control event is broken

The control event appears broken when trying to take control from more than one agent. Even putting two separate scripts in two separate prims, the key passed to it seems to be the key of the first person to give permissions, not the key of the person actually doing the input.

I've reported this as a bug, but does anyone know any workarounds, short of creating a prim that's not part of the same linkset to take controls from the second agent?

Here's one potential test script that reproduces the bug, we've tried everything, but no matter what, the control event doesn't get the right keys (UUID type keys).

Put this in a prim and have two people touch it and then trigger some control inputs.

default
{
    touch_start(integer num)
    {
            llRequestPermissions(llDetectedKey(0), PERMISSION_TAKE_CONTROLS);
    }
    
    run_time_permissions(integer perm)
    {
        if(perm & PERMISSION_TAKE_CONTROLS)
        {
            // take all of the user controls
            llTakeControls(CONTROL_UP | CONTROL_DOWN |
                           CONTROL_FWD | CONTROL_BACK |
                           CONTROL_LEFT | CONTROL_RIGHT |
                           CONTROL_ROT_LEFT | CONTROL_ROT_RIGHT |
                           CONTROL_ML_LBUTTON | CONTROL_LBUTTON, TRUE, FALSE);
                           
        }
        else
        if(perm == 0)
        {
            llReleaseControls();
        }
    }
    
    control(key id, integer held, integer change)
    { 
        llSay(0,(string)llKey2Name(id));
    }
}

Gigs, its broke and has been for a while.

The only solutioon I've ever found was to only get control permission in child prims and send infor about the results via link messages.

If the root prim has 'em, it gets all the control messages, exclusively, and mis-identifies them as comming from whomever gave the root prim script permissions.

what about using the returned "key" to filter?

Jillian,

Thanks, that's one thing I hadn't tried, not having the root prim involved at all. Our tests were 1st av parent, 2nd av child. I'll try that and see how it goes.

Kyrah,

The key passed to the event gives you the wrong key! That's the problem I'm talking about here.