Wrong position for transformation about playrho HOT 5 CLOSED

I'm fairly certain that changing it to pos1 would be correct, because my game state gets desynchronized after serialization and deserialization if I just use the current pos0 in the function.

from playrho.

I've run into this problem before too. But is this more of a wrong position, or more of a it-doesn't-work-as-expected problem?? I'm wondering this myself.

Even when I recently added the ability to set the sweep, I ran into this. Also, I ran into the gotcha of having to set resetMassData to false.

from playrho.

But is this more of a wrong position, or more of a it-doesn't-work-as-expected problem??

I'm not sure either. I really have to emphasise that when I filed these issues I was almost completely blind to how things logically should work; I was solely looking at how I can recreate an exact copy of a serialized world, and I just found I had to change this function getting pos0 to pos1 for the equality to be true.

from playrho.

Some notes on this, at least for myself:

1. In commit a709055, playrho/d2/BodyConf.hpp was changed to add support for configuring with Sweep information, instead of being limited just to location and angle (that form a Position that would set Sweep::pos0 and Sweep::pos1 to that value).
2. This meant that then one variable (sweep), spoke for what used to be two separate variables (location, and angle), and could then also convey state independently for the other stage of the sweep (so sweep.pos0 and sweep.pos1 can now both be set to what that had been - and not just to the same Position).
3. This also meant that Body::m_sweep could then be initialized by BodyConf::sweep, instead of having Body::m_sweep initialized to the more limited Sweep{Position{bd.location, bd.angle}}.
4. The question at hand seems to be whether the appropriate semantics of the location, angle and the Transformation of the former BodyConf have been preserved for the new BodyConf while allowing the other Position of the Sweep to be set independently.
5. The address of this question seems to be predicated on whether the changes to BodyConf match the behavior for Body with respect to functions like Body::GetTransformation, which returns Body::m_xf, which itself is set from GetTransform1 of the body's sweep, which in turn is the value of GetTransformation(sweep.pos1, sweep.localCenter).
6. In other words, I'm seeing this as a relational consistency question.

So would it be more consistent for GetTransformation(const BodyConf& conf) to return GetTransform1(conf.sweep) (which is GetTransformation(conf.sweep.pos1, conf.sweep.localCenter)) than {conf.sweep.pos1.linear, UnitVec::Get(conf.sweep.pos1.angular)}? I believe so.

I'm gonna try this and see what unit testing reveals.

from playrho.

I've merged in changes now to the master branch to address this. Closing this assuming that it does. Please let me know though either way. Thanks!

from playrho.