[Updated on: Fri, 18 July 2008 02:46]

Report message to a moderator

[Updated on: Fri, 18 July 2008 03:05]

Report message to a moderator

Report message to a moderator

gible wrote on Fri, 18 July 2008 09:02

So theoretically, I suppose you could dispense with the crash-sweepers-must-have-lower-fleet-ids issue by making them target another fleet

In practice, however, it's very advisable to testbed that idea 1st.

And please publish the results!

Report message to a moderator

Adacore wrote on Fri, 18 July 2008 08:46

I understand that, normally, fleets move in an order determined by fleet number, but what happens when a fleet has an order to intercept another fleet? Does the fleet move after all the fleets without interception orders (ie are there two 'passes' through the fleet list, the first for standard movement, the second for interceptions)? Does the fleet move immediately after the fleet it is to intercept?

There's enough serious bugs related to fleet interception (the "overshoot" and the "stuck pursuer" to name but two of the most common) that anything related to it should be thoroughly testbedded anyway.

Report message to a moderator

So we have an estimation of "smooth" chasing (when the positions of ships are being updated in steps) only in the case of circular reference? And when there is no circular reference then the chased fleet is moved and then the chaser tries to get to that position, right?

At first it seems there is no problem about it. But theoretically this can be exploited. For example imagine small enemy fleet positioned at (1000;1000). It is quick (having warp 10) and targets your fleet with some unprotected cargo ships at (1080;1000). Cargo ships are very slow (warp 4) and can't reach any homeworld within the next turn. The good news is that you have a large army positioned at (1012; 1000) (that is between the enemy and cargo fleets). The bad news is that this army is slow (warp 7).

Scenario 1: cargo ships try to run away to the nearest homeworld, army tries to intercept the enemy. But the both fail - the enemy fleet cathes the cargo ships the next turn and destroys them all. The next turn enemy has no problem running unpunished saying "haw-haw-haw".

Scenario 2: having read Jeff's explanation our commander suddenly gets the cargo saving plan: he just needs to target his army with his cargo ships! By doing this he gets circular reference: enemy -> cargo -> army -> enemy ->... and the movement is calculated in steps. So the movement begins. In fact there is no difference whether the enemy fleet moves first or the army does. Enemy fleet moves 10 * 10 / 10 = 10 l.y. each step and army 7 * 7 / 10 = 5 (if it rounded to 4 the result of the example is the same). If enemy fleet moves first then: enemy -> (1010; 1000) and army has no problem catching it. If army moves first then: army -> (1007; 1000), enemy -> (1010; 1000) and army catches it. Hurray! The cargo is saved!

I know this example is pretty unreal but there are many others which, I believe, are encountered often. Every time we have our ship try to really intercept (and not move to that target waypoint 1 position).

So I think it's a flaw in game that the movement is not always calculated "smoothly". The reason why it is so is clear: in order to get proper distances traveled by ships which are not involved into any chasing the program must use floating values while calculating ships positions at each step and rounding the position only at the final step (or else if we use rounding on each step the ship with warp 6 can travel 40 l.y. instead of 36).

Report message to a moderator

paul_ik wrote on Mon, 21 December 2009 13:06

Scenario 1: cargo ships try to run away to the nearest homeworld, army tries to intercept the enemy. But the both fail - the enemy fleet cathes the cargo ships the next turn and destroys them all. The next turn enemy has no problem running unpunished saying "haw-haw-haw".

Nothing new here. Where's the exploit?

Quote:

to get proper distances traveled by ships which are not involved into any chasing the program must use floating values while calculating ships positions at each step and rounding the position only at the final step

I have never seen ships notoriously misbehaving in that case, except if the two known movement bugs are triggered. Is there supposed to exist a new exploit?

Report message to a moderator

[Updated on: Mon, 21 December 2009 12:43]

Report message to a moderator

paul_ik wrote on Mon, 21 December 2009 12:06

Scenario 2: having read Jeff's explanation our commander suddenly gets the cargo saving plan: he just needs to target his army with his cargo ships! By doing this he gets circular reference: enemy -> cargo -> army -> enemy ->... and the movement is calculated in steps. So the movement begins. In fact there is no difference whether the enemy fleet moves first or the army does. Enemy fleet moves 10 * 10 / 10 = 10 l.y. each step and army 7 * 7 / 10 = 5 (if it rounded to 4 the result of the example is the same). If enemy fleet moves first then: enemy -> (1010; 1000) and army has no problem catching it. If army moves first then: army -> (1007; 1000), enemy -> (1010; 1000) and army catches it. Hurray! The cargo is saved!

OK. Your army has caught the enemy in the first iteration of the 10.
On the second iteration the faster moving enemy will just outrun your slow moving army.
i.e. you temporarily catch them, but then get stranded by them.
Battle will not occur at the split second that you happen to have caught them, but at the end of the 10 iterations.

Report message to a moderator

Report message to a moderator

paul_ik wrote on Mon, 21 December 2009 18:42

the second one seems to be illogical but leads to the desired result. That is why I think the system of movement calculation in stars a bit buggy. Just think - we have army that is positioned on the path of the enemy. And still the army can't intercept the enemy.

Weird, perhaps. Buggy, at least in the case you outline, not quite.

You cannot stop a fleet traveling at warp. Period. The only things that can stop such a fleet are lack of fuel, engine malfunction and minefields. Try sending a fleet _through_ a stellar mass. It will safely arrive to the other side, every time.

