[MLton] Trip Report -- Commercial Users of Functional Programming

Daniel C. Wang danwang@CS.Princeton.EDU
Mon, 27 Sep 2004 18:14:32 -0400


In all seriousness, I suggest there be some serious consideration to taking 
the F# fronted and retrageting it to run on MLton.  I really, like to think 
of MLTon as some amazing compiler technology that happens to have an SML 
frontend. If MLTon suddenly started to suport a very interesting OCaml 
subset I think many people would take notice.

Since, I think F# bootstraps. The low fuss hack is to take the frontend and 
add a prettying pass that spits the elaborated IL out as a valid SML program.

Then you can compile that with MLTon. With a little extra dancing you can 
avoid the pretty-printing phase and just link the F# frontend converted to 
SML as if were native SML code.

I suggest stealing the F# frontend, rather than the actual Ocaml frontend 
because I suspect it is less crufty than the current Ocaml frontend, and 
I'll bet it is more suited to this high-level source to source translation 
approach than the current Ocaml compiler from Inria.


Stephen Weeks wrote:
>>Were there any SML users?
> 
> 
> Other than me, sadly not.  At least not currently.  Galois did use
> SML for one of their early projects where the CKit came in handy.  But
> they're pretty much a Haskell shop these days.
> 
> I of course mentioned the wonders of MLton/SML to those who would
> listen. :-)
>