I think the previous structure with TestCase was much better. My suggestion with the swapped scenarios was meant to be handled in the TestCase definition, not as a separate test function.

Thanks for the comment.
Could you tell me why using TestCode is better than using TestCodeSource for this test?

I find the whole new object[] a bit overkill, that's it. I guess it's more a matter of personal taste.

Thanks for the comment.

it's more a matter of personal taste.

Could you show the reason?

The reason for what exactly? I prefer compile-time checks and try to avoid object as much as possible

Thanks for telling the reasons.

Done in the commit cd80cf1.
I changed the test-code as you required. I'm not sure that the revised test-code helps the reader understand at a glance that parameters are kept intact except its order.

Thank you. This is not what I meant, we can augment the existing tests with a reversed line

var line = new Line2D(Point2D.Parse(p1s), Point2D.Parse(p2s));
// existing test

var reversedLine = new Line2D(Point2D.Parse(p2s), Point2D.Parse(p1s));
// distance test with reversed line

This way you keep the TestCase cases at a minimum

Thanks for showing the example closely.

It is not recommended to test 2 things in a single unit-test.
Or do you want to say that these 2 tests are equivalent because of the symmetry?

It is not recommended to test 2 things in a single unit-test.

Fully agree

Or do you want to say that these 2 tests are equivalent because of the symmetry?

Basically yes