Random Hits OXP

[Screet]

This is a weird mix.

The fer-de-lance was the Constitution's mother? As the latter can only come from OSE on your system, that's impossible. They are never ever escorts?

UPS attacking Random Hits with an OSE ship in tow. Bizarre.

A friendly fire chain reaction perhaps?

The consitution I don't know about, it suddenly was gone while I was still fighting the asps...the asps had this upsescort script and were attacking in a way I've never seen before. I only noticed that these belong to the ferdy because I did shift-h them, but I shot down the ferdy before I checked the log and it wasn't flying that close together with the asps. You can see it on the screenshot - I did not notice it during that situation!


Screet

[Eric Walch]

Is UPS now attacking space bars?


Sadly, I shot down their master too early to know which AI he had to begin that attack. It was a pretty impressive sight and strangely the bar did have defenders in the air...there were a few yellow blips, but they did not fight these ships!

Screet

His master is a fer de lance and has the roleset:

Code: Select all

roles = "ups_pirate_leader hardpirate";

It is one of the more tricky pirates in ups therefor I also gave it a hardpirate role. So it could have been another oxp that called this one. ups itself calls him only for one or two special missions.

[Screet]

It is one of the more tricky pirates in ups therefor I also gave it a hardpirate role. So it could have been another oxp that called this one. ups itself calls him only for one or two special missions.

That hardpirate thing would well explain why they attacked the space bar! It was really a strange thing to happen, especially these asps flying like they've been set up to attack the bar on purpose. Never seen anything like that before, they would not even turn hostile to me but try to shoot down the bar, approaching slowly and continuous lasering it instead of the usual NPC flying style...and as you can see on the screenshot, they pretty nicely flew in formation!

Screet

[Screet]

In some systems I do get the same target systems with the same victim name, ship and description almost 100% of the time. I really do wonder who is resurrecting these people...

I'm not sure if RH does determine these things similar to galactic navy, but that oxp also had a similar problem. I fixed it by using a slight modification to the code, based upon the clock:

Code: Select all

this.ReserveDutySystemDetermination = function()
{
	var sec = clock.secondsComponent;
	for(var i = 0 ; i <= sec ; i++) {
		this.ReserveDutySystemDeterminationPt2();
	}
}

this.ReserveDutySystemDeterminationPt2 = function()
{

Screet

[pmw57]

Curses: should have replied, not edited

[Screet]

How on earth does that solve the problem? It looks more like a hack that relies on a side-effect.

It's a fix for Galactic Navy, put probably Random Hits does something similar.

The original code does use the many switches to determine the target system. That's done randomly from an array - but since random numbers in Oolite are pretty reproducible, it often happens that the player will get the same mission at the same target planet again and again.

By simply taking the current time into account to have the system being determined randomly multiple times it does change these numbers, not only for the target system but also for other things to happen

Random Hits could probably do something as simple as to use the current seconds to request a random number up to seconds times just in order to get more different random numbers on later usage of random numbers

Screet

[pmw57]

By simply taking the current time into account to have the system being determined randomly multiple times it does change these numbers, not only for the target system but also for other things to happen

I added the ps. part on up above, but I'll reiterate here for clarity.

The method now takes on average, 30 time longer to run. Sometimes nearly 60 times longer!!

[Screet]

The method now takes on average, 30 time longer to run. Sometimes nearly 60 times longer!!

So what? It's not noticeable. All it executes is one line!

Screet

[pmw57]

The method now takes on average, 30 time longer to run. Sometimes nearly 60 times longer!!

So what? It's not noticeable. All it executes is one line!

Screet

The original this.ReserveDutySystemDetermination method consists of 738 lines of code. It's a large switch statement (that really needs to be rewritten) that is determined entirely upon galaxyNumber and system.ID

From what I see in that piece of code, its behaviour will not change no matter how many times it is run.

[Kaks]

So what? It's not noticeable. All it executes is one line!

Err, if I may... You might have noticed that trunk is more responsive than 1.73.4
Why? Because we do remove inefficient code when we find it, and replace it with faster one. If you make changes that slow down oxps & Oolite, sooner or later - and my bet is sooner - you'll start to notice that the gameplay itself is compromised.

However, if you're happy to create inefficient code, or make not-that-efficient code more inefficient, I suppose it's really up to you. And yes, on average executing a 738 instruction 'one-liner' 30 times makes that part of the code 30 times slower than before, and is the equivalent of running 22140 lines.

