It’s been almost four years since I released the rats upon the unsuspecting world of Ruby on Rails. Since then, I’ve learned a lot about writing web applications – so I’ve been meaning to update this fun little game with some of the tricks I’ve learned. But, you know, it hasn’t been exactly a priority.
Then a few weeks ago a prospective client asked me about Django. I had never used it before, so prior to a conference call with them I downloaded it and worked through a couple of tutorials.
Python is not my favorite language, though it’s far from my least favorite. A lot about Python is really clean, but some of it just seems anal-retentive – especially the indentation rules. Oh, it’s consistent all right, but it just strikes me as a Simon Says Rule. I mean, c’mon – you can’t even use tabs instead of spaces? Between that and always forgetting the colons I end up wasting a lot of time on simple syntax errors. At least the interpreter gives helpful error messages.
The web development framework Django seems to be a study in powerful simplicity. It makes the usual things dead easy, and the less usual things still easy — just a little less so. A web project requires a lot less scaffolding to sort through than in Rails. The project/application generator writes a lot less code for you (almost none, in fact), not only because the design of the framework is simpler, but also because it relies more on inheritance and less on generated code. I have never lost track of where code lives in a Django project the way I used to forget where I put my Rails code.
As an exercise, I decided to rewrite the Rat Race in Django, which you can download below. This version also makes liberal use of JavaScript, especially the jQuery library (which I’ve included in the directory where the app expects to find it). If you try to run the race with JavaScript disabled in your browser, you’ll get a nice red and yellow banner telling you that you need it.
To try this app out, unzip the download wherever you like. Make sure you have Python and Django installed. I’d recommend using the latest versions (Python 2.6, Django 1.1.1), as I haven’t tested with any others. Go to the “ratrace” directory and start the included server:
manage.py runserver
Now you can bring up a browser (anything except Internet Explorer) and navigate to http://127.0.0.1:8000/ — which should start the next race. I’ve included a sqlite3 database containing 72 rats. You can administer this database by navigating to http://127.0.0.1:8000/admin/ – and if you don’t like my rats, you can just delete the database (rats.sqlite), regenerate it:
manage.py syncdb
and enter your own. You must have at least three rats to run the race.
In this version of the rat race (back story here), I use the Ajax and JSON capabilities of jQuery and Django to drive the race. jQuery’s animation lets me smooth the rat movements so you don’t get quite the jumpy effect of my earlier versions, and it allows me to ping the server for updates a little less frequently. You can run multiple browsers pointed at the same server and get essentially the same race except for unimportant rat gestures. I’ve tested it with the latest versions of Chrome, Firefox, Safari, Opera (some minor display issues) and Internet Explorer (no worky, not sweating it).
As in all earlier versions, each rat’s likelihood of moving forward or backwards in the race is controlled by a random number generator that’s biased according to their track record. So a rat that has done well in the past is more likely to do well in the future. But that isn’t set in stone, and I’ve introduced another random factor that can’t be seen by the user. Instead of basing the odds strictly on that bias, I now simulate between $200 and $10,000 in bets of $2, $5, or $10 each. Each bet is randomly chosen, but skewed according to the rat’s track record and an additional random factor. This is intended to simulate a hunch, a whim, or individual knowledge that may or may not be in line with the additional random factor controlling the rat’s performance in that particular race. The result is, I think, fairly close to true life experience – in which the best rat usually does well but doesn’t always win, while the poorer rats hardly ever place but sometimes surprise everyone.
Some interesting coincidences with real life fell out of algorithmic decisions I made. In order to avoid divide-by-zero errors, I added one to the number of races run and then offset that by a random number. This has the result of making the track record count more for more experienced rats, and conversely newer rats are less predictable.
I placed a chunk of cheese at the end of the race as an incentive for the rats. When they reach it, they appear to eat it by overlapping its image. At that point, I change the motion algorithm to avoid backing up (and thus regurgitating the cheese), and also slow it down a little. This increases the excitement at the end of the race, because even if a rat reaches the cheese first, another rat may still be able to come up and polish off his/her chunk of cheddar to win. The transition from the normal speed down to the eating phase also improves the favoring of rats with a better bias, so the rats who got near the end on sheer luck may not be able to hold out all the way to reaching the cheese.
The only thing I might add in the future is the ability to track bets from the user.
Enjoy the races. For background music, may I recommend the works of Django’s namesake?
