Make Gizmos::rect_2d draw the joint of the upper left corner of the rectangle

[uckelman]

Make Gizmos::rect_2d draw the joint of the upper left corner of the rectangle

Objective

• Gizmos::rect_2d does not draw the line joint between the left and top line segments. This looks odd when the line joint is not GizmoLineJoint::None.

Solution

• Gizmos::rect_2d draws the rectangle using Gizmos::linestrip_2d, which doesn't draw a closed shape.
• In order to make linestrip_2d to draw a joint at a corner, that corner has to be in the points list with a predecessor and a successor.
• tl has no predecessor when it's the first element in the list, and no successor when it's the last.
• Appending tr to the list is enough to put tl between two other points, which causes the joint to be rendered.

Testing

• I tested my change against a demo where I draw a selection box.

The test code is just this, and observing the result captured in the screenshots below:

    gizmos.rect_2d(selection.rect.center(), selection.rect.size(), Color::srgb_u8(0xFF, 0, 0));

---

Showcase

Incorrect GizmoLineJoint::Miter joint without this change:

Correct GizmoLineJoint::Miter joint with this change:

[github-actions]

Welcome, new contributor!

Please make sure you've read our contributing guide and we look forward to reviewing your pull request shortly ✨

[alice-i-cecile]

I don't love that we're double-drawing the one line segment, but this is a very unobtrusive fix. Are other polygons affected?

[uckelman]

I don't love the double drawing either, but I didn't spot any underlying mechanism for drawing closed shapes. Not sure about other polygons---I found this in the context of drawing a selection box. I'll check..

[uckelman]

rounded_rect_2d has a correct upper left corner, but I think doesn't use line joints? This also seems not to be a problem with ellipse_2d. I don't see any general polygon drawing functions in Gizmos.

[uckelman]

Aha, I found primitive_2d for drawing polygons. Hadn't spotted that before. primitive_2d does have the same problem with the joint on the first vertex. E.g.:

These all use linestrip_2d in the same way, so should all have the same defect:

GizmoPrimitive2d<Rhombus>
GizmoPrimitive2d<Triangle2d>
GizmoPrimitive2d<Rectangle>
GizmoPrimitive2d<Polygon>
GizmoPrimitive2d<RegularPolygon>

Given that, a better fix would be a lineloop_2d (or some other better name?) they could all call.

[alice-i-cecile]

Given that, a better fix would be a lineloop_2d (or some other better name?) they could all call.

Agreed, this would be my preference. We could fix #22081 in earnest, and then resolve all of these problems at once.

[uckelman]

Closing this in favor of #22085.