llSensorRepeat is not functioning properly

Did anyone else notice that the recent results from llSensorRepeat point over an intresting problem with the function?

While running llSensorRepeat at a high-prim enviroment, I noticed that the results will change from listing nearby objects to objects that are located somewhere completly else on the sim.

The results keep alternating between proper replies (objects in the nearby's object proximity radius) to bad results from somewhere completly different in a consistant fashion.

Just check out this sensor swip log:

The scan parameters are: PASSIVE, 94m range, every 4-seconds.

The object "Crystal's" is the closest object placed next to the sensor to be used as the test object for the sensor to find.

The scan was performed at the RELIC shop. Most of the items in the sensor radius should have the name "Relic" in their name field.

Swipe A
---------
- Crystal's
- RELIC - The Lost Garden Building Kit, IRON RAILING, 1 prim
- RELIC - "The Lost Garden" Building Kit, DBLE TXTRD RD DECK, 1pm
- RELIC - "The Lost Garden" Building Kit, IVY WALLa, 1 prim
- RELIC - "The Lost Garden" Building Kit, IVY WALLa, 1 prim
- RELIC - "The Lost Garden" Building Kit, LG CVRD ARCHWAY, 13 prm
- RELIC - "The Lost Garden" Building Kit, IVY WALLa, 1 prim
- RELIC - "The Lost Garden" Building Kit, CURTAIN-B, 1 prim
- RELIC - "The Lost Garden" Building Kit, IVY WALLa, 1 prim
- Object
- RELIC - "The Lost Garden" Building Kit, WALL W/WINDOWa, 1 prim
- RELIC - "The Lost Garden" Building Kit, TOWER PIECEb, 31 prim
- RELIC - "The Lost Garden" Building Kit, IVY WALLb, 1 prim
- Object
- "The Lost Garden" Building Kit, WALL W/WINDOWa, 1 prim
- Pine w/ snow 6.5m

Swipe B
---------
- Pine w/lot of snow 10m
- Pine w/ snow 10m
- Tiki Torch
- Candle Burgendy
- Candle Burgendy
- Object
- Object
- Object
- Object
- Object
- Mountain Home
- Object
- Object
- Object
- Object
- Pine w/ snow 6.5m

Huh... it almost seems like each subsequent pass is detecting 'the next 16 nearest objects', or something like that. I was wondering what was going on with my particle-based radar system... it was giving very weird, spurious results like that, too.

This seems to be more or less consistent with a test I did a long time ago trying to detect
agents.

I went to Ahern to try the sensor with a lot of av's around, and it would give highly varying results.
As soon as there was more than the 16 results that the llSensorXX can report, it would miss people that were right beside me, despite the doc saying it will report in order of proximity.

I built a radar last year. On occasion it would report someone as being 250 or so feet away. Sure enough, I'd look at the mini-map and there would be a dot - clear over in the next sim, much further away than the 96 meters that sensors are supposedly limited to.

I consider any case where a sensor would match more than 16 obejcts/agents to be a divide-by-zero-like "undefined" behavor.

A maximum of 16 objects are passed to the sensor event. This has been added to the LSL reference docs.

Dan Linden