Event queuing

default
{
touch_start(integer total_number)
{
llSay(0, "Touched.";
llSleep(1.0);
}
}

Now click on that a few times quickly. What would you expect? I thought the script had an event queue that would hold up to 64 events, so I'd expect it to say "Touched" once a second, for as many times as I clicked on it. What I see, makes it look like the script receives no events while it's in the llSleep call. I saw the same thing with an llInstantMessage in place of the llSay and no llSleep. So it looks like, when a script is blocked, it doesn't queue up any events.

Maybe it's always been this way, and I've never noticed it? I thought events got queued up, and you'd only lose events if you get past 64 to any individual script.

My understanding is that when a script 'sleeps' it is dead to the world. Events don't register at all, therefore nothing is cued. But, the cue works when... say running a lengthy user-function, and someone clicks during the function... once the function clears, the click is registered and processed.

The sleep = dead to the world is useful for times you may want a script to ignore everything for a while. Say, a collision detector in a goal which needs to register one, and only one collision event with a ball... but the ball might spam several collisions if left alone. So, you llSleep() the script for a few seconds after the collision, to allow time to kill the ball.

I imagine llInstantMessage() has a built-in penalty sleep which works the same as llSleep().

Not a perfect explanation but... says:

More specifically:

Using tough_start as a test may not be the what you want in this case, since, in theory, rapid sequential touches may not trigger multiple events even without a sleep. Instead, they are added to the total_number parameter, and you are expected to process them one by one yourself. (While this seems rare, and usually you can get away with llDetectedKey(0), multiple touches could require a for loop, and llDetectedKey(1), for instance, to process all the touches).

I doubt I was clicking fast enough to get multiple hits in the same touch_start. And it registered for each touch fine without the llSleep.

So either touch happens only once, or I only get one event per type. I guess I should have checked the Wiki first I could test this with a timer I guess and see what pattern that follows.

A simple test of mine seems to show that it behaves as described in the Wiki. I used a long sleep, and touching it any number of times after the first touch causes it to enter the touch handler exactly once after the sleep. The count was always one.

default
{
    state_entry()
    {
        llSay(0, "Hello, Avatar!");
    }

    touch_start(integer total_number)
    {
        llSay(0, "Touched " + (string)total_number);
        llSleep(10);
        llSay(0, "sleep done");
    }
}