[Eric Walch]

The original this.ReserveDutySystemDetermination method consists of 738 lines of code. It's a large switch statement (that really needs to be rewritten) that is determined entirely upon galaxyNumber and system.ID

From what I see in that piece of code, its behaviour will not change no matter how many times it is run.

When you want better random system numbers, those numbers should not be selected from the description.plist but selected with Math.random() from a number array within the JS script.

I think that the bad randomness of numbers is caused by patrolling police ships regularly asking for the witchpoint coordinates. To calculate that position it needs to reset one of the random seeds.

[pmw57]

And yes, on average executing a 738 instruction 'one-liner' 30 times makes that part of the code 30 times slower than before, and is the equivalent of running 22140 lines.

Well if it helps, here is a 20-line version for galnavymain.js instead of the 738-line one.

Code: Select all

this.ReserveDutySystemDetermination = function ()
{
	// sector is organised as [galaxy][system]
	var galaxySectors = [
		[0, 24, 193, 154, 188, 125, 109, 249, 150, 80, 175, 120, 220, 158, 15],
		[0, 248, 135, 106, 82, 150, 182, 33, 48, 227, 96, 127, 202, 243, 170],
		[0, 31, 165, 245, 84, 36, 21, 26, 139, 251, 205, 223, 204, 17, 247],
		[0, 39, 221, 188, 86, 9, 57, 186, 237, 49, 103, 254, 66, 140, 110],
		[0, 249, 71, 57, 144, 198, 153, 204, 73, 102, 222, 29, 233, 152, 190],
		[0, 151, 227, 6, 51, 60, 129, 8, 53, 85, 120, 208, 220, 146, 66],
		[0, 214, 161, 250, 127, 23, 154, 234, 24, 55, 64, 231, 140, 202, 35],
		[0, 26, 218, 9, 174, 156, 251, 75, 136, 184, 87, 212, 116, 98, 55]
	],
		sector = galaxySectors[galaxyNumber].indexOf(system.ID),
		description = 0;
	if (sector > 0) {
		description = expandDescription('[galaxy' + galaxyNumber + '_sector' + sector + ']');
	}
	missionVariables.reserve_planetnum = description;
};

[Screet]

The original this.ReserveDutySystemDetermination method consists of 738 lines of code. It's a large switch statement (that really needs to be rewritten) that is determined entirely upon galaxyNumber and system.ID

From what I see in that piece of code, its behaviour will not change no matter how many times it is run.

It does change it, try that modification which I did post and then add a message to log the target systems number each time this is run

Most probably, it would have been enough to simply request a random number and then finally call that other method once.

It's just when I did make that modification several months ago (for a much slower oolite version, BTW) I did want something which I could quickly do without much knowledge about those things.

As to the question of slowing things down, it's only run once when the player is docked and requests a new mission - and really, even on the much slower computer I had then with the much slower Oolite I was running then I did not notice any change in responsiveness

Guess it's all simply running inside of the cache anyway, and although the switches are there, it's not that all those lines inside there would be executed: First you can divide it by 8 due to the galaxy switch and then it's still only checking values, which also is pretty fast, until it finally found the proper planet where it then only has to execute one line.

Of course, your shorter version does look much nicer. Would have to test, though, if it's also creating the desired randomness...

Screet

[pmw57]

It does change it, try that modification which I did post and then add a message to log the target systems number each time this is run

So that we're on the same page, are these changes being made to galnavymain.js ?

[Screet]

It does change it, try that modification which I did post and then add a message to log the target systems number each time this is run

So that we're on the same page, are these changes being made to galnavymain.js ?

[/quote]

I made them to my local version of that file, yes. It's simply something which I thought I should suggest now, as it's a simple modification that solves the only problem there and I also did post it here because Random Hits suffers from the same problem (although I did not yet look at the cause). Really, most previous messages would have better been written in the GalNavy thread

What I did to ensure that it's really changing the values selected, I did add something like
log("GalNavy",missionVariables.reserve_planetnum);
temporarily. When the selection is done multiple times, there's also a different random value being used for the mission selection, thus it's also helping to get more different missions, not only more different target systems.

Thus, should your much more readable code still have the old tendency of selecting the same systems, maybe it's enough to simply request n times (for n seconds) a random number before running that code - and I'd guess that the next version of GalNavy would then come with a shorter and better script

If, as Eric writes, the Random Seed is reset during flight, RH most probably also would only require to use time or something like that to request random numbers it ignores before it begins to use some. That's (for seconds) at least giving 60x more diversity

Screet