[Haskell-cafe] An ugly zip3 problem..

[Michael Feathers]

One thing that gets me about this solution.. as I was structuring mine I 
noticed that I was ending up with types like FilterWindow3 and functions 
like lowPass3.  Inlining does eliminate them, but I wonder whether there 
is a good way to structure the computation generically so that it can be 
performed with windows of 5 as well as 3.  The cons pattern matching 
here would get in the way, and in my original solution, the fact that I 
was using tuples got in the way also.

Would Haskell's type system allow you to pass a function of arbitrary 
arity, discern its arity, use that information to construct the 
appropriate structure for iteration, and then apply it?


Michael

Tillmann Rendel wrote:

> 
> by inlining the definition above, this can be given as a four-liner now:
> 
>   smooth n = (!! n) . iterate f where
>     f ds = [(g y z + 2 * y + x) / 4.0 | x:y:z <- tails (head ds : ds)]
>     g x []    = x
>     g _ (x:_) = x
> 
> :-)
> 
>   Tillmann
> 


-- 
Now Playing: http://www.youtube.com/watch?v=SsnDdq4V8zg