Bugfixes + Improvements (RWR, Explosion Modelling, HMD)

[BillNyeTheIE]

• Fix TGP NRE
• Fix robotic parts with turretID = 0 getting sent back to original position by AI
• Massive RWR overhaul
• Improved explosion modelling for both standard and shaped charge explosions, particularly with regards to armor, as well as very large explosions
• More granular control over notch chaff effects
• Improve uncagedLock behavior and new HMD
• Various minor bug fixes
• Update KKV to use BDCustomWarhead
• Update AIM-9 to use uncagedLock as well as improve cfg documentation

[BrettRyland]

Note to self: sync this to the other localisation files once merged.

[BillNyeTheIE]

Actually, I was thinking about this, but perhaps it'd be better to sync these to the other localization files as just the English text, rather than what it was previously translated to, since otherwise translators might miss the fact that this was changed, and that the meaning is different.

[BrettRyland]

There's a lot of duplicate code here setting ml.customTurret[i].slavedGuard to true or false. I'd suggest putting it into a single function, e.g.,

public void SlaveCustomTurrets(bool slaved)
{
    for (int i = 0; i < customTurret.Count; i++)
    {
        if (customTurret[i] == null) continue;
        if (customTurret[i].vessel != vessel) continue;
        customTurret[i].slavedGuard = slaved;
    }
}

in MissileBase.cs and just use ml.SlaveCustomTurrets(true); here to make this easier to maintain in the future.

Similarly, the while loop for aiming the slaved turrets (both custom and MissileTurrets) could likewise be moved into common coroutines for ease of maintenance, which would reduce these blocks to something like

ml.SlaveCustomTurrets(true);
yield return ml.AimCustomTurretToTarget(targetVessel, Mathf.Max(targetScanInterval / 2f, 2));
ml.SlaveCustomTurrets(false);

The slaving calls could even be moved inside the coroutine if those are always required, making these all into one-liners where MissileBase.cs (and MissileTurret.cs for regular turrets) contains something like

public IEnumerator AimCustomTurretToTarget(Vessel target, float timeout)
{
    SlaveCustomTurrets(true);
    var wait = new WaitForFixedUpdate();
    float angle = 999;
    float turretStartTime = Time.time;
    while (Time.time - turretStartTime < timeout && targetVessel && angle > 5)//customTurret[0].fireFOV
    {
        for (int i = 0; i < customTurret.Count; i++)
        {
            if (customTurret[i] == null) continue;
            if (customTurret[i].vessel != vessel) continue;
            angle = VectorUtils.Angle(MissileReferenceTransform.forward, customTurret[i].slavedTargetPosition - MissileReferenceTransform.position);
            customTurret[i].slavedTargetPosition = MissileGuidance.GetAirToAirFireSolution(this, target.CoM, target.Velocity(),
                customTurretLoft, customTurretLoftFac);
            customTurret[i].AimToTarget(customTurret[i].slavedTargetPosition);
        }
        yield return wait;
        //Debug.Log($"[PD Missile Debug - {vessel.GetName()}] bringing radarMsl custom turret to bear, angle {angle:F2}...");
    }
    SlaveCustomTurrets(false);
}

and a similar function for taking in TargetSignatureData instead of Vessel.

I see some of the turret aiming code may be doing other stuff that would make this more complicated, but I think it should generally be possible to condense most of these large blocks of code to common functions that are easier to maintain.

[BillNyeTheIE]

Hrrrm I'm trying to implement this but it'll need some kind of method to tell GMR it has completed before GMR can go about trying to fire the missile. I was thinking of using a boolean field that it can write to which I can then read from GMR, but it's getting a bit convoluted in terms of predicate inputs for the different conditionals and if we ever want to allow simultaneous launches a single field or even an array of fields is gonna be a bit messy to deal with. I think it might be wiser to just keep the while loops on the outside.

I have packaged up the turret aiming code for all missiles into a single function in MissileBase though (with an override in MissileLauncher to account for the additional turret types they can be mounted on), so it is still more maintainable, with only a single loop required.

[BrettRyland]

This could be written more succinctly as

groundStabilized &= gimbalLimitReached || slewingToPosition;

[BrettRyland]

These could be combined to the following without losing any performance:

MethodImpl(MethodImplOptions.AggressiveInlining)]
public static float Angle(Vector3d from, Vector3d to)
{
    double num = from.sqrMagnitude * to.sqrMagnitude;
    if (num < 1e-30)
    {
        return 0f;
    }

    double num2 = BDAMath.Clamp(Vector3d.Dot(from, to) / Math.Sqrt(num), -1.0, 1.0);
    return (float)(Math.Acos(num2) * 57.295779513082325);
}
[MethodImpl(MethodImplOptions.AggressiveInlining)] public static float Angle(Vector3 from, Vector3 to) => Angle((Vector3d)from, (Vector3d)to);
[MethodImpl(MethodImplOptions.AggressiveInlining)] public static float Angle(Vector3 from, Vector3d to) => Angle((Vector3d)from, to);
[MethodImpl(MethodImplOptions.AggressiveInlining)] public static float Angle(Vector3d from, Vector3 to) => Angle(from, (Vector3d)to);

[BillNyeTheIE]

I mentioned this in my commit, but I was thinking, given that Vector3 is automatically (as far as I'm aware) typecast to Vector3d when passed into a Vector3d function, would these lambda expressions even be needed? Does it change how the typecast functions?