Priorities about mini-runner

[drkibitz]

To converge on the priority stuff, I'm seeing rather quickly that we would need to wrap your items in a listener object. Nothing to do with scoping the functions, it is just to have a higher level object to hold the reference, with description properties next to it, e.g. priority.

Here's what I'm thinking (note the changes to the loop in generateRun):

p.add = function(item, priority) {
    if (!item[this._name]) return;

    this.remove(item);

    const listenersCopy = this.listeners.slice();
    const l = listenersCopy.length;
    const listener = { item, priority };

    if (l > 0) {
        let i = l;

        while (i--) {
            // from end - if next has lower priority, insert now after
            if (listenersCopy[i].priority < listener.priority) {
                listenersCopy.splice(i + 1, 0, listener);
                break;
            }
            // from start - if next has equal priority, insert now before
            else if (listenersCopy[l - (i + 1)].priority === listener.priority) {
                listenersCopy.splice(l - (i + 1), 0, listener);
                break;
            }
        }
        this.listeners = listenersCopy;
    }  else {
        listenersCopy.push(listener);
    }
    this.listeners = listenersCopy;
};

MiniRunner.generateRun = function(name, argsLength) {
    const key = name + '|' + argsLength;
    let func = MiniRunner.hash[key];

    if (!func) {
        if (argsLength > 0) {
            let args = 'arg0';

            for(let i = 1; i < argsLength; i++) {
                args += ',arg'+i;
            }
            func = new Function(args, 'var listeners = this.listeners, i = listeners.length; while(i--) listeners[i].item.'+name+'('+args+');');
        } else {
            func = new Function(args, 'var listeners = this.listeners, i = listeners.length; while(i--) listeners[i].item.'+name+'();');
        }

        MiniRunner.hash[key] = func;
    }

    return func;
};

[drkibitz]

@GoodBoyDigital so I made some changes locally, and I'm seeing something odd, too good to be true even. Take a look at this:

benchmarking signals...
signals time 1.129935
listener updates 2000
benchmarking mini-signals...
mini-signals time 0.1500250000000001
listener updates 4000
benchmarking events...
events time 0.08157999999999993
listener updates 6000
benchmarking runner...
runner time 0.00013999999999987268
listener updates 8000
runner is 8070.964285721625x faster than signals
runner is 1071.607142858118x faster than mini-signals
runner is 582.7142857148152x faster than events

This is consistent... compared to the 25x, 2x, 1x average without my changes.

I can't believe my changes would have increased performance 500%, my expectation was that it might have dropped it by like 0.1% or something. I added the listener updates # log to make sure things were being called. I'm kind of at a loss here. See anything out of the ordinary, or curious? Make a PR? ;)

[drkibitz]

Nevermind I found the issue, definitely too good to be true.