I’ve just returned from the week-long CppCon 2014 in Bellevue, Washington. Here’s what I experienced.
I’ve absorbed a great deal from a variety of C++ developer conferences – CppNow, Going Native, C++ And Beyond – but always virtually, via video and webcast. This was an opportunity to jump into the thick of things and participate in person. With community heavyweights like Herb Sutter and Scott Meyers in attendance I knew the content would be stimulating and informative. (Honestly, the speaker list featured nearly every name in the “C++ royalty” that you could imagine. I smiled to myself seeing Bjarne Stroustrup standing in the registration line like he was just another attendee.) So when the conference’s early-bird admission opened in March, I eagerly sent in my hard-earned dollars and blocked off the week of September eighth on my calendar.
The continuing dissent and confusion about unit testing of private class methods surprises me.
The access specifier is much like your choice of software license: it exists to limit consumers’ actions, not to limit yours. A method’s access specifier is completely irrelevant to testing, and only describes what you want the consumer to use; any code that takes inputs and produces outputs, private or not, should be tested.
The opponents of private-method testing tend to argue in quasi-religious terms: that private methods are mere hidden implementation details; that users of the class will only care about the public API; that testing of private methods breaks encapsulation. A typical unhelpful “solution”: private methods should be put into a different class and made public there.
To argue against granular testing of private methods is to mean well while being thoroughly unhelpful. The purpose of testing is more than just to guarantee the viability of your public interface – it is also to examine the inner machinery and support routines of your class to ensure that they themselves function correctly for a spectrum of inputs and edge cases. The private implementation will contain non-trivial complexities that are more readily and precisely tested directly than via the public API.
I just released my Steering Behaviors package to Github, and an accompanying Gem to Rubygems.
If you’re building a game, you’ll want your game agents and characters to exhibit realistic motion. A standard way of doing this is with ‘steering behaviors’.
The seminal paper by Craig Reynolds established a core set of steering behaviors that could be utilized for a variety of common movement tasks. These include such behaviors as predictive pursuit, fleeing, arrival, and wandering. This Ruby library can accomplish many/most of those tasks for your Ruby / JRuby game.
The basic behaviors can be layered for more complicated and advanced behaviors, such as flocking and crowd movement.
Embellishments and expansions are planned, but this is working software you can use to drive your own game’s characters. (I’m using it in my own game programming.) The Github repo includes working graphical examples, and you can install the Gem for easier and more direct use in your own game.
Pull requests are enthusiastically encouraged.
I’ve just released to Github my working fuzzy logic module, Fuzzy Associative Memory.
A Fuzzy Associative Memory (FAM for short) is a Fuzzy Logic tool for decision making. It uses Fuzzy Sets to establish a set of rules that are linguistic in nature; examples might include:
- “If the room is a bit warm, turn the fan up a little bit”
- “If the orc’s hit points are a little low, retreat from the enemy”
- “If the ship is off course by a little bit, correct just a little to the right”
- “If the bird is much slower than the flock, speed it up a lot”
As you can see, the rules are deliberately vague and use qualifiers like “a little” and “a lot”. This is the nature of fuzzy sets; they capture such human fuzziness in a way that extracts highly natural behavior from the fuzzy rules.
It has a wide range of applications:
- Industrial control, such as governing a fan to keep a room at the “just right” temperature
- Game AI, such as giving human-like behavior capabilities to NPCs
- Prediction systems
This is working, functional software. It currently supports:
- Triangular fuzzy sets for input/output
- Larsen Implication (scaling)
- Atomic antecedent propositions (
if A then Z)
- Trapezoidal (and other shapes) for fuzzy sets
- Hedges (‘very’ and ‘fairly’)
- Mamdani Implication (clipping)
- Composite antecedent propositions (
if A or B, then Z)
- Additional examples
bin directory contains the following examples:
hvac_system_example illustrates how a FAM could govern an HVAC fan unit to maintain a constant, comfortable temperature
Get it here!
Our Lunar Lander game is somewhat playable by this point but it still lacks some key features. After all, it would be nice if we could detect collisions and determine if the lander has safely landed on the pad. Let’s see how our flexible Entity-Component system permits us to expand our game with minimal fuss.
First, a frank disclaimer: the following collision detection algorithm is entirely inefficient. It’s kept simple for our basic teaching purposes here but is probably undesirable in a game of any scale. But that’s OK: E-C will permit you to swap in a much more advanced collision detection system when you’re ready. 🙂
Read the rest of this story…
Entity-Component systems, we’ve learned, are easy to implement and maintain; the elegance is basically “baked in” due to the way components and entities are married in the Entity Manager.
One particularly tidy aspect of an entity-component system is how well it lends itself to data persistence, or in practical terms: saving game state. Let’s take a look.
Where Is State?
In a conventional object oriented design, state is scattered all over the place, embedded in your far-flung object instances. But in E-C everything is neatly gathered together under one roof: the Entity Manager. This manager knows every entity “instance” along with every entity’s components, which are where the entity state data are stored.
Therefore, persist the entity manager to disk and you’ve saved the game in its entirety. Load from disk to memory and you’ve just loaded the game. It really is that easy.
Read the rest of this story…
Now that we have laid the entity-component foundation and introduced some necessary libGDX concepts, we can finally get around to putting together a little game. Let’s make a “Lunar Lander” type game to illustrate all the concepts we’ve learned so far.
Remember that the source code for this Entity-Component Framework and the game we’re writing is all available at Github. The Github version is, of course, the “final version” that includes features I might not have addressed so far in the blog series, but if you’re eager to jump ahead…
Entities and Components
For this exercise let’s begin by defining our entities and some of their relevant components.
What are we going to need for this game?
- The lunar lander module
- A platform for it to try to land on
- Ground to collide with
That seems like a fair assessment of our initial entity needs. Now, moving on to components. What are some of the aspects / behaviors / features that we should provide to our entities?
Read the rest of this story…