Example that passes:

1
2
3

assert_random do
    rand(1000)
end

Example that fails:

1
2
3

assert_random do
    1000
end

Basic stuff but I plan to extend it in the future so something like this will fail:

1
2
3
4

counter = 1
assert_random do
    counter += 1
end

I used the gem Jeweler. I got up and running very quickly and was able to focus on the functionality of the gem and not all of the meta details a gem needs. All that was generated for me. Awesome really.

Source: http://github.com/stewartmatheson/assert-random
RubyGem: http://rubygems.org/gems/assert-random

Bookmark It

This weakness was exposed today when I was trying to write an acts_as_method. I am currently working on a rule based AI system and I wanted to separate the underlying code of the AI system from the rails models that provide the knowledge. Thats where the acts_as_method comes in. However it came in rather slowly.

My first hurdle was looking at the require vs include. Both sound the same but do different things. Its important to understand this. The require opens reads and evaluates the code with in a ruby file. So for example say I have a file called Person.rb. In this file there was a class called “Person”. With me so far? I am working out of a file next to Person.rb called test.rb. Test.rb needs to know about Person.rb so that the Person class can be used. To do this we simply use

require 'Person.rb'

Simple enough. Ruby also supports shortcuts and will auto append the .rb so

require 'Person'

will also work. Again simple enough however all of a sudden it looks like we are talking about a class or a module here. This is what threw me a bit at first but remember it is the file. Directories work as you would expect. Say our person.rb is now in a models directory. Just use

require 'models/person'

Again no .rb but remember it is the file we are talking about. That brings us to include. Include in c++ for example includes a file the the way require works as described above but this is not the case in ruby.

In ruby we have mixins that allow use to add a module full of methods to any class we like. This is great and very handy as ruby does not support multiple inheritance. But who cares. Mixins are simple to use and make more sense. So lets consider the following

class Person
  def say_goodbye
    "goodbye"
  end
end
 
module PersonMethods
  def say_hello
    "hello"
  end
end
 
Person.include(PersonMethods)

Here we have a person class. We have a module with a method and we are using the include method to add the module methods to the person class. Now the person class will have both the say hello and say goodbye methods. The include is working on the program level. Not the file level. Very different to #include in c++ for example. I think form memory PHP has something smilier.

Who cares? Why do I need to do this? Why not just put both methods in the person class. Well you could do that and in this case its fine. However think about developing a Rails application for example.  You might have some logic that you want to use in a number of Rails applications. Cuting and pasting code from one model to another is one way to it but what about the following?

ActiveRecord::Base.send :include, PersonMethods

This is one of my favorite things about ruby. You can add in methods to classes that have already been declared else where.  You can add them when you need and you can break up your methods in different ways and add them at different times. All of a sudden your writing extra features in to rails with out even touching the source. Well the rails models at least.

This does not just stop at rails. You can apply this to any class anywhere. You can even apply it to core ruby classes. You can change the way ruby works at runtime! Check out the ruby homepage for lots more info.

Bookmark It

Events are a vital part of developing with ActionScript 3. Events are a very flexible way to make your code moduler and loosely coupled. Flush supports a number of events prepackaged. These include mouseclick events and keyboard events.

Whilst these are important for making any kind of user interface its custom events can I am going to explain how to use today. In the example I set up custom event for a unit within a strategy game I am developing. Think of civilisation and moving units around on a map.

When creating a custom event the first thing to do is write the custom event class. The custom event class must extend the event class which can be found in the flash.events package.

In this example the unit event class covers all events that relate to units. Obviously. Specifically the events I wish to send include movement, when the unit is selected, when the unit is deselected and when the unit has finished its movement.

In the class there is a static string declared representing each of these event types. The event model in action script uses this string to decide what listeners to call. Therefore this string is needed when adding an event listener.

In the class is the constructor. The constructor accepts one variable which is this event string. Make sure when you declare the constructor that the super call with the command string is present. This is all ActionScript needs to handle the event.

