Do you think it's useful and possible to extend the D type inference (deduction) for templates to support something like this (I know there are different ways to do this in D)?


struct Tree1 {
    static struct Node { int x; }
}
struct Tree2 {
    static struct Node { int x, y; }
}
void foo(T)(T.Node n1, T.Node n2) {} // ***
void main() {
    Tree1.Node n1, n2;
    foo(n1, n2);
}


(I am asking this because nested types are becoming increasingly common in D code.)

Bye,
bearophile

On Monday, 17 September 2012 at 14:54:48 UTC, bearophile wrote:
> Do you think it's useful and possible to extend the D type inference (deduction) for templates to support something like this (I know there are different ways to do this in D)?
>
>
> struct Tree1 {
>     static struct Node { int x; }
> }
> struct Tree2 {
>     static struct Node { int x, y; }
> }
> void foo(T)(T.Node n1, T.Node n2) {} // ***
> void main() {
>     Tree1.Node n1, n2;
>     foo(n1, n2);
> }
>
>
> (I am asking this because nested types are becoming increasingly common in D code.)
>
> Bye,
> bearophile

I can't say I've ever thought to have such. But it seems it could simplify something. Though I'd think the syntax:

Void foo(T: T.Node) (...

Or more proper.