A simple append function like this (in F#):
let rec app s t =
match s with
| [] -> t
| (x::ss) -> x :: (app ss t)
will crash when s becomes big, since the function is not tail recursive. I noticed that F#'s standard append function does not crash with big lists, so it must be implemented differently. So I wondered: How does a tail recursive definition of append look like? I came up with something like this:
let rec comb s t =
match s with
| [] -> t
| (x::ss) -> comb ss (x::t)
let app2 s t = comb (List.rev s) t
which works, but looks rather odd. Is there a more elegant definition?
In addition to what Juliet posted:
Using sequence expressions
Internally, sequence expressions generate tail-recursive code, so this works just fine.
let append xs ys =
[ yield! xs
yield! ys ]
Using mutable .NET types
David mentioned that F# lists can be mutated - that's however limited only to F# core libraries (and the feature cannot be used by users, because it breaks the functional concepts). You can use mutable .NET data types to implement a mutation-based version:
let append (xs:'a[]) (ys:'a[]) =
let ra = new ResizeArray<_>(xs)
for y in ys do ra.Add(y)
ra |> List.ofSeq
This may be useful in some scenarios, but I'd generally avoid mutation in F# code.
Traditional (not tail-recursive)
let rec append a b =
match a, b with
| [], ys -> ys
| x::xs, ys -> x::append xs ys
With an accumulator (tail-recursive)
let append2 a b =
let rec loop acc = function
| [] -> acc
| x::xs -> loop (x::acc) xs
loop b (List.rev a)
With continuations (tail-recursive)
let append3 a b =
let rec append = function
| cont, [], ys -> cont ys
| cont, x::xs, ys -> append ((fun acc -> cont (x::acc)), xs, ys)
append(id, a, b)
Its pretty straight-forward to convert any non-tail recursive function to recursive with continuations, but I personally prefer accumulators for straight-forward readability.
If you love us? You can donate to us via Paypal or buy me a coffee so we can maintain and grow! Thank you!
Donate Us With