On Sun, Mar 28, 2010 at 3:47 AM, Kai Sterker <kai.ster...@gmail.com> wrote:
> On Sun, Mar 28, 2010 at 7:22 AM, Tyler Nielsen <tyler.niel...@gmail.com> > wrote: > > > I'm working on writing a few unit tests for placeable so that when I make > > the change, I can be reasonable sure it's correct. One question though, > I > > believe EntireMinPos should represent the Bottom, Left, Forward point, > and > > EntireMaxSize is how far it extends in each direction from there. Is > that > > correct? > > Yes, that's correct. > > > A specific example is, a cube3 of: min = (0, 0, -5) max = (40, 40, 0) > (and > > the points that match that). Should have a EntireMinPos of (0, 0, -5) > and a > > EntireMaxSize of (40, 40, 5). > > Exactly. > Kai, Let me start off by saying, enjoy your easter. if you can't get to this for a few days, no big deal. Ok I know this e-mail is from a while ago, but this I found the problem with jumping and it relates directly to this. Basically we have a position of the tile you stand on when starting of: [256, 480, -5] - [384, 624, 0]. When we check for a collision (in collide_with_objects) we call object_in_bbox, which takes the position and adds SolidMinPos([0,0,-5] in this case) and doesn't find the collision until we fall too far. I sort of found this out already (but didn't realize the significance) when I modified the line: shape->collide (collisionData, (*i)->Min-(*i)->get_object()->entire_min()); In doing that I 'normalized' the position on the objects (0,0,0) so when it adds shape later it is correct. Basically there are two ways of fixing this. One option is to change the definition of EntireMinPos and SolidMinPos, so they are always relative to the lowest values. (In the case above, SolidMinPos would then be 0,0,0). But I'm afraid that will have far reaching effects that I can't anticipate. Two, I can subtract out entire_min when we are getting the solid_min and solid_max from chunk_info. We are already doing it in a few other places, so maybe I can clean the whole thing up a bit as I go. I just wanted to run the whole thing by you and get your take before I go too far down this path. Tyler
_______________________________________________ Adonthell-devel mailing list Adonthell-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/adonthell-devel