[Haskell-cafe] Google Summer of Code idea of project & application

[Damien Desfontaines]

2012/3/19 Richard O'Keefe <ok at cs.otago.ac.nz>
> On 19/03/2012, at 8:01 AM, Damien Desfontaines wrote:
> > The project I suggest is mainly inspired by Ticket #1555 [1] : I think that
> > would be a great idea to make it possible to call some Haskell code into
> > OCamL.  In particular, this would contribute to the spreading of Haskell in
> > countries where OCamL is proeminent, mainly France and Italy. The idea would
> > be the following : building a translator which would turn Haskell code into
> > (purely functional) OCamL code, in order to enable the use of Haskell
> > functions and libraries within OCamL programs, in a "human-readable" way
> > (the OCamL source code generated would ideally be understandable enough to
> > be manually modified).
>
> You might want to consider targeting F# as well as (or instead of) OCaml.
> I've had nothing but trouble with GODI, to the point where I gave up on OCaml
> entirely.  On the other hand, F# came with Mono...

Thank you for answering that fast and for your advices.

I'm afraid I have absolutely no experience with F#. I guess I can learn it in
several months, I heard it is derived from OCaml, but I think I would be really
less efficient working with a brand new language such as F# instead of OCaml,
which I already master.

But the real question is : what would be the most useful ? I am quite convinced
that I could work faster and most efficiently with OCaml than with F#, but if
the community considers that such a project would be far more useful if I would
"translate" Haskell into F# instead of OCaml, I can adapt.

> F# has built-in support for lazy evaluation (although it is not the default),
> so this might simplify your task.  Indeed, F# has comprehensions too, so the
> main impedance mismatch would be typeclasses.  This would make an F# target a
> sensible half-way point for an OCaml target.

OCaml has a built-in module for lazy evaluation as well (even if it is not in
the Pervasives (= default) module, and a syntax for list comprehensions as well.
So, in my opinion, the main challenge would be dealing with typeclasses, exactly
like in F#.



2012/3/19 Stephen Tetley <stephen.tetley at gmail.com>:
> Hi Damien
>
> A translator might be a lot of work.
>
> Matthew Naylor had a translator between Haskell and Clean [1], which
> performed well according to [2]. The translator was his Master project
> in the UK so I think that means it would represent approximately a
> years work.

Thanks for your answer. I must admit that I do not really realize how much work
such a project represents. I will probably need the help of someone who is more
experienced than me to decide my timeline, and perhaps to restrict the final
goal of my work (perhaps to a syntaxic subset of Haskell ?).

If someone is interested in mentoring me for this work, I would be glad to
discuss those technical details with him.

Damien D.