Description

Fixes #4627

Fixes a <= that should be a <. Also removes some redundant > that could create similar issues.

More detail: After projections and truncating to Float32, p[5], p[6] both have an x value of 1, i.e. they are exactly on the clipping edge. (In clip space the x value range is always -1..1.) Clipping is checked using the distance to clip planes, effectively d = 1 - x, so for our points d = 0f0.

For the p[4] -- p[5] segment it does not detect them as both outside with d <= 0 (correct), and then also not as one outside with d < 0 (fine). In other words it detects both as inside. Neither point needs to be adjusted, and only the first is pushed. If p[5] were outside, p[5] would get translated towards p[4] until it is inside and then get pushed to finish the segment.

For the next segment p[5] -- p[6] it does detect them both as outside with d <= 0, which is an invalid state. Both points being outside implies that p[4] -- p[5] detected p[5] as outside, which would have pushed an adjusted point between the two. So this iteration believes the previous segment to be finished and pushes NaN to cut the line instead. And thus the inside -> edge segment disappears.

Type of change

• Bug fix (non-breaking change which fixes an issue)

Checklist

• Added an entry in CHANGELOG.md (for new features and breaking changes)
• Added or changed relevant sections in the documentation
• Added unit tests for new algorithms, conversion methods, etc.
• Added reference image tests for new plotting functions, recipes, visual options, etc.