Refactor lockstep and simplify opApply string mixin generator.

[ghost]

No description provided.

[ghost]

Unfortunately the diff view doesn't give it justice, not even with ?w=0.

[denis-sh]

First of all, we don't need std.range.lockstep at all. What we really need is to document ability to expand tuple in foreach (implemented by Kenji of course) and add such example to std.range.zip.

[ghost]

First of all, we don't need std.range.lockstep at all.

We'll need at least a backwards-compatible template or function which forwards to zip, to avoid breaking code.

[denis-sh]

You are right of course, I mean as a permanent documented member.

[monarchdodra]

We'll need at least a backwards-compatible template or function which forwards to zip, to avoid breaking code.

AFAIK, zip's front is return by value, so you we use zip in a foreach with ref parameters. You can with lockstep:

import std.range;
import std.stdio;

void main()
{
    int i = 0;
    int[] odds = new int [](4);
    int[] evens  = new int [](4);
    foreach(ref a, ref b; lockstep(odds, evens)) //Works
    //foreach(ref a, ref b; zip(odds, evens)) //Fails
    {
        a = ++i;
        b = ++i;
    }
    assert(odds == [1, 3, 5, 7]);
    assert(evens == [2, 4, 6, 8]);
}

So we can't currently outright make lockstep an alias of zip.