llSetKeyframedMotion() loses power crossing sim border

[Jaiden Magic]

I have been attempting to produce a train of sorts using llSetKeyframedMotion(). It works beautifully in the preliminary tests within one sim, but when I tried to cross a sim border with it, the function loses the ability to do anything after crossing into the next sim. The sensor that powers it is still going, triggering the no_sensor events that set off new commands of the next keyframe. But it will not move, nor will it resist my manual movement of the object as it would when working normally in the sim it started in. Moving it back into the sim it was in lets it resume keyframed motion as normal. Stopping and playing keyframes doesn't work, nor does resetting the script or dragging a new copy in the new sim. The only way I can get it to work in the new sim is to pick it up into my inventory and re-rez it across the border, after which it works normally. So it seems to me that llSetKeyframedMotion only works in the sim that the object is natively rezzed to.

The LSL wiki says "Inter-region are supported."

Has anyone else tried inter-sim motion? Thoughts? Suggestions?

[Dora Gustafson]

Just a wild guess: will it work with grid coordinates?

[Jaiden Magic]

How would I use them? The movement vector passed into llSetKeyframedMotion isn't a "Go to this spot" vector as much as a "Go west three paces" vector, not caring where the object currently is.

[Dora Gustafson]

You are right, silly me, I didn't think:smileyembarrassed:

[Jaiden Magic]

:matte-motes-smile: It's fine, I've just looked at this function 20 diffferent ways trying to fix this problem. Thanks for trying.

[Nika Talaj]

I've just run into this same problem.  However, in my case the motion will resume if I PAUSE and PLAY it (which isn't helpful, but is interesting).  I don't see anything in Jira about it.   Did you find out anything more, Jaiden?

 

Alternatively, does anyone have a keyframed motion that DOES successfully cross a sim boundary?

[Dora Gustafson]

---

Nika Talaj wrote:

Alternatively, does anyone have a keyframed motion that DOES successfully cross a sim boundary?

---

Since my last post I verified that the motion dries out a few meters across the sim border.

I can only conclude that the specification: "Inter-region are supported" is wrong

[Haravikk Mistral]

Have you tried using any of the other physics function such as llSetForce, or toggling the physical state of the object with llSetStatus? Not that either of these are really any kind of solution, since the goal is smooth movement between sims.

I've posted a JIRA issue for the sake of tracking this problem, but it'll probably be easier to get this issue fixed by attending the right user group meeting to get it some attention.

[Nika Talaj]

---

Haravikk Mistral wrote:

Have you tried using any of the other physics function such as llSetForce, or toggling the physical state of the object with llSetStatus? Not that either of these are really any kind of solution, since the goal is smooth movement between sims.

 

---

llSetKeyFramedMotion does not work on physics-enabled objects.  It is more akin to llSetPos than llMoveToTarget.

 

ETA: Thanks for filing the bug, Haravikk!   (I wouldn't be surprised if this had to be changed to an SVC bug, though.)  It looks like the appropriate meeting is Kelly's scripting User Group, which is meeting today at 9am SLT, after having been on hiatus for a bit.

http://maps.secondlife.com/secondlife/Ambleside/157/91/30

It's unlikely that I'll be able to make the meeting today.

[Jaiden Magic]

The bug must have gotten some attention. After a few days' break from the project, I came back to it, using non-keyframed motion to cross the border, and it works perfectly. The bug must've gotten some attention and got fixed. Thanks for the help, everyone!

[Nika Talaj]

The bug Jira Haravikk opened appears to be addressed in the Pathfinding code, which is rumored to be headed for release around mid-July.   I gather that there have been a few oddities noticed in llSetKeyframedMotion across sim boundaries; hopefully they'll all be fixed in the pathfinding code!