-----
void main()
{
    int[] source = [0, 1];

    int[] arr1 = new int[](2);
    arr1[] = source[];

    assert(arr1 !is source);  // ok

    int[] arr2 = source[];  // looks like a copy
    assert(arr2 !is arr2);  // assert fails: but in fact it's not!
}
-----

The 'int[] arr2 = source[];' syntax appears as though the source contents are copied into arr2, but in fact this is the same code as 'int[] arr2 = source;'.

Since [] has a very special meaning, the above should either:

1) Become an actual deep copy, meaning arr2 would have to allocate space first and then copy contents. This would be a breaking and negative change due to performance implications.

2) Not compile. It looks like a deep copy but it isn't, and this can cause issues down the road (for example using memcpy or even OpenGL functions can create hard to track problems due to using the source and target arrays which point to the same memory).

I'd vote heavily towards 2.

Of course one can always use 'int[] arr2 = source.dup'.