I have three shafts that I did a simultaneous breakthrough into a creeper chamber. They all have shields, then blasters, then missiles--everything is symmetric. (This is a sideways-build map, once the weapons have moved on I'm filling them with reactors.)
For a while I was watching missiles come up out of one shaft and dive back into another to attack the creeper--never mind that the direct route was to head down the shaft. It wasn't consistent, either--I even saw a missile dive down a shaft another missile had just come up out of.
It stopped once there was no more creeper touching the ends of the shafts, now the missiles all head down the shafts as expected. The only thing that makes any sense is that the computer was trying to path the missiles through a minimum amount of creeper.
Quote from: Loren Pechtel on May 28, 2012, 09:49:02 PM
The only thing that makes any sense is that the computer was trying to path the missiles through a minimum amount of creeper.
That's possible. there is in fact consideration in one of the path calculation algorithms for creeper avoidance. I remember in the early days of testing the Conversion Bombs ad that algorithm and avoided creeper. :)
We'd be able to judge better if you could attach the map or images or something.
Quote from: Karsten75 on May 29, 2012, 03:36:39 AM
Quote from: Loren Pechtel on May 28, 2012, 09:49:02 PM
The only thing that makes any sense is that the computer was trying to path the missiles through a minimum amount of creeper.
That's possible. there is in fact consideration in one of the path calculation algorithms for creeper avoidance. I remember in the early days of testing the Conversion Bombs ad that algorithm and avoided creeper. :)
We'd be able to judge better if you could attach the map or images or something.
The problem went away before I thought of a screenshot. If the missiles are using the same pathing routine as units do the behavior makes sense.
Can you recreate the incident? You still have the map, right? I'll do some tests when I get home.
It's fully possible for this sort of thing to occur without needing the missiles to avoid creeper (and my testing reveals no such tendency). A simple geometric explanation suffices.
EDIT: Here is a map to facilitate testing of this. It's more fun to figure out on your own.
Quote from: mthw2vc on May 29, 2012, 12:48:20 PM
It's fully possible for this sort of thing to occur without needing the missiles to avoid creeper (and my testing reveals no such tendency). A simple geometric explanation suffices.
EDIT: Here is a map to facilitate testing of this. It's more fun to figure out on your own.
The missiles were going much farther than they needed to and at no point did they save a single square.
Actually counting the squares may help you to realize that this is not so, much as your intuition may try to tell you otherwise.
Quote from: mthw2vc on May 29, 2012, 06:11:45 PM
Actually counting the squares may help you to realize that this is not so, much as your intuition may try to tell you otherwise.
A perfect count of the squares wouldn't be possible as I had multiple launchers together--no way to see for sure where a missile came from. The situation (simplified), however:
.*.*.*
.*.*.*
M*M*M*
M*M*M*
B*B*B*
B*B*B*
.*.*.*
c*c*c**<missile impact area>
cccccccccc
*<missile impact area>
. = open
* = solid
M = missile launcher
B = blaster
c = creeper
There is *NO* way the path up around the top is shorter than heading down. The roundabout paths did involve fewer squares with creeper, though.
Ah, I assumed you were referring to the test map I provided in my post. If you use that simplified situation and then count squares, you will see that the launcher's odd-looking path choices do not actually increase the distance traveled.