All submissions to this site are governed by Second Life Project Contribution Agreement. By submitting patches and other information using this site, you acknowledge that you have read, understood, and agreed to those terms.
VEHICLE_REFERENCE_FRAME appears to now be global rather than relative to the vehicle, which is to say if the reference frame "tilts" the vehicle backwards when the vehicle is level and facing east, it tilts the vehicle forward when it's level and facing west.
The repro is very easy, simply set the reference from to anything non-zero and turn the vehicle with an angular motor. This is likely a known bug, but I can't find anything on the jira.
Description
VEHICLE_REFERENCE_FRAME appears to now be global rather than relative to the vehicle, which is to say if the reference frame "tilts" the vehicle backwards when the vehicle is level and facing east, it tilts the vehicle forward when it's level and facing west.
The repro is very easy, simply set the reference from to anything non-zero and turn the vehicle with an angular motor. This is likely a known bug, but I can't find anything on the jira.
Yes, there were some new parts of the vehicle code (for making Havok4 ground vehicles behave more like Havok1 ground vehicles) that was not using the VEHICLE_REFERENCE_FRAME feature at all yet – they now do.
Also, there was a bug where some old Havok::Quaternion math was converted to use LLQuaternions, but the developer who did it did not know that the LL quaternions operate from the other direction. That is hkQ_AC = hkQ_BC * hkQ_AB but LLQ_AC = LLQ_AB * LLQ_BC, so a quaternion transformation was in the wrong order.
Finally, I discovered that the VEHICLE_ANGULAR_FRICTION feature has never been using the VEHICLE_REFERECE_FRAME properly Most vehicles don't use this feature very much, and the problem would only show up well if the eigen values of the vehicle's inertia tensor were very divergent – which is probably why no one ever complained; the percentage of vehcles affected by this problem is probably a tiny fraction of 1%.
Andrew Linden added a comment - 19/Oct/07 09:19 AM Yes, there were some new parts of the vehicle code (for making Havok4 ground vehicles behave more like Havok1 ground vehicles) that was not using the VEHICLE_REFERENCE_FRAME feature at all yet – they now do.
Also, there was a bug where some old Havok::Quaternion math was converted to use LLQuaternions, but the developer who did it did not know that the LL quaternions operate from the other direction. That is hkQ_AC = hkQ_BC * hkQ_AB but LLQ_AC = LLQ_AB * LLQ_BC, so a quaternion transformation was in the wrong order.
Finally, I discovered that the VEHICLE_ANGULAR_FRICTION feature has never been using the VEHICLE_REFERECE_FRAME properly Most vehicles don't use this feature very much, and the problem would only show up well if the eigen values of the vehicle's inertia tensor were very divergent – which is probably why no one ever complained; the percentage of vehcles affected by this problem is probably a tiny fraction of 1%.
Torley Linden added a comment - 21/Dec/07 12:06 PM========-
Changed status to "Fixed" – I'm going through the Issue Tracker and updating issues that've been publicly fixed!
========-
Bug
Closed
Critical
VEHICLE_REFERENCE_FRAME broken on preview
Also, there was a bug where some old Havok::Quaternion math was converted to use LLQuaternions, but the developer who did it did not know that the LL quaternions operate from the other direction. That is hkQ_AC = hkQ_BC * hkQ_AB but LLQ_AC = LLQ_AB * LLQ_BC, so a quaternion transformation was in the wrong order.
Finally, I discovered that the VEHICLE_ANGULAR_FRICTION feature has never been using the VEHICLE_REFERECE_FRAME properly Most vehicles don't use this feature very much, and the problem would only show up well if the eigen values of the vehicle's inertia tensor were very divergent – which is probably why no one ever complained; the percentage of vehcles affected by this problem is probably a tiny fraction of 1%.