You may require to send additional information with the event. I have done this by using get and set methods. As the event is simply a class there are no limits on what functionality you can implement within the event itself.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19

package com.events {
 
	import flash.events.Event
	import com.geom.MapPoint;
 
	public class UnitEvent extends Event {
		public static const MOVEMENT:String = "movement";
		public static const SELECTED:String = "selected";
		public static const DESELECTED:String = "deselected";
		public static const MOVEMENT_COMPLETE:String = "movement_complete";
		private var targetLocation:MapPoint;
		private var startLocation:MapPoint;
		public function UnitEvent(command:String) { super(command); }
		public function getTargetLocation():MapPoint { return targetLocation; }
		public function setTargetLocation(t:MapPoint):void { targetLocation = t; }
		public function getStartLocation():MapPoint { return startLocation; }
		public function setStartLocation(t:MapPoint):void { startLocation = t; }
	}
}

Once we have the event class written we a free to dispatch it. Despatching an event can be done from any class that extends IDispatcher. IDispatcher is built into ActionScript.

The following example is an excert from a class that handles the unit. The code creates a new unit event. The new unit event takes one constructor which is the movement complete static variable of the event itself. This calls any listeners that are subscribed to that particular event type. After the locations are set the dispatchEvent method is called. This takes one parameter which is the event instance.

1
2
3
4
5
6
7
8
9

if(movesLeftThisTurn <= 0)
{
	var movementCompleteEvent:UnitEvent = new UnitEvent(UnitEvent.MOVEMENT_COMPLETE);
	movementCompleteEvent.setStartLocation(startLocation);
	movementCompleteEvent.setTargetLocation(currentLocation);
	startLocation = m;
	movesLeftThisTurn = movement;
	dispatchEvent(movementCompleteEvent);
}

So now we have a custom event class, we have code that dispatches the event however we do not have any way of listening for the event. A listener is set up with an addEventListener method. This method takes two parameters.

The first parameter is the string representation of the event. Remember when we first created the instance of the event? We passed it a single string parameter. When we add an event listener that listener will only be called if the event is created with that particular string parameter. This allows us to create different types of events in a single class. It also allows us to blind behaviours for different events. A very powerful system!

The second parameter that this method takes is a string representation of the function that the event is calling. Make sure to omit the brackets at the end of the function as we dont want to call function we simply want to tell the listener about it.

1
2
3
4
5
6
7

function unitFactory():Unit
{
	var u:Unit = new Unit():
	u.addEventListener(UnitEvent.MOVEMENT, unitView.moved);
	u.addEventListener(UnitEvent.MOVEMENT, activeState.selectNextUnit);
	return u;
}

The final step is to add the actual listener. The listener is a function in itself. The listener accepts one variable. The event that you have passed. This was the event that was sent in the dispatchEvent method. Make sure that every event listener returns void. If the event listener does not have a return type void then the event will not be fired properly.

1
2
3
4
5
6
7
8

class UnitView extends View {
	...
	public function moved(e:UnitEvent):void
	{
		x  = e.getTargetLocation().xpos;
		y  = e.getTargetLocation().ypos;
	}
}

In the above event listener we update the units position as this event listener is fired when the unit is moved.

I hope this helps. Please give me feedback. If there is any other topics you would like me to blog about please don’t hesitate to ask.

Bookmark It

A very rough draft of what the first map in the game might look like.

I mentioned in my last post that started work on a hush-hush project. When I say hush-hush I really mean a project in which the details have not quite been refined. I can now reveal that the project I’m working on is a Flash-based Web game. The game is going to be a turn-based strategy game. The game will feature an isometric view with tiles. Here’s my first rough version of the tiles that I want to include in the game. The colours are not quite right and there needs to be more variety however it is a good starting point. Don’t be too rough on me. I’m not really a designer.

I can also tell you that this game will feature an artificial intelligence system that I have not seen anywhere before. Instead of the artificial intelligence been coded into the game the artificial intelligence is housed within a Web application. The game requests back to the Web application asking the new artificial intelligence instructions. This is my dissertation for my masters. It’s my final research peace. A system like this probably doesn’t have any place in a commercial title. What I am really hoping to achieve is a proof of concept. Can a single artificial intelligence system operate across a distributed player network? I’m not sure but stick around if you would like to see the results.

Bookmark It

I decided to cut my losses and go to Ruby 1.9.1. Again I ran into trouble. This time up problems came when I was attempting to save records into the database. This error is known to the community and I got the impression that it was the version of Ruby I was using.

In the end I decided to cut my losses and go back to Ruby 1.8.7. As Ruby 1.8.7 has proved time and time again its the version to go with. Everything is working now and I am currently using rails3.

