return prims

[Beth Marama]

it would be nice if the store gave away modifiable prims that we can shape and texture, (invisibly would be my choice) that TP the cats home when they touch it

this way we could put them under our skyboxes, on our property edges, etc, and they'd TP home automatically

[Tad Carlucci]

How would this be conceptually any different than setting the Range?

Points to ponder:

*) the suggested prim would only work when the cat is actively walking/running. When the cat is not waling/running, it's not physical, so no collision detection would occur, even through it appears to be 'moving' (ie., sitting, standing, laying, rolling around).

*) for this to work, the cat MUST NOT be allowed to touch the collision-script prim under normal conditions. So you'll need two prims, a non-collision-script floor and the scripted floor, So consider error modes such as 'forgot floor' and 'floor below detector.'

*) consider the case where Home is inside or close to the scripted prim. This leads to a high-speed collision/teleport loop.

*) consider a thin prim. The cat can still wiggle through by stopping (becoming non-physical) and rolling around a bit,

*) consider a sim with stalls using these prims, and, say, 1,000 cats. At some point all 1000 will be awake, walking on the prim, causing 20 collisions/sec/per cat. What are the lag implications of 20,000 collision events per second? Is there a throttle? Is so, the system fails because some cats won't collide and will still fall through the prims.

[Beth Marama]

ok, collision spam would be an issue, but maybe it could be scripted so that if it does get stuck, the prim shuts down automatically (say, 5 collisions with the cat in a second and the script stops?)

the basic idea for the prim isnt to keep them in range... youre right, the range setting is great for that... but range doesnt help vertical limits (maybe that would be a better solution actually, but let me finish this first)... the idea is to return the cat if it falls out of a skybox... the prim would be put down under the skybox as a net that auto returns the cat to its home point

but ok, thank you, you made me think about it and i have a different suggestion...

maybe script the range part of the cats so that if they slip out of range they tp back to home? that way if they fell out of a skybox they'd get TPd back, without any collisions, and hanging out at the range limit wouldnt cause any problems

very good points listed above, thank you very much, i wasnt thinking about collision spam... maybe i should ask make a new post and ask for this one to be deleted

[Tad Carlucci]

There are a lot of issues with self moving prims, such as KittyCatS.

Your second idea leads me to consider an alternate solution. Have two Ranges. When you set a distance-based range (since this won't work with Parcel and Sim ranges), set another range, say, 5 meters larger. If the cat finds itself outside the set Range, and moves beyond this second, fall-back range, it ceases movement and attempts to 'teleport' home as if it had fallen asleep.

This is not a full solution, but it would speed recovery in most cases. Instead of waiting 8 hours to go to sleep, it jumps Home early because it left the fall-back extra range.

Note, however, I said "attempts", quoted "teleport," and said "most cases."

This recognizes that cats do NOT actually teleport. That is not possible. Instead, what they do is move it fast, short jumps. (See Warpspeed teleport in the SL Wiki for how this is done.)

The problem is, it is possible for the cat to move in such a way that, when told to Jump Home (or TP To Me, or any other rapid, long distance move), that the cat can NOT get there in a straight line. There are two possible solutions to that problem: record every move and reverse course EXACTLY (which is infeasible due to script memory limits) or attempt a Drunkard's Walk (which may take a long time and is a probabilistic solution which may fail given any time constraint whatsoever).

The Path Finding added to Second Life a couple years ago was an attempt to reduce the size of the search space for a Drunkard's Walk. But it can't help, here, because it presupposes facts which are not true: that the region had an accurate Path Finding map (rarely true in my experience) and that the moving prim is walking upon a Path Finding surface (the ground or a prim), which is obviously not the case when falling from a skybox.