UNT0034 and UNT0035 need more explaining?

[MightyCreak]

Hi!

I wonder about the relevancy of the UNT0034 and UNT0035 as to why they are in the "Performance" category.

As far as I can understand, in the example given in UNT0034, it's advised to go from there:

        Vector3 v3 = ...;
        Vector2 v2 = ...;
        var distance = Vector2.Distance(v2, new Vector2(v3.x, v3.y));

to there:

        Vector3 v3 = ...;
        Vector2 v2 = ...;
        var distance = Vector2.Distance(v2, v3);

But, if I'm not mistaken, the only thing it does is calling the implicit cast here: https://github.com/Unity-Technologies/UnityCsReference/blob/master/Runtime/Export/Math/Vector2.cs#L387

public struct Vector2 : IEquatable<Vector2>, IFormattable
{
    public static implicit operator Vector2(Vector3 v)
    {
        return new Vector2(v.x, v.y);
    }
}

And, to me, it feels like the result is exactly the same: instead of doing a new explicitly, it does a new implicitly.

And, as a result, I think that mixing Vector2 and Vector3 makes the code less readable and potentially less maintainable. (it's obvious in this example where variables are named v2 and v3, but it's a bit different when you have a bunch of variables in your algorithm).

So my question is: am I missing something that would explain why there is a performance improvement? (if I am, could you explain it to me? I'm really interested to understand why); or should these diagnostics be re-categorized? to "Correctness", maybe?

[jbevain]

I think those diagnostic ended up in that category simply for the lack of a better worded category. It's not related to Performance or Correctness. Maybe Readability?

But yeah there should be no performance difference if the implicit conversion is inlined.