Mikez2015

**ccdVec3PointTriDist2** The function incorrectly considers the distance from the zero point to the triangle with the following coordinates: A(1.650970, -0.102323, -2.994949); B(-1.613953, -0.102323, 2.534958); C(-1.613953, -0.102323, -2.994949); The function returns...

> One of the factors that is important, from a geometry sense, is the scale of the problem. The default epsilon value works best when all of the parameters are...

I began to research this comparison on the correctness of work: https://github.com/danfis/libccd/blob/master/src/vec3.c#L176-L180 I found out that if the function does not work properly, the condition is not met `(ccdIsZero(t) ||...

> For the "erroneous" triangle, you assume the correct distance is 0.102323 which implies the origin projects to a point _inside_ the second triangle. However, that appears not to be...

> I was worried that might be a bit opaque. Sorry about that. I'm happy to walk through it. > > > I correctly understood that you are claiming that...

> Correct. Everything else in the message was about showing that to be the case. I double-checked, you are absolutely right. Then it turns out that the issue is that...

> Are you sure about that? It's reporting the correct distance. It seems highly unlikely that it can simultaneously classify it as projecting onto the triangle _and_ compute the correct...

> Are you sure about that? It's reporting the correct distance. It seems highly unlikely that it can simultaneously classify it as projecting onto the triangle _and_ compute the correct...

> * Are you invoking `ccdVec3PointTriDist2` with or without witnesses? > > * What is your type of `ccd_real_t`? Double or float? With witnesses... probably witness is not equal to...

> Are you sure about that? It's reporting the correct distance. It seems highly unlikely that it can simultaneously classify it as projecting onto the triangle _and_ compute the correct...