If you’re using a Mac like I am I would recommend compiling a fresh version of Ruby from source. This way you don’t interfere with your previous Ruby installation. If you need to run multiple versions of Ruby I suggest you check out Ruby version manager.

I am very excited about a lot of the changes in rails three. Full credit to David and the team for acknowledging that there still is quite a hefty barrier of entry for new users. The bunder gem is a welcome addition. I think this will ease some of the pain for new users. Especially new users joining an existing rails project.  In addition there have been several refinements in the API. The routing API has been considerably refined and it’s a much cleaner interface.

I can’t wait to get moving. I got some work to do today but once I get that finished in going to start specing out my first rails3 application. WOO!

Bookmark It

There are some problems. The HTML does not validate completely however this is due to the plug-ins I am using. I noticed that all of the code causing breaking HTML validation is outputted by the various plug-ins I have installed. No bother the page seems to render okay. That said I have not tested in Internet Explorer yet.  I’m not sure about the links on the left hand side. The event model is not working correctly sometimes the links jut up and down. I might disable the JavaScript if it gets really annoying.

I have also added my latest twitter status to the page. I love twitter and use it constantly. It will be good to have my status there. At least something on this blog will get regularly updated . My use of twitter has increased greatly over the last few months and I have noticed a distinct lack of any decent twitter client for Windows. That’s why very shortly I will be posting my own twitter client that quite frankly leaves all of the other twitter clients I have used on Windows in its wake. It doesn’t say much about existing social applications on Windows as it only took me two or three days to write.

With the new theme for my blog and a completed twitter client I can now focus on my dissertation. I will not have much time to post on this blog. I hopefully however will put my dissertation online and make it available for download. As many of the details are not finalised I will not go into them yet.  I can tell you I plan to build a web server that encapsulates an artificial intelligence algorithm to serve out artificial intelligence as a normal web server would serve documents. Hopefully that wets your appetite enough to come back to this space for updates in the future.

Bookmark It

http://mirror.lividpenguin.com/pub/apache/httpd/httpd-2.2.15.tar.gz

http://subversion.tigris.org/downloads/subversion-1.6.9.tar.gz

http://nginx.org/download/nginx-0.7.65.tar.gz

Here are the apache configurations I used

<IfModule mpm_prefork_module>
    StartServers          1
    MinSpareServers       1
    MaxSpareServers       1
    MaxClients           10
    MaxRequestsPerChild   0
</IfModule>
 
<Location /repos>
    DAV svn
    SVNPath /usr/local/nginx_svn_test/testrepos
    SVNAutoversioning on
    Options Indexes FollowSymLinks
    Order allow,deny
    Allow from all
</Location>

This is the Nginx proxy pass configuration

location / {
  proxy_pass http://localhost:86;
}

The following is the line that configures the apache compile.

./configure --enable-dav --enable-dav-fs --prefix=/usr/local/nginx_svn_test --disable-alias --disable-userdir --disable-actions --disable-dir --disable-negotiation --disable-cgi --disable-cgid --disable-status      --disable-autoindex --disable-asis --disable-mime --disable-version --disable-setenvif --disable-env --disable-log-config --disable-include --disable-filter --disable-auth-basic --disable-authz-default --disable-authz-user --disable-authz-groupfile --disable-authz-host --disable-authn-default --disable-authn-file

The following is the configure line for the SVN install.

./configure --prefix=/usr/local/nginx_apache_svn_test --with-apxs=/usr/local/nginx_apache_svn_test/bin/apxs

Bookmark It

One of the things I have come to love about C++ is the flexibility. You can code C++ in virtually any style you want. It supports a higher level object oriented design but you can also get down and fiddle with bits if you need to. At the moment I’m interested in the high-level features of the language, the low-level features can wait.

One thing I have been working a lot on is the understanding of programming patterns in relation to C++. For the kind of work I am doing at the moment understanding programming patterns is almost as important as understanding the language itself. You can code in an ad-hoc way however it’s when you start to introduce programming patterns when things start to make sense and become… dare I say fun. Just look at the impact Ruby on Rails had to the Web development world. In a sea of ad hoc development one framework comes along that implements one gigantic programming pattern and it’s hailed as the second coming of Christ.

