Slow speed physics broken?

There seem to be several problems with physical objects moving at slow speed.

These may be related to Jira bug VWR-86, which unfortunately it appears LL has decided not to fix anytime soon so we may have to learn to live with them.

1. At speeds below about .5 - 1 m/sec vehicle motion is very jerky. A related issue is the vehicle buoyancy floating is also very jerky for buoyancies near neutral that would cause slow motion. Rather than slow smooth movement, the object repeatedly jumps. There seems to be no fix for this.

2. If you set the vehicle buoyancy to be very close to neutral, in my case this is about .8, the vehicle stops moving even though the linear motor is running. Changing the vehicle buoyancy to above .82 or below .78 avoids the freezing problem. Calls to llGetVel return 0 and the position stops changing.

3. Not using vehicle calls, just using impulse, the motion is smooth, however, there seems to be an issue that for slow velocities velocity changes sometimes fail. That is, rather than the object attaining the new velocity it stops moving, calls to llGetVel return 0 and the position stops changing. This is very random but is a bigger problem for slow velocities and appears also dependent on how far the object has been moving before the change in velocity - shorter distances are worse. Generally this stopping occurs when the velocity is changed, however occasionally it will just stop moving without any impulse being applied, and occasionally spontaneously start moving again. Usually though, if it stops it will never restart unless a new impulse if applied, and sometimes not then. This can be avoided by monitoring the actual motion and repeatedly applying impulses if it stops moving.

This is really annoying, and it's bothered me a lot. I have to ask, though... isn't this the way things have been even before VWR-86 reared its ugly head? I mean, I thought slow-moving things have had a tendency to "settle" and stop moving. I thought that was just how the physics system tried to save itself work. I've seen stuff like that happen for most of the time I've been in SL.

Lex - the slow speed vehicle jerkiness and buoyancy related freezing at least are new, I've been working on a project for awhile and before Jan, they were moving very nicely, now they are jerky and pretty useless.

Since that killed my vehicle based project, it forced a rewrite for me which is when I found the impulse issue. I'm not sure if the impulse stopping is new or not, but it's suspiciously similar to the vehicle bug.

The "jerkyness" in llApplyImpulse() got introduced at the same time the jerkyness got introduced in other physical movement commands, like the vehicle-system.

It seems like LL has done some small changes in the couple of latest patches that makes it "abit better", but its still "jerky as hell" in my eyes.

*crossing fingers about LL fixing it soon(tm)* - I loved playing with llApplyImpulse until this problem occured. Now llApplyImpulse() just makes me want to puke.

I've got a low-mass (< 1 unit) object which uses llApplyImpulse to move at a low speed (1 m/s), and I've noticed this behaviour.

Being a total noob I assumed this was some friction thing. I got it to some semblance of working by setting bouyancy to 0.9 and specifying a force 3 times the desired force, and doing the impulse every second. Beautiful - not!!!