Everyone was ready for a nice relaxing day at the beach when they showed up for my CUFP tutorial…
So, after realizing how boring it would be teach thirty flavors of factorial to experienced functional programmers for four hours, I thought I’d spice things up a bit. AI is always fun and I do love ants, so why not a contest where attendees would write Ant AI to gather food? For a bit of additional motivation to get everyone over the post-lunch slump, I decided to give away one of my Visual Studio 2010 Ultimate Edition MSDN subscriptions as a prize.
In thinking about how I would run this contest, it occurred to me that the AI better be pretty frickin easy to write. Four hours is not a lot of time to both learn a language and write something cool in it. Below is a modified (to fit nicely on my blog) version of the Ant AI for the red colony above, which is what I provided as a baseline sample for participants to test against.
match me with
| HasFood
| HasMaxFood ->
match locations with
| HasUnownedFood cells
when not (here.HasPheromone me.Color) ->
DropPheromone (here, 100)
| NearHome homeCells ->
match homeCells with
| CanDrop dropCells -> DropFood dropCells.Head
| CantDrop -> Move homeCells.Head
| AwayFromHome allCells ->
if not homeDirectionCell.ContainsAnt
then Move homeDirectionCell
else match randomEmptyLocation locations with
| Some location -> Move location
| None -> Nothing
| HasNoFood ->
match locations with
| HasUnownedFood foodCells ->
TakeFood (maxFood foodCells)
| SmellsPheromones pheroCells ->
Move (maxPhero pheroCells)
| _ -> match randomEmptyLocation locations with
| Some location -> Move location
| None -> Nothing
We all had a great time. Most didn’t even take the given breaks other than to run out quickly to the bathroom or grab a coffee. When time finally ran out I was proud to have three finalists who stayed and watched their ants battle it out, one of whom completed the entire project in Mono/Linux (Mono on MacOS, not so much).
Paul’s rather clever ants did what they could to block the opponent’s nest while gathering food. Phil, despite having some delays getting going due to my lack of experience with mono, made an excellent showing. However, in the end little could be done to stem the tide of Frank Levine’s pheromone gradient following ants.
As runner up, Phil will be getting a fresh copy of Professional F# 2.0 as soon as they hit my doorstop. Frank won the day and took home his very own MSDN Ultimate subscription.
All of the tutorial feedback was great. To my surprise, those who had trouble getting my code running on their Macs seem to have had a fun time. For days afterward I received emails from people who kept working on the problem even after they went home from the conference!
Currently I’m working on putting together a larger version of this contest online. I hope to extend the model used in this to include combat, spawning new ants and maybe even decaying dead ants. Send me a message on twitter if you’re interested in helping to beta test the engine.
I want to thank all of the attendees for coming, having fun, and putting up with the various cross platform issues. I also want to thank Martin Logan and Anil Madhavapeddy for helping to make this session happen. Finally, I want to thank Paul Greene for reporting a behavior related bug he found afterward through experimentation. It’s fixed now in my local source, but not in the files you can sneakily grab from the Mono bug report above ;).
I hope all of you CUFP tutorialists will join me in the next iteration of this contest. It’s sure to be a blast.