On Thursday, October 11, 2012 18:45:36 Rizo Isrof wrote:
Hi,
Why the `in` expression can only be applied to associative
arrays and cannot be used with static or dynamic arrays as
it is possible with, _e.g._, Python?
The following code is not legal:
int[] a = [1,2,3,4,5];
if (1 in a) { }
Are there any technical explanation for this limitation?
Because that would mean than in was O(n), whereas it's generally assumed to be
at least o(log n) (which is what you'd get in a balanced binary tree such as
red-black tree). AA's do it in O(1), so they're okay, but dynamic arrays can't
do better than O(n). And if you can't rely on a certain algorithmic complexity
for an operation, then it becomes problematic for generic code. std.container
specifically lists the complexity requirements for every function in there
which is standard across containers, and any container which can't define such
a function in the required complexity, doesn't define that function.
There has been some discussion of allowing in on dynamic or static arrays in
some restricted cases where the array is known at compile time, but nothing
has come of it as of yet. Regardless, there's pretty much no way that arrays
in general will ever work with in, because it would be too inefficent.
You can use std.algorithm.find or std.algorithm.canFind. e.g.
int[] a = [1, 2, 3, 4, 5];
if(canFind(a, 1))
{
...
}
- Jonathan M Davis
