some thoughts on Clean 2.0

[rinus plasmeijer]

Dear Panitz,

> The other day I found an old clean manual on my book shelve:
> 'concurrent clean language manual version 0.8.3 february 1993'.
> In those days I was playing around with clean on a small
> apple powerbook (with 4mb memory..)
> 
> Well that is some time ago! Nowadays I mainly use clean on my small
> linux laptop or an old mac desktop.
> 
> I remember the fist time when Rinus Plasmijer presented clean 1.0 on a
> conference and how I was looking forward to that release.
> Now we are waiting for the second mayor clean release: Clean 2.0!
> 
> If I understand the information on the web site correctly this will
> be a completely new dimension for clean and clean no longer looks
> like a fast Haskell compiler with a little bit different syntax
> and a good IO library, but a more powerful language with a lot of
> additional features:

> dynamics:
> is this what I think? like in good old lisp days converting a string
> into an epression? Wow! I could have used that in a program, where I now
> wrote a parser for certain expressions.

This is true. We now have the first examples running.
One can dump any Clean expression in a dynamic on file (including
closures), read it in another Clean
program where the corresponding code is linked in dynamically.
No parsers needed !

> 
> btw: those very long threads in comp.language.functional 'static vs dynamic
> typing' will have a simple answer in the future: Clean! Especially I wonder
> what the Erlang people will say. They insist that dynamic typing and linking
> is needed in their area of application.

I agree. With dynamics you have both static as well as dynamic typing.

> 
> Since the new Clean compiler is written mainly in clean, does that mean
> that with dynamics we also get a Hugs like interpreter for clean?

Maarten de Mol is working on the proof system for Clean.
It has a great user-interface including a Clean interpreter which can do
symbolic reduction as well.
It might be a good idea to use the dynamic linker to include code
dynamically such that one has an ideal mixture of interpreter and
compiler.

> 
> And you are also planning to make the parts written in clean open source?

All Clean source code will be put on the net.


> Well who knows what will happen with that. Maybe someone who does not know
> what to do with his/her spare time will write a Haskell frontend, so
> that one can have mixed Haskell/Clean projects (and moving from
> Haskell to Clean will be fairly easy, or using Haskell libraries will be
> possible).

As a matter of fact, I am looking for volunteers to make such a
front-end.
The idea to change the Clean parser to accept haskell code.
It would be nice if Clean and Hassle modules could be combined into one
application....

> How difficult would it be to write such a Haskell frontend?

Most of it will be very straightforward.
The main problems I see will be caused by the different module structure
of Clean and Haskell.

> I guess the Clean compiler has some intermediate language,
> so a Haskell parser and type  checker as well as a
> transformer to that intermediate language would be needed.

No, I think you only need to change the parser to accept Haskell syntax.
Clean compiles to Core-Clean. One can also translate Haskell to
Core-Clean.
The main problem will be the module structure (Clean has dcl and icl
modules which are much more convenient than the Haskell module
structure).

> And these things could be derived from the Clean parser and typ checker
> or from some Haskell compiler written itself in Haskell (e.g. nhc and
> then the 'n' of nhc could have a new meaning ...:-)).

This would be one way to do it. I would prefer just to change the
current Clean parser...

> 
> Well, I am looking forward to Clean 2.0, but I know, in order the get the
> quality we are used from the team in Nijmegen, we simply have to wait
> a bit more.

Sorry about that. We could release the compiler today.
But, we would get many bug reports.
We rather test it until we believe the system is fine.


By the way: we will release Clean 1.3.3 very soon. Next week. This week
we are almost all joining IFL in Aachen.
The 1.3.3 system contains some small improvements over 1.3.2 and
contains some support for Clean 2.0.
We have tested it for over a year.

> 
> Regards Sven Eric,
> (who still could not convince his boss to make a project in
> Clean and is therefore forced to make his living with Java hacking :-( )

Life is hard..

Greetings,

Rinus