Having problems with PlayGame.jar or ShowGame.jar?

[a1k0n]

There are a bunch of bugs in both of the starter packs that have been released so far (and it's kind of hard to tell which one you already have). Most of the known bugs in the engine and visualizer have been fixed, but the updated starter packs haven't been released yet (there are several other pending issues, notably getting everyone's games to run and keeping the site up).

So I have just updated these:
http://a1k0n.net/temp/PlayGame-1.2.jar
http://a1k0n.net/temp/ShowGame-1.2.jar

Bugs fixed by this release:
- Fleets randomly disappear right before capturing a planet
- When fleets from both players land on a neutral simultaneously, the resolution was inconsistent and wrong. This has been corrected according to the rules set forth in some other threads on the msgboard here, and will also be described in detail on the wiki.
- ShowGame.jar takes forever to load games of any reasonable length
- stderr from each bot is swallowed by PlayGame; now it is output to PlayGame's stderr, prefixed with "Player <n>: <stderr line>"
- If the locale was one in which , and . are used differently than they are in the US, PlayGame would fail to parse the maps
- Windows line-endings would break PlayGame
- PlayGame runs much more slowly than it should (several un-optimizations were removed)

Known bugs:
- 0-growth planets show up with 0-size circles, which is very confusing.

Try them, see if they fix your problem.

I believe the plan at this point is to release a set of starter packs at version 1.2, and they will almost certainly contain the above two jars unless any other bugs get fixed before the release.

[tjkoblentz]

Thanks a1k0n! I'll link to this thread from the one I just posted regarding the 0-growth rate graphics bug. It's nice of you to do this.

[a1k0n]

There is an bug logged in the issue tracker for that, BTW: http://code.google.com/p/ai-contest/issues/detail?id=71

[tjkoblentz]

Sweet the competition's on Google code (I actually read that somewhere earlier)! Thanks for the link.

[temp_dummy]

Anyway, I do have a patch for the viewer that would allow it to play backwards (in theory).
I do not have a working Java environment, but just looking at the code I can see that the forloop does not take into account the direction.

svn/ trunk/ planet_wars/ viz/ ViewerPanel.java

// Makes the game objects
public void run()
{
// <SNIP> find the loop after: tGame.Init();

Replace For with:
int i = 0;
int direction = 1;
if(!forward){
direction = -1;
}

While(i >= 0 and i < turns.length)
{

// <SNIP> Goto end of the loop
i = i + direction;
}
If this compiles, it will play backwards.
---------------------------------------------
NOTE : Just took the latest version and it does run a lot faster !
Makes testing against all maps easier.

Feature request:: Display the move number in the visualizer.
This can help with debugging as I write each gameboard/move to disk and load re-load it for debugging,
If the move number is displayed I know which file to pick. (thank you)

[temp_dummy]

I found a bug in which the (new version of the ) game engine does not allow me to take over an planet, no matter how many ships I send to it.

The bug is consistently reproducable:
Windows,
C#
map100.txt
Fighting "Rage" (in more ways than one.)

I have included my game trace:
1) Around move 150 you can see that player two has almost complete over the map.
2) In move 166 there are just two planets left, with two ships each.
An entire Armada is coming to get them, but they never will.
3) this continues until move 1001 where Player 2 is rightfully so declared the winner

I hope this has been informative enough?
The big problem is that if we can not trust the engine, how can we trust our bots?

[kukems]

Hi, I'm using ant to build and run my bot, but I have issue with ShowGame-1.2.jar.

[exec] Exception in thread "Thread-2" java.lang.NumberFormatException: For input string: "Turn 20"
[exec] at java.lang.NumberFormatException.forInputString(NumberFormatException.java:48)
[exec] at java.lang.Integer.parseInt(Integer.java:449)
[exec] at java.lang.Integer.parseInt(Integer.java:499)
[exec] at ViewerPanel.run(ViewerPanel.java:108)
[exec] at java.lang.Thread.run(Thread.java:619)

http://www.filefront.com/17283967/gamelog.txt See gamelog.txt.

[temp_dummy]

You are likely running the code on a machine that does not use the american 'locale' for the decimal separator.

I fixed this by Using the 'invariant' local, which according to the help files of my development tools was designed for inter-machine communication.

//This is what is looks like.

Code: Select all

   public static int ParseInt(string value)
   {
      return int.Parse(value, CultureInfo.InvariantCulture);
   }
   public static Double ParseDouble(string value)
   {
      return Double.Parse(value, CultureInfo.InvariantCulture);
   }


There other ways to change the local, you can set it on the thread,
change your windows settings,
and quite possibly somewhere in config files.

*I did not bother because this worked*

[Highie]

Thanks a lot!
It fixed my problems with

Code: Select all

java -jar tools/PlayGame.jar maps/map7.txt 1000 1000 log.txt "java MyBot" "java MyBot"  | java -jar ShowGame.jar
WARNING: player 1 timed out.
WARNING: player 2 timed out.


[Tarski]

Code PlayGame.jar be modified to display errors from bots crashing? E.G. If my bot throws an exception, stderr is swallowed by PlayGame.jar and the bot times out. This makes it difficult to debug. Could the bot's stderr be redirected to stdout by PlayGame.jar?

Cheers