Vector subtraction in function 'rod::point_lies_on_rod_line_element()' gives inconsistent results

[Ryan Cocking]

The vector subtraction:

vec3d(n){vec3d(n){r1_c[n] = c[n] - r1[n];}

will inconsistently give strange results, such as [1,0,0]-[0,0,0] = [1.07,0,0], usually appearing to add something on when attempting to subtract.

This only seems to happen when the function is called from a larger function, rod::get_point_on_connecting_line(), during the test ffea_test::line_connecting_rod_elements(). Are the arguments being passed in wrong?

The independent unit test of rod::point_lies_on_rod_line_element() does not show such behaviour.

This bug could interfere with assigning point of connection during rod collision.

The attached output shows an example (line 47). There is an error of -0.0669873 in x, and -0.25 in z.

‌

[Ryan Cocking]

Printing c - r1 per-element seems to solve this problem. See second attached text file.

[Ryan Cocking]

However, replacing the vec3d statement with individual subtractions replicates the error!!

[Ryan Cocking]

Turns out, doing the vector subtraction before vs. after the if-statements that assign c to a rod node or not, gives different outcomes! (subtracting before gives the buggy behaviour)

[Ryan Cocking]

Did some fiddling around and it seems to have solved itself. I have a strong suspicion that I was using the wrong values of r1 and c to check if the answer to the function was correct. No apparent 'error' in the calculation now occurs! Hooray for silly mistakes!!