[Haskell-cafe] "automonadization" of code?

[Adam Megacz]

Is there any work on automatic translation of code in some tiny
imperative language into Haskell code that uses the ST and/or IO
monads (or perhaps even pure functional code)?

For example, the user writes something vaguely like

   array = newArray (1,100) 1
   for x=2 to 99
     array[x] := array[x-1]+array[x+1]

And it is transformed into something like

   foldl
     (>>=)
     (newArray (1,100) 1)
     $ map (\n arr -> do a <- readArray arr (n-1)
                         b <- readArray arr (n+1)
                         writeArray     arr n     (a+b)
                         return arr)
           [2..99]

Obviously the "small imperative language" would have to be highly
restricted and carefully chosen in order for the translation to always
work and be predictable.  I'm interested in any existing work on
choosing such a sublanguage.

Thanks!

  - a

-- 
PGP/GPG: 5C9F F366 C9CF 2145 E770  B1B8 EFB1 462D A146 C380