A benefit of Javascript would be easyer contributions from a larger community.
In the private company we mostly care about async/await support at the moment. We are using generators in the meantime like this

If considering switch from Coffeescript, please explain your reasons.
If we have no choice we will eventually switch mostly for async/await support, better tooling, easier adoption for new people.

If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?

The fact that in CS everything is an expression, the cleaner syntax, more powerful and expressive. Small things like?.this?.one isnt 'trascurable' for most in people when not most.crazy may look stupid but they can really make things more concise, readable and most of the times you don't need external libs like underscore/lodash.
Its true that it's harder to find coffeescript devs and it takes time to learn it, but it pays off.

Babel is the only way to use ES6 stuff BUT it adds a build step OR it's super-slow (using require-hook). I also had problems with exception stacks not being readable and source maps not working as well as coffee. I don't consider it a decent backend replacement because it just adds overhead on our case. It's ok for a webapps indeed we do use ES6 js + jsx for frontend code.

[coffeescriptbot]

From @jcollum on 2016-09-13 19:47

CS project summary.
Quick summary: 1 developer, several projects, about 3k lines of code including tests, Hapi.js and pure node, 18 months old for the oldest

Is the project considering a switch from Coffeescript to ES6/other?
No, not the project. Unless we decide it's very important to re-do the codebase. Unlikely.

If considering switch from Coffeescript, please explain your reasons.
Yes, considering switching new codebases to ES2015. Mostly to avoid getting left behind if I ever need to look for a new job.

If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?
I would prefer to stay with CS (vastly prefer the syntax) but the lack of ES2015 features gives me fear of a dusty resume. FODR instead of FOMO? Coffeescript was a reasonable "sell" to a potential employer but now it's outdated and I can't recommend it for new codebases.

[coffeescriptbot]

From @jcollum on 2016-09-13 20:14

@davibe I have to ask if you've considered LiveScript. Much more functional / closer to Haskell than CS.

[coffeescriptbot]

From @davibe on 2016-09-14 05:59

@jcollum no. Quickly looking at it the syntax already has horrible things that i don't need to know % *+like |> <- this. 'cmon. I would even remove @, => and maybe class from coffee if it was just for me. Sorry for being grumpy, I am exaggerating a bit i know :)

[coffeescriptbot]

From @JimPanic on 2016-09-14 06:03

No need to apologize, @davibe! I can relate re @, => and class. Although the pipe operator |> (also found in Elixir) is very handy sometimes! ;)

[coffeescriptbot]

From @mrmowgli on 2016-09-14 10:21

CS project summary.
2 developers, 35k+ lines of code, 4 years of projects and contract work in CS.

Is the project considering a switch from Coffeescript to ES6/other?
No, with some caveats. Frameworks like Meteor have already switched to shimmed ES6, and I am running into lots of issues updating projects. For the most part code is still CS, with some ES6 wrappers to deal with framework peculiarities.

If considering switch from Coffeescript, please explain your reasons.
Not interested in a world without CS. Good god, back to all those semi-colons and matched braces all over the place?

If not considering switch from Coffeescript (or are, but would ideally like to remain with CS), why are you happy with CS?
CS is by far the best way to approach Javascript. I know the ES6 features are the new hotness, but lord, I am happy with the CS classes, happy with the syntax, happy with conciseness and readability. I know everyone wants to keep up with the joneses, but CS is still a completely workable language form many tasks. Once we get some of these big ticket items (Thank you @JimPanic & @GeoffreyBooth for module loading!! ) cleared up then I can get back to clean beautiful code! Also I would love to see a day when all the ES coders end up adopting CS as the OFFICIAL version of ES, instead of just taking useful parts and mangling them. * sigh *

[coffeescriptbot]

From @jcollum on 2016-09-21 16:34

@GeoffreyBooth suggested I repost this here:

Hey I don't know where else to put this so:

I'm working on a new Hapi.js project the last few days. Decided to do it in ES2015 since a couple of junior devs are helping me. HOLY CRAP I MISS COFFEESCRIPT. All these damned extra braces and semicolons and such, clutterin' up my code. And the lack of easy key/value list comprehension. Etc.

So, thanks so much for participating here. Know that your work is appreciated.