Why I cannot use a tuple in `map` directly?

General Usage
1

Why I cannot do this

map((i,j,v) -> v/(i*j), zip([1,2],[10,20],[100,200]))

But this is just fine

[v/(i*j) for (i,j,v) in zip([1,2],[10,20],[100,200])]

And instead I have to write

map(x -> x[3]/(x[1]*x[2]), zip([1,2],[10,20],[100,200]))

which I consider unnecessary as they should be syntactic sugar from the same jar. :wink:

2 
map(((i,j,v),) -> v/(i*j), zip([1,2],[10,20],[100,200]))
2 Likes
3

You can use map, but the syntax is slightly different:

  • (i,j,v) -> v/(i*j) is an anonymous function with three arguments, but zip(x, y, z) creates three tuples which map then passes to the function.

To make the example work, either

  1. pass an anonymous function with a single argument destructuring into a three tuple:
    map(((i,j,v),) -> v/(i*j), zip([1,2],[10,20],[100,200]))
  2. or just pass all vectors as individual arguments to map (it will then call the function on three arguments instead of zipping them into three tuples):
    map((i,j,v) -> v/(i*j), [1,2],[10,20],[100,200])
8 Likes
4

Thank you both for clarifying my misunderstanding that a function that accepts a 3-tuple may also take three inputs and my ignorance that map accepts multiple collections.

I wonder whether the destructuring might be more visible if a keyword as were used, for this example, (t as (i,j,v)) -> v/(i*j) to denote something like

t -> begin
  i, j, v = t
  v/(i*j)
end

similar to the grammar used in SML/NJ and Haskell to bind to a subpattern identifier.

5

The closest shape I can think of

map(t-> ((i,j,v)=t; v/(i*j)), zip([1,2],[10,20],[100,200]))