|Failed to load latest commit information.|
-- RMU SESSION #1 | PROBLEM 2 -- Submissions due 12:00 UTC 2010.09.15 If in doubt about how to submit, see SUBMISSION_GUIDELINES file. This week, we'll be experimenting with building a git themed achievements system. The emphasis of the exercise is on data modeling and requirements discovery, so you'll need to ask questions to help get the job done. Imagine you have just been hired by Github, and you have an idea for a brand new feature to make working with git more fun. You've decided it's time that Github embraces game mechanics by rewarding users with frivolous badges for every imaginable action under the sun. You made your first commit, you get a badge! You made ten commits? A bigger, shiny badge for you! You've got 50 forks of your project? Well, then you win the Bad Mother Forker badge. While implementing achievements may seem simple, there are a number of things to consider about performance. The more badges you have, the more conditions you need to check each time an event occurs that could cause a new achievement to be attained. That means some optimization is needed. Here are some ideas to get you started: * Certain achievements are dependent on others, and their conditions do not need to be checked until the badge they depend on is attained. For example, if you have badges for 1 commit, 20 commits, and 50 commits, you don't need to run the check for the 50 commit badge until both 20 commits and 1 commit have been attained. * Most achievements, once gained, are valid forever. For example, a user who has made 50 commits will never have made less than 50 commits, so the condition does not need to be checked any more once the badge has been attained. * However, certain badges *are* dynamic, and can be lost if certain events occurs. For example, if a user has a project that is in the top 10% for number of watchers, this can change over time. But these should be treated as special cases rather than evaluating all conditions by default. Your goal is to come up with a system to model the range of possibilities you might run into if you were creating these sorts of achievements for Github. You can stub out the source data, but you should build a running prototype through which you can define a list of achievements and determine whether a user has earned them or not based on their raw stats. This exercise is pretty open ended, but please follow the guidelines listed below. == GUIDELINES * You are encouraged to discuss this exercise both on IRC and the mailing list, and it's okay to share details about the general strategies you'll be using, as long as you're not giving away every last implementation detail. (I.e, discussion should mainly be done in words, and not code) * The stub data you'll create is open ended, but should be enough to demonstrate various kinds of conditions I've mentioned in the project description, and possibly some others you've come up with. * Your submissions should include runnable examples which demonstrate how your system works. As a prototype, it's okay if some features are missing, but make sure that whatever you submit is cleaned up and relatively self-contained. * There are likely some undiscovered requirements here, and I will be happy to clarify things as they come up. Treat me as the 'customer' for the purposes of this exercise. == QUESTIONS? Hit up the mailing list or IRC.