One of the most important programming patterns is the subject observer pattern. This is what I am detailing in the post. Subject observer fits in with model view controller. Basically a change to the model occurs, the model notifies the views on the subject and the pattern is born. In this particular example the subject is the model and the Observer are the views. It doesn’t have to be particular to any other class and the example I give allows you to inherit from the subject observer classes and implement them in any way you want. One of the goals with this exercise was to keep the solution is generic as possible.

The subject observer pattern is powerful because it allows you to loosely couple your views and your models. Who doesn’t love that. In C++ terms the model contains a vector pointers to its observers. The model is dumb in this sense. It doesn’t know who its observers are and nor does it care. Observers can be added and removed from the model at any time as long as some sort of a array/vector/list of the observers is maintained. Note that this “model” solution does not include away to detach subjects from observers. It should not be too difficult to work this out but if you have problems ask me.

The following example has been tested with the GNU C++ compiler. In this case I am running it on a Macintosh. The code should be fairly generic and you should be run at different compilers. If you can’t let me know.

In the code the horn class observes the alarm class. Thus making the alarm class the subject. Note how the horn and alarm class inherit from observer and subject respectively. The alarm gets triggered by the yes you guessed it triggered method. This it is an update to the model and as a result calls the notify observer method. This method which is inherited from subject loops through a vector of observer pointers calling the notify method on each observer. The notify method is a member of observer and inherited in the Horn class. The notify method gets passed a subject. This is so the notified object can extract information from the subject. While some solutions may not require the notified object to obtain any data from the subject the solution allows the Observer complete access to the subject.

Note in the horn class that there is a typecast. This typecast converts the subject pointer and alarm pointer allowing methods on the alarm to be called from the Horn class. In this case the Horn class fetches the ID of the alarm but it could be any kind of data.

I hope this explanation of the subject observer pattern helps. If you have any questions please post a reply to this comment. I will try to answer questions to the best of my knowledge bearing in mind that I am new to C++ myself.

Anyway time for code.

#include <vector>
#include <typeinfo>
#include <string>
 
class Subject;
 
class Observer
{
public:
    virtual void notify(Subject* s) = 0;
    virtual ~Observer() {};
};
 
class Subject
{
    std::vector observers;
protected:
    void notify_observers()
    {
        std::vector::iterator iter;
        for (iter = observers.begin(); iter != observers.end(); ++iter)
            (*iter)->notify(this);
    }
 
public:
    virtual ~Subject() {};
    void register_observer(Observer* o)
    {
        observers.push_back(o);
    }
};
 
class Alarm : public Subject
{
public:
    Alarm()
    {
        std::cout << "alarm created" << "\n";
    }
 
    void triggerd()
    {
        std::cout << "The alarm has been triggerd" << "\n";
        notify_observers();
    }
 
    int const get_alarm_id(){ return 100; }
};
 
class Horn : public Observer
{
public:
    virtual void notify(Subject* s)
    {
        Alarm *a;
        a = dynamic_cast<Alarm*>(s);
        std::cout << a->get_alarm_id() << "\n";
    }
};
 
int main ()
{
    Alarm a = Alarm();
    Horn h = Horn();
    a.register_observer(&h);
    a.triggerd();
    return 0;
}

Bookmark It

Box 2-D

Box 2-D is a fantastic physics library. I’ve been programming quite a lot using box 2-D and I really enjoy it. Box 2-D supports joints, bodies and the number of various other components that can be used to model real worlds. It’s written in C++ but there is also a port available in action script. There are a number of flash games that are using this port to various levels of success. The possibilities are endless and I am sure I can find a way to incorporate this library into some Web work or to use it to animate an interface that is not necessarily game orientated. Something I would have never considered before starting this course.

Simple fast multimedia library

The simple fast multimedia library is a tool for producing desktop games. It supports Linux, Macintosh OS X and Windows. It can produce both 2-D and 3-D games. It produces 3-D games by providing interfaces to OpenGL classes and methods. It produces 2-D games by providing classes that handle the displaying and updating of sprites.  It’s also got classes for sound and controller imports. There’s a handy integration project for Xcode so you can develop easily on Macintosh OS X if you choose to. I’m pretty sure the same exists for visual studio on Windows but I am not 100% sure. A welcome change as there is a lot of computer game engines that specifically run on Windows. Not that I would hold a grudge