Space has its quirks. Look up how to intercept packets, or the MT. It always works that way.

Report message to a moderator

Report message to a moderator

In my example cargo and army were unable to reach each other in one turn. Targeting army with cargo was supposed to produce circular reference and calculation of movement in 10 steps (instead of plain ordered movement of fleets one by one), which, as I've thought, will lead to army being capable of catching the enemy. But as mazda has stated the battle doesn't occurs if fleets meet during one of the steps. So my theoretical exploit doesn't work.
Sorry for disturbing!

Report message to a moderator

paul_ik wrote on Tue, 22 December 2009 19:22

In my example cargo and army were unable to reach each other in one turn.

With the speeds and distances you stated they cannot meet, thus the enemy can kill the cargo and run away laughing. That's why it usually pays to have enough fuel for your fleets, as it allows better manoeuvering.

Still no exploit that I can see. Or have your testbeds shown otherwise?

[Updated on: Wed, 23 December 2009 20:19]

Report message to a moderator

m.a@stars wrote on Thu, 24 December 2009 03:18

Still no exploit that I can see. Or have your testbeds shown otherwise?

mazda has killed my hope on exploit, but I'll run the test today's evening in order to be sure.

Report message to a moderator

In my testbed the enemy was able to reach the cargo ships. The army could do nothing. At the end of the turn the army was positioned at (1041; 1000). Which is rather strange and I can't explain this position. The army should at some step (either first or second) be at (1010; 1000) which means the army should then fly 31 l.y. in either 8 or 9 steps. This makes no sense with warp 7 (in fact the army was expected to travel either 0.8 * 49 = 39.2 or 0.9 * 49 = 44.1). Even if the ship ability to travel 4.9 l.y. each step is rounded to 4 l.y. the army was expected to travel at least 32 l.y.
So there is some magic behind the scenes!

Report message to a moderator

Yes, there is. Because the way the game calculate travel, travelling in diagonal can sometimes create "magical" outcomes.

I've used this features several times to be able to bring a fleet which, theorically, didn't have enough fuel to do the full trip (used it twice to catch far away MTs!). And I learned this technique from the bests.

-EoF

Report message to a moderator

paul_ik wrote on Thu, 24 December 2009 19:16

In my testbed the enemy was able to reach the cargo ships. The army could do nothing. At the end of the turn the army was positioned at (1041; 1000).

What was the position of the cargo/enemy? Did they travel at the speed you mentioned?

Report message to a moderator

m.a@stars wrote on Fri, 25 December 2009 21:03

paul_ik wrote on Thu, 24 December 2009 19:16 In my testbed the enemy was able to reach the cargo ships. The army could do nothing. At the end of the turn the army was positioned at (1041; 1000). What was the position of the cargo/enemy? Did they travel at the speed you mentioned?

m.a., has that diagonal movement exploit been taken into account for the Stars! Trip and Fuel Usage Calculator ? If it hasn't, then it should if it can be verified.

Report message to a moderator

I have good news! The exploit works! The cargo is saved! Hurray!

So the thing are a bit more complicated than we expect them to be.
This time I decided to run different tests and here are the results.

I've created two players game. Both players were JoaT. The attacking enemy was player #1. The fleet was stalwart defender. The army was stalwart defender as well and the cargo was santa maria. I've set the conditions described in my example: positioned fleets on (1100;1200), (1112;1200) and (1180;1200) (yeah, I've added 100 to each x); set circular targeting with warp 10, 7 and 4 respectively.

magic #1: the army catched the enemy! The battle took place at (1120; 1200). I've run the situation several times (with the same fleets, until my poor defender was destroyed by warp 10 ) and the result was the same. I've tried changing the battle orders for enemy to attack only unarmed ships, this had no effect.

Then I removed circular reference, making santa maria running away in the opposite direction. The next turn enemy catched it and the army was positioned at (1161;1200). No intercepting, nice.

magic #2: then I've changed the sides, making player #2 attacking. Set circular targeting. This time the result was as I've described earlier: the enemy hit the cargo, the army was positioned at (1141;1200).

So, the exploit seems to work depending on the fleets numbers And... a new bug?

Report message to a moderator

craebild wrote on Fri, 25 December 2009 22:52

m.a., has that diagonal movement exploit been taken into account for the Stars! Trip and Fuel Usage Calculator ? If it hasn't, then it should if it can be verified.

Not only was it taken into account, it's at the heart of the multi-hop calculator.

Report message to a moderator

paul_ik wrote on Sat, 26 December 2009 00:51

So, the exploit seems to work depending on the fleets numbers And... a new bug?

Okay, that potential nightmare deserves further testing. Hopefully I'll have some free time one of these days.

Meanwhile, you could continue your tests. See if you can find a pattern or something.

Report message to a moderator

paul_ik wrote on Fri, 25 December 2009 23:51

I've set the conditions described in my example: positioned fleets on (1100;1200), (1112;1200) and (1180;1200) (yeah, I've added 100 to each x); set circular targeting with warp 10, 7 and 4 respectively. magic #1: the army catched the enemy! The battle took place at (1120; 1200).

Interesting.
So the two warfleets have only moved 20 l.y. and 8 l.y. respectively ?
Where did the cargo ship finish ?

Report message to a moderator

Report message to a moderator

Report message to a moderator