Re: FC++ in Boost, Spirit, Preprocessing

From: David Abrahams (dave_at_[hidden])
Date: 2003-07-31 16:42:14

Jan Kybic <yg-boost-users_at_[hidden]> writes:

>> Regarding the preprocesor, I've had the idea once but I guess I don't have
>> enough time to pursue it. I'm not sure but are you implying that you want
>> to volunteer to write a Haskell preprocessor that emits FC++ code?
>> Possibly using Spirit? That would be awesome!
>
> I do not think it would be worthwhile to use Haskell as the input
> language of the preprocessor, because then you could just use the ghc
> Haskell compiler. Also, even though I quite like Haskell and I have
> been using it for some time, I found it a bit too lazy for the
> my applications involving large-scale numeric computing where
> efficiency and control over memory management is necessary. I found
> C++ to be a better language in that aspect. 
>
> Therefore, I propose to use the best from both worlds, the Haskell
> type system for defining function signatures and C++ for writing the
> function bodies. The first step could be just a very stupid
> preprocessor to save some typing. I do not really have a clear idea at
> the moment, it needs to be thought out well, it could be something as
> simple as:
>
> @fun plus :: A a -> B b -> A result
> @{
>         @let c = a+b   // type automatically deduced
>         return static_cast<A>(c) ;
> @}
>
> which would provide a polymorphic functor plus.
>
> As for me volunteering to write the preprocessor, I am obviously
> tempted, but very busy at the moment and also I think it might be
> better to wait a month or two whether there is any outcome from the
> cross fertilization of Spirit/FC++/....

You guys oughta check out the Felix project:
http://felix.sourceforge.net/

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com