segDirection should be normalize ?it looks like a bug.

[zyxia]

https://github.com/davideberly/GeometricTools/blob/4a8f9d6803ed2210cfe8bb47343d3bb468b7dd34/GTL/Mathematics/Distance/3D/DistSegment3Triangle3.h#L42C55-L42C55

[davideberly]

I believe the code works correctly. The GTL file DistLine3Triangle3.h has a comment on line 12

// The line is P + t * D, where D is not required to be unit length.

The following sample code shows that it does not matter that the segment direction is unnormalized.

    Triangle3<double> triangle{};
    triangle.v[0] = { 1.0, 0.0, 0.0 };
    triangle.v[1] = { 0.0, 2.0, 0.0 };
    triangle.v[2] = { 0.0, 0.0, 3.0 };

    // The segment has endpoints {4,5,-1} and {6,7,1}. The non-unit-length
    // direction vector is the difference {2,2,2}.
    Line3<double> line{};
    line.origin = { 4.0, 5.0, -1.0 };
    line.direction = { 2.0, 2.0, 2.0 };

    DCPQuery<double, Line3<double>, Triangle3<double>> query{};
    auto result = query(line, triangle);
    // result.distance = 3.7416573867739413
    // result.sqrDistance = 14.0
    // result.parameter = -1.0
    // result.barycentric[0] = 0.0
    // result.barycentric[1] = 1.0
    // result.barycentric[2] = 0.0
    // result.closest[0] = { 2.0, 3.0, -3.0 }
    // result.closest[1] = { 0.0, 2.0, 0.0 }

    Normalize(line.direction);
    result = query(line, triangle);
    // result.distance = 3.7416573867739413
    // result.sqrDistance = 14.0
    // result.parameter = -3.4641016151377539
    // result.barycentric[0] = 0.0
    // result.barycentric[1] = 1.0
    // result.barycentric[2] = 0.0
    // result.closest[0] = { 2.0, 3.0, -3.0 }
    // result.closest[1] = { 0.0, 2.0, 0.0 }