Type combinations

Is there any way to express an "intersection" type, analogous to a
union type? I can see there may be fewer use cases that there are for
union types, but they could be useful in the following cases:

1. An argument must implement two interfaces, but it's not practical
to introduce a 3rd interface to unify them as a single type.
2. An argument must be some subtype of a given type, but also
implement one or more additional interfaces.

A trivial example (using dubious syntax):

function replicateObject(obj:(DisplayObject & IClonable)):void
{
   var copy:DisplayObject = obj.clone();
   parent.addChild(copy);
}


I know that proposals are closed now, but this seems simpler than -
and complimentary to - union types.

Peter

On Nov 24, 2007 3:37 PM, Peter Hall <peter.hall at memorphic.com> wrote:
> Is there any way to express an "intersection" type, analogous to a
> union type?

None at present.  I toyed with similar things for a while, being able
to subtract types from unions and so on.

> I can see there may be fewer use cases that there are for
> union types, but they could be useful in the following cases:
>
> 1. An argument must implement two interfaces, but it's not practical
> to introduce a 3rd interface to unify them as a single type.
> 2. An argument must be some subtype of a given type, but also
> implement one or more additional interfaces.
>
> A trivial example (using dubious syntax):
>
> function replicateObject(obj:(DisplayObject & IClonable)):void
> {
>    var copy:DisplayObject = obj.clone();
>    parent.addChild(copy);
> }
>
>
> I know that proposals are closed now, but this seems simpler than -
> and complimentary to - union types.

I think TG1 should look into various kinds of type constraints for the
next iteration of the spec -- this would be interesting for the
parameterized types, also.  For now, proposals really need to be
closed if we're going to finish the spec within a reasonable time.  I
will write up this as a ticket in Trac and then immediately resolve it
as "Deferred"; the intent is that we will produce a document with
deferred ideas by and by.  There are lots of them.

--lars

> I think TG1 should look into various kinds of type constraints for the
> next iteration of the spec -- this would be interesting for the
> parameterized types, also.  For now, proposals really need to be
> closed if we're going to finish the spec within a reasonable time.  I
> will write up this as a ticket in Trac and then immediately resolve it
> as "Deferred"; the intent is that we will produce a document with
> deferred ideas by and by.  There are lots of them.
>
> --lars
>

Thanks.

I think the important thing is that the syntax for union types is
designed with more complex set operations in mind. Otherwise we could
end up with some really ugly syntax if they get added in a later
revision.

Peter