smiths
commented
5 years ago @oluowoj, for your T5 (Collision Detection) we want to actually have an equation. Something like the following:
In the description you can clarify that we have a sequence of bodies B. Each body is made up of a set of points in space. The set of points changes as the body moves. Two bodies have collided if they have intersection of their sets of points.
This will only bring us part way to what we eventually need. Knowing there is a collision is only part of the story. What we really need is to return the collision point or points. We should have an invariant that two bodies never actually intersect, so in our 2D examples we shouldn't have an area of intersection. Intersections can either be at a point, or on a line. We really should return the points and lines that intersect.
This discussion highlights that our SRS is missing more information. What is a point? a line? a body? These things are implicitly assumed, but not made explicit. At some point we need to have another serious pass over this document, but I don't think now is the time for this. We should discuss this with @JacquesCarette when we are planning for your tasks going forward. (We also need to discuss whether we have a sequence of bodies, or the more general set of bodies. The literature on game physics always talks about a sequence, as far as I can tell, but this seems like a programming idea, not an abstract view of the problem.) My instinct is that we don't want to get too distracted at the moment and that we are better off proceeding with the bigger picture stuff.
My advice is to fix T5 along the lines I have explained above. We can worry about the other things I mentioned later. It will also become easier to write the SRS when we introduce the notion of types to our specifications. (It would be nice to have a type for points, lines, etc.)
Your IM4 (Collision detection of rectangles) also needs some work. Can you please make the following changes:
- add a reference for the axis aligned bounding box method
- clarify your assumption that rectangular bounding boxes are adequate (in general the boundaries are convex polyhedra)
- unbold x_1, x_2, etc - these aren't vectors
- you need to re-state your output. It can't just be True (or False). We need to know what two bodies have collided. Don't you also need the point of collision?
IM4 also highlights that we really will eventually need the notion of a body. I know what you mean by rect_1.x1, but this isn't really defined anywhere.
oluowoj
@smith, as discussed SRS is located @ https://github.com/smiths/caseStudies/tree/gamephy_finaldoc/CaseStudies/gamephys/docs/SRS.
Please let me know if it makes any sense to you before I proceed with adding it on Drasil.