• 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.
MAINTENANCE ANNOUNCEMENT - JIRA will undergo maintenance starting 1:00am PDT through 3:00am on Saturday 2010.03.20. Please do not enter issues during this time as the system maybe restarted.
Issue Details (XML | Word | Printable)

Key: SVC-1703
Type: Bug Bug
Status: Resolved Resolved
Resolution: Fixed
Priority: Critical Critical
Assignee: Unassigned
Reporter: Sabine Stonebender
Votes: 8
Watchers: 4
Operations

If you were logged in you would be able to see more operations.
2. Second Life Service - SVC

Physics driven roller coaster consistantly breaks in Havok 4

Created: 28/Jan/08 01:11 PM   Updated: 26/Feb/10 02:01 PM
Component/s: Physics
Affects Version/s: Havok4 Beta
Fix Version/s: None

Time Tracking:
Not Specified

File Attachments: None
Image Attachments:

1. bugreport-rollercoaster.png
(699 kB)
Environment:
CPU: AMD (Unknown model) (795 MHz)
Memory: 1023 MB
OS Version: Microsoft Windows XP Service Pack 2 (Build 2600)
Graphics Card Vendor: ATI Technologies Inc.
Graphics Card: RADEON XPRESS 200M Series SW TCL x86/MMX/3DNow!/SSE2
OpenGL Version: 1.4.5524 WinXP Release
LLMozLib Version: 1.1.0 (Mozilla GRE:1.8.0.12_0000000000)
Packets Lost: 17/8969 (0.2%)
Viewer Digest: 58f7490d-4744-f65f-8845-3f6b997152e6

Linden Lab Issue ID: DEV-11224


 Description  « Hide
I created a Physics driven roller coaster ride. It is essentially a car/vehicle which follows a prim rail guide using a tori mounted to the top of the car. This creation has been essentially stable since the summer of 2005 when it was made but has a repeatable bug in Havok 4 that breaks it as described below.

-Attached picture shows summarized views of the vehicle and track in operation and in the midst of breaking

The ride consists of 2 main parts: the rails , and the car.
All prims used in the construct are material type glass for least friction.

-Rails/track: Consist of a closed loop made from cylinder type prims with a single rail containing a rezzer loaded with the car.
All rails other than the rezzer are not scripted.
Rails are NON-physical and set up well away from sim borders.

-Car: 10 prim physics driven vehicle using a limited vehicle script that makes it a constant acceleration forward once started. Steering is accomplished by a tori mounted to the top of the car which encloses the rail and follows it based purely upon collisions as shown in the attached image.
The car is a temp rez object when rezzed and expires when the rider unseats from the car.

Order of Operations:
-The rider rezzes a new car by touching the touch active rezzer rail
-The cars pilot chair is a sit active sculpt prim that upon being touched,
--seats the rider
--engages Physics and forward force of 1600(viewable copy of drive script to be forwarded to Spike Linden for full params)
--link messages several article and sound scripts in the vehicle's child prims to activate.
--Vehicle and rider follow the rail guide through the course of the loop until they unseat from the ride.
– Once rider has left the chair ,physics disengage,all particle and sound scripts are link messaged to halt and the car derezzes due to its temp status.

*Please note that using a non temporary rez car has shown no change in the operation of this construct other than leaving the car around after riding*

Havok 4 issues: 40 test rides were observed as a sample to find this bug

-20 seconds to 2 minutes into the ride, the rail guide will phase in the manner of a phantom object right through the track sending the vehicle on a pure ballistic trajectory off the rails.
--This behavior, while variable in its timing never took longer than 2 minutes and manifested in ALL 40 test rides
– the car after leaving the rail proved to be fully solid and not Phantom

Unpredictable issues:

  • during the phasing thru the rail, about 1 in 5 attempts would also cease forward motion and begin to roll on Y and Z-axis just off of the rail.

-A full copy of this construct and the relevent drive scripts will be forwarded to Spike Linden for review

Thankyou

Sabine Stonebender



 All   Comments   Work Log   Change History      Sort Order: Ascending order - Click to sort in descending order
Christophe003 Carter added a comment - 02/Jul/08 03:58 AM
High-velocity objects act as phantom since havok4, and it's not good.... in havok1 a much higher velocity was required for a same object to act phantom. – Chris

Sabine Stonebender added a comment - 26/Feb/10 01:33 PM
After a discussion with the current Linden in charge of physics, Falcon, a working solution to create a working rail guide was devised. The main fix is to eliminate any and all tori,tubes, and rings from the rails and the guides.
Rails and guides based upon cubes and cylinders appear to be the most stable under current Havok 4 conditions. This fix also applies to hinges requiring bearing type construction and seems to render those builds to more stable working order.Apparently the underlying reason is that at high speed prims collisions surface revert to a simpler form.While this is less of an issue for cubes and cylinders, it's catastrophic for tori as their bounding box reverts to a cube with no holes and simply phases off the rail from inter penetration rules.
Thanks for your time and the tips Falcon.
High speed Physical Roller coasters are once again a feasible option
cheers
Sabine Stonebender

TigroSpottystripes Katsu added a comment - 26/Feb/10 01:42 PM
won't that workaround (i refuse to call it an actual fix) require a significantly bigger number of prims for a smooth ride?

Sabine Stonebender added a comment - 26/Feb/10 02:01 PM
My initial rig did add about 12 prims . I'm currently testing lower prim configurations. As far as this being a workaround vs actual fix, the issues I had with collision based steering were fixed to my satisfaction. Following the discussion with Falcon I took a ride on a well built Japanese coaster that used a 2 prim rail guide with dual rails and was both fast and smooth using this method of omitting tori in favor of cubes. If not being able to use tori makes this more of a workaround and not a "fix" to your sensibilities, then by all means file your own JIRA to that effect. My problem is solved so I consider this JIRA resolved.