This is something I'd love to see implemented, but, I don't know if we should change the `array_*` functions. By they starting with "array_", we presume that one of the arguments is gonna be an array, not an iterable.

I can try to create POC within the next weeks and open for discussion on internals@ :)

Actually, these functions will still return an array as a result :) Also, I can't find any suitable name for such functions. `iterable_diff()`? Looks ugly. And array will still be the valid input.

Some comments that will likely arise:
* The single-pass functions (eg, map, column, chunk, combine) could benefit from this
* The more complicated functions (eg, diff) would need iterator_to_array
* iterator_apply would presumably be made an alias of array_map
* These functions can be called multiple times on arrays but not necessarily so on iterators
* What happens when the iterator is exhausted?

That last point is the main reason why this question tends to be resolved one way: using iterator_to_array is a benefit, not a hindrance, as it reminds the developer that the source was not a simple array that can be easily reused.

Personally, I think it is better to provide a namespace like `STD` and then, in that namespace, implement functions polyfilled for array. iterator and all possible foldable interfaces.

For that library, we can use an PHP implementation as the first release, and then make it to C and merge it to main PHP branch when the API seems stable, For a long time, we are in a lack of standard library with consistency.

Hi guys, is there any update on this proposal? It sounds reasonable, doesn't it?

Thanks!