Question

Does it make sense to have such functions defined

let squash12 (e:('a*('b*'c)   )) = e |> (fun (a,(b,c)  ) -> (a,b,c  ))
let squash21 (e:(('a*'b)*'c   )) = e |> (fun ((a,b),c  ) -> (a,b,c  ))
let squash13 (e:('a*('b*'c*'d))) = e |> (fun (a,(b,c,d)) -> (a,b,c,d))

let seqsquash12 (sa:seq<'T>) = sa |> Seq.map squash12
let seqsquash21 (sa:seq<'T>) = sa |> Seq.map squash21
let seqsquash13 (sa:seq<'T>) = sa |> Seq.map squash13

I could not find another way to make my core code recursive (leading to nested tuples), yet expose simple function that maps to generalized n-dimensional coordinates.

Was it helpful?

Solution

I would have marked your functions as inline so that they could just be

let inline squash1 (a,(b,c)) = (a,b,c)

Also, you don't need the lambdas (fun ...)

OTHER TIPS

Yes, it makes sense to do so. The suggestion is avoiding lambda to make these functions easier to read:

let squash12 (a, (b, c)) = a, b, c

If you encounter inner tuples with varied arity very often, converting them into lists is not a bad idea. For example, e becomes a tuple of two lists:

(a, (b, c)) ~> ([a], [b; c])
(a, b), c) ~> ([a; b], [c])
(a, (b, c, d)) ~> (a, [b; c; d])

And we only need one function for sequence:

let seqsquash sa = sa |> Seq.map (@)

The problem is you lose the control over size of input. Pattern matching on list could help:

let squash12 (xs, ys) = 
   match xs, ys with
   | [a], [b; c] -> xs, ys
   | _ -> failwith "Wrong input size"
Licensed under: CC-BY-SA with attribution
Not affiliated with StackOverflow
scroll top