Hacker News Reader

Comment by emmelaich | Hacker News Reader

You wonder why Clean isn't as popular as Haskell and Ocaml.

I've read it's the lack of community. The name doesn't help either.

The only reason why clean has always been confidential is this:

"Developers wishing to distribute commercial applications (either publicly or privately) can purchase a commercial license."

Although there seams to be a dual license now? Can anybody with more understanding of legal verbiage confirm that if I choose to use the bsd+lgpl then the above restriction don't hold anymore?

No change, according to c2wiki[0] it has always been phrased like that.

[0]: http://wiki.c2.com/?CleanLanguage

According to the wiki linked into the other answer

> The company is not claiming that you can't sell your programs. The licensing only matters when it comes to modifying the compiler itself.

which it makes sense. The language runtime is LGPL but the program you write is anything you want.

At http://clean.cs.ru.nl/download/Clean24/CleanLicenseCondition...

> Clean is available under a dual license. Users can choose which of these two licenses they wish to operate under:

> 1 The Simplified BSD License (see below) applies to the libraries, runtime system and examples, the LGPL the standard GNU Lesser General Open Source license (see below) to the rest. The libraries, runtime system and examples consist of the files in the following directories (including subdirectories of these directories):

> - Libraries and Examples (versions for Windows)

> - StdEnv, data and examples (versions for Linux and Mac OS X)

> - libraries, RuntimeSystem and CleanExamples (source code).

> 2 A commercial license (see below) that can be purchased. Information on that license can be obtained from [check the original document for name and email]

So, it should be ok as using any LGPL library on any proprietary program. I wonder if they are making any money from the commercial license or this confusion is only harming the language.

No, I don't. I actually wonder why it gets mentioned. It is some professor's research project not an actual programming ecosystem. The documentation on the site is more than a decade old, it covers version 2.2 released in 2006. Want documentation on what has been happening since then? Pick up a research paper. By comparison Haskell has a dozen books on Amazon and plenty of newbie level documentation since multiple universities use it as a teaching language.

It took ages for Haskell to become "popular". My personal impression is that OCaml was more popular in the past and lost ground in recent years. OCaml's story demonstrates though how difficult it is to turn an academic research language into a widely accepted tool that is (financially) supported by the industry.

Clean had a performant compiler when Haskell was mostly interpreted and slow. Why didn't it take off? Maybe it wasn't the main goal of its developers once the publications were done.

Languages don't win. Package management and distribution wins. Access and energy required for a viable solution.

C won befoee package managers and Linux distributions existed.

Because it was the language of UNIX distributions (not called that, but it's the same concept).

They went hand in hand. I know lots of people that installed Linux in the 90s so they could use GCC.

well, unix was written in it, making it a natural choice if you're writing unix programs, or if you are writing programs on a unix.

And symbiotically, C was written to write unix.

edit: to note that by 'written' i mean designed and initially implemented.

If any of its competitors had had package managers as good as some of the modern ones, C would have lost.

Excellent point. Haskell got Cabal in 2005, and even if it to this day continues to be rather awkward to use, it is something. Clean never developed anything for this.

I believe a large part of Rust's popularity is due to its superb library infrastructure via Cargo and crates.io. Furthermore, I believe a large part of its contemporary competition (say, go, swift and crystal) tripping on its own feet is the lack (or weaknesses) of such a thing.

> Why didn't it take off?

afaict there was a moment when most of the "pure" "lazy by default" functional language projects coalesced into Haskell to build working mass -- except the developers of Clean continued to see value in strictness analysis and uniqueness typing but that limited the numbers developing libraries and tools.

An additional hurdle may be it's use of uniqueness typing (http://clean.cs.ru.nl/download/html_report/CleanRep.2.2_11.h...), which it uses for IO instead of monads.

In my experience with Clean, uniqueness typing has been some kind of magic.

Funny how interpretations differ. I always considered the uniqueness typing approach to IO superior to the monadic approach.

I don't know if it's changed much, but I remember that several years ago I thought about trying to learn Clean but I gave up because it looked like Linux was a second-class platform and the Clean project mostly focused on a Windows IDE.

Can anybody comment on this?

Their documentation states: "The latest version 2.0 of the IDE is currently only available on the Wintel and MacOS/PowerPC platform. On the other platforms a much more restricted and older version of the IDE is available. "

So it sounds about right to me.

iirc the compiler and libraries worked fine with Linux, but work on the IDE was not being ported to Linux.

Example, Clean programs working fine on Linux --

https://web.archive.org/web/20130101063215/http://benchmarks...

it was being developed on windows, and the gui library didn't work on linux. at least that's what kept me from picking it up, and i suspect a lot of the language enthusiasts who might have adopted it were similarly linux users, especially in the early 2000s

Reply on news.ycombinator.com