Introversion

You've given the SetActionTarget wrong coordinates, thus silos are firing at 0,0 on the grid. Perhaps you have x and y reversed or something. It's been a while.

Edit: Hehe, didn't read the com chat fully, didn't see the 'bad argument' comment. (trying to pass a table as x,y)

I've got it fixed, but i thought you all would enjoy sharing that fail moment with me.

Here's the updated source-in-progress. The bot is very random, aggressive, and uncoordinated at this point. http://owenjohnson.info/dev/defcon/strangebot.zip

YAY!!
More YAY!!

Congrats!

BUT!
First match:
Dr. Strangebot placed in enemy radar range and still has 15 nukes.

Second match:
Spreaded silos? Your Bot will have no chance vs human players. You know, there are guys out there who can triangulate silos via their AA fire?
Still 20 nukes left.

Anyway, if your Bot is better than the CPU you have made huge jump.
At this stage, is it possible to play against your Bot?

The placements are random at this point, and the navy is hardcoded, which surprisingly can beat the cpu. It's nowhere near done, but if you want to play against it, the bot's latest version will be at http://owenjohnson.info/dev/defcon/strangebot.zip

I'll be devouring your code as much as I can once I have some free time. Code from a 'working' bot would probably help me out a bunch. Keep up the coding because winning vs the CPU is a big step, especially when you (other people out there) consider you're starting from scratch (no placement, no nothin').

Just an FYI for anyone making a bot: if your bot sends too many messages during 1 tick, it can cause other clients to become out of sync so bad they can't catch up again. Found that out the hard way.

same for WhiteBoard drawings. I for myself wrote a little wrapper. It will always just draw 30 lines, queue the rest for later and drop the oldest ones if it lacks more then 10secs.
Chat I send to console or a file instead of chat window, no speed concerns, copy and paste ability and the chat window is just terrible to scroll.

trickser wrote:same for WhiteBoard drawings. I for myself wrote a little wrapper. It will always just draw 30 lines, queue the rest for later and drop the oldest ones if it lacks more then 10secs.

Hmm, I may have to steal that idea for my Whiteboard Library.

The problem i was running in to was that during defcon 1 I updated all of my orders for every unit every tick. All the other clients dropped or went out of sync.

I rewrote the defcon 1 calls so it doesn't do that anymore.

Joshua does absolutely everything he ever does within long running tasks (see my multithreading library), so Joshua never hangs unless I issue too many commands without a Yield in there somewhere, which is nice and easy to fix.

poptart911 wrote:The problem i was running in to was that during defcon 1 I updated all of my orders for every unit every tick. All the other clients dropped or went out of sync.

I rewrote the defcon 1 calls so it doesn't do that anymore.

http://forums.introversion.co.uk/defcon/viewtopic.php?t=7217

Okay, so I had the bot put silos into clusters, and it seems that clusters result in much lower score than random placement. The ICMBs have a slightly higher hit rate, but the clusters seem to fail to protect my cities, and in about half of the test, the silos were all destroyed.

Is there a good place to put the cluster? Right now, it places the cluster near the mean long,lat of all cities.

The random silo bot beats the cluster silo bot every time.

I am very surprised at this, and I want to know what you all think.

EDIT: I fixed the navy for the most part. Still buggy though, but gets all the nukes out.

poptart911 wrote:Is there a good place to put the cluster?

Outside enemy radar range, preferably also far enough from enemy fleet placement zone. Try to place it to protect at least one big city (or several small).

As for lower protection result with cluster silos, more info needed. I personally don't see how default AI can destroy a cluster Can you upload a recording of a few matches for us to see what's happening?

A cluster should produce significantly better hitting results, not just slightly better, do you launch silos all at once or one by one/random?

poptart911 wrote:seem to fail to protect my cities

They should only protect the ones they are directly next to, all other cities are usuallu just left there since it's impossible to protect them. A cluster protects at least some cities, spread silos protect no cities at all.

I'll make a couple of pictures to show optimal cluster positions in various situations.

.