Adventures in F#--Corecursion

MSDN Blog Postings » Adventures in F#--Corecursion

MSDN Blog Postings » Adventures in F#--Corecursion — Mon, 24 Sep 2007 21:12:32 GMT

PingBack from http://msdnrss.thecoderblogs.com/2007/09/24/adventures-in-f-corecursion/

re: Adventures in F#--Corecursion

Jon Harrop — Wed, 26 Sep 2007 07:24:25 GMT

I found it enlightening to write a pair of mutually recursive functions that extract alternate elements from a given list to return a pair of lists that contain the elements with even and odd indexes, respectively.

Tail

Finnsson — Thu, 27 Sep 2007 15:51:20 GMT

Cool. So F# is using ”tail” after all. What I still don't understand is why F# bother with the rec keyword, since it should be trivial to calculate if tail call optimization is possible or not (and since rec anyway have to make sure that tail call optimization is permitted).

Another strange thing is that F# is making f1 and f2 generic since neither f1 nor f2 can return anything else than int. I would love if F# could type infere f1 and f2 to both int and float in this case, but I don't think it can - so why bother with the generics? Is it something I'm missing here?

re: Adventures in F#--Corecursion

Jon Harrop — Sun, 04 Nov 2007 14:29:26 GMT

Finnsson,

F# has a "rec" keyword because it is often very useful to supercede definitions in a functional language. So you often want a not-rec let binding.

Regarding the inferred types of the "f1" and "f2" functions, these functions never return so F# correctly infers generic types for their return types. This is just one example of type inference providing you with assurances about your code. If the compiler inferred a return type of "int" for a function that should never return then you know that you have made a mistake.

You can learn to leverage F#'s type inference and static type checking to write more robust code. This is described in detail in some of our F#.NET Journal articles. These tricks make it much easier to write complicated code in F#.

Discoveries This Week 06/21/2009

Rick Minerich's Development Wonderland — Mon, 22 Jun 2009 04:39:34 GMT

Almost too much to keep up with this week.  Most notably, Roger Castillo gave a fantastic talk on

re: Adventures in F#--Corecursion

James Hugard — Mon, 22 Jun 2009 17:46:44 GMT

Just as a note, tailcall is disabled in debug mode as of Beta 1. This preserves stack frames, often simplifying debugging.