Algorithms and slices

[Timon Gehr]

On 08/26/2011 08:24 PM, Mafi wrote:
> The algorithms in std.algorithm are great. They operate the same on
> arrays/slices and generic ranges. But they return their own range-types.
> Often this ok but sometimes you need a T[]. You may say to just use
> array() but this allocates a new array! I think you sometimes want to
> get a slice to the original array. I wrote this function. It checks if
> it is ok what it is doing at runtime.
>
> ////////////
> import std.stdio, std.algorithm, std.range, std.conv, std.exception;
>
> void main() {
> auto x = [0, 1, 2, 3, 4, 5];
> auto t = x.take(3);
> foreach(i, ref elem; t) {
> writefln("%s : %s (%s)", i, elem, &elem);
> }
> foreach(i, ref elem; array(t)) {
> writefln("%s : %s (%s)", i, elem, &elem);
> }
> foreach(i, ref elem; slice(t)) {
> writefln("%s : %s (%s)", i, elem, &elem);
> }
> }
>
> /**
> Retuns a slices of the data the given range represents.
>
> throws: ConvException if the data is not continous.
> *
> ElementType!(R)[] slice(R)(R range) if(isInputRange!R) {
> alias ElementType!(R) E;
> E* addr = null;
> size_t length = 0;
> foreach(i, ref e; range) {
> static assert(is(typeof(e) == E));
>
> if(addr is null) {
> addr = &e;
> } else {
> enforce(&e is (addr + i), new ConvException("Could not create a
> continous slice."));
> }
>
> length++;
> assert(i == length-1);
> }
> if (addr is null) return null;
> else return addr[0.. length];
> }
> //////////////
>
> Is there functinality like this in phobos? Does such a function fit into
> the range world.

That exposes the implementation details of all range-based functions. 
Are there even any range types in Phobos that do save their elements in 
a continuous array?