Adding multiple paths into a compound path

Question

If I use the gui to add multiple paths into a compound path all I have to do is select multiple paths and choose Object->Compound Path->Make (or ctrl-8). When I do this then the bottom path which is filled has holes 'punched through' creating a window in the shape of the smaller paths which were on top of the larger one. For example, a large rectangle with a white fill color is the bottom item and two smaller paths are on top. I create a compound path and whatever is below the new compound path shows through in the largest shape of the previous two smaller paths.
If I use JavaScript and create a new compoundPathItem and move 3 paths into it, with the largest being first, then I get a compound path but only the first of the smaller objects creates a transparent space.
Here is a simplified example:
    var grpMask = doc.groupItems['mask group'];
    var compound = grpMask.compoundPathItems.add();
    var pathobj = grpMask.pathItems['bigbox'];
    pathobj.move(compound,ElementPlacement.INSIDE);
    pathobj = grpMask.pathItems['PathLeft'];
    pathobj.move(compound,ElementPlacement.INSIDE);
    pathobj = grpMask.pathItems['PathRight'];
    pathobj.move(compound,ElementPlacement.INSIDE);
In the above example the 'bigbox' path has the left and right paths on top of it but only the left path creates a transparent space in the original box.
I also tried using PLACEATEND instead of INSIDE and grouping the two smaller paths together and moving the group to the compound path.
Merging path items together won't work - I want to keep the original path's closed and intact without drawing a line between them so I can't just add their pathpoints together.
Is move() the best way to add existing paths to a new compound path?
Thanks!

Answer 1

Im at work at the moment so only have access to CS2 where the second script works… The first script worked at home with CS5…
here were my test files… Before
and After…
As you can see with this I did not change the polarity of the last path as in this test case I knew it was the largest of the shapes not sure if that had any bearing on my results…?