HingeConstraint / Point2PointConstriant

Topics: Developer Forum, User Forum
Developer
Apr 5, 2007 at 1:35 PM
Edited Apr 5, 2007 at 1:44 PM
Does anybody know if these constraints (HingeConstraint / Point2PointConstriant) should work well in the current version of the managed bullet engine? I don't get them to run well, they seem to be unstable (e.g. object are jumping ariund like crazy). Maybe I missconfigured something and need to tweak even more, if someone else made a working example. (lil notice: unconstrained rigid body motion works well!! Nice work!). But if the developers know already some issues, I will not waste any more time trying to get it work without manipulating the c# bullet source code.

IF THERE ARE KNOWN ISSUES: Can give some of the developers give me a date, until when it will be fixed maybe? Are you actively working on that already? Or give me a starting point of the so far known details and I will fix it and submit a patch to solve the problems. Because we would like to use the managed bullet engine and XPA for a XBOX360 game project that is just about to start and ends in 12 weeks. So, when needed, we will also spend some time in bugfixing the bullet c# port, while finishing our game...

cheers, b3n
Coordinator
Apr 5, 2007 at 6:12 PM
Hi,
I'm right now working on the constraints, yes they are jumping around like crazy :)
Another basic shapes looks to work well. And I'm planning to get mesh collision to work after april XNA update.
Those are the main issues right now: constraints and meshes.

Regards,
KleMiX
Developer
Apr 9, 2007 at 3:25 PM
Edited Apr 9, 2007 at 3:26 PM
Thanx for your answer, KleMiX.

But im not sure about what exactly u told me about the constraints...

"Another basic shapes looks to work well."
-> What basic shapes do already work? U mean with constraints or generally as rigid body motion? I tied out constraints with boxes and spheres and they didnt work... almost at all. So i did nothing with meshes so far and I think meshes are not the most important part to have ported as soon as possible, because u can model a lot with the basic shapes already. But I guess, constraints in the opposite are very important to have!

You told me about when mesh collision will be ready, thats nice. But I was asking for constraints, because I really would need them asap, if we want to use bullet already for our project. (Would save us a lot of work, isntead of implementing our own XNA phyiscs engine) I would also invest work into ur Project to get the constraints working. You just need to give me a short description about ur exact state (what classe u are currently changing, in what direction to do something and so on.)
Coordinator
Apr 10, 2007 at 6:29 PM

b333n wrote:
U mean with constraints or generally as rigid body motion?
In general :)
I can agree on constrint importance that's why it's the primary goal now :)
Now I again have less time to work on project... so I can't tell exact times... If you're interested in fixing it I can add you as a developer and check-in my testbed with constraints which is the same as C++ appConstraintDemo for debuging ease. There's some problems with calculating - in C++ version I get difference in 0.1 to 0.01 but in C# I get exactly 0 - that's can be the point... I can also try to get some guesses from Erwin (Bullet author)...
Developer
Apr 10, 2007 at 7:34 PM

Now I again have less time to work on project... so I can't tell exact times... If you're interested in fixing it I can add you as a developer and check-in my testbed with constraints...


That would be very cool, if you would add me and check-in that testbed. Because im really planning to fix the problem and use your bullet port in our project. I now need to evaluate if its possible to use this contraints for our project during this or at least during the next week,... So I think I'll have some time to spend in fixing the issue,... hope I'll get it.
Coordinator
Apr 11, 2007 at 6:08 PM
OK, I checked in code and added you as a developer. Good luck :)
Coordinator
May 2, 2007 at 7:28 PM
Today I fixed this issue and I will check-in it in some time
Developer
May 3, 2007 at 11:40 AM
Oww, well! What was the problem? Could you tell me, if it was just a small thing? Something else: Will you ever update the bullet-port to the actual bullet C++ version? Because some parts of the constraints solver are quite old already, as i noticed compared to the actual version of c++ bullet...
The reason I didnt fix the code was cauze I had some personal problems, had an accident and got unfortunately a cranial fracture. Therfore was a week in hospital and stuff like that... now I feel a bit better and could continue to develop... but you seem to be faster ;-) Thanx anyway!
Coordinator
May 4, 2007 at 11:35 AM
The problem was in the Vector3.Transform function, Bullet used the same but with matrix rows and cols swapped so it's a small thing.
Yes, I'll update source to more recent version in the near future.
If you want to know what exactly changed you can compare existing files with older versions, for example Point2PointConstraint.cs with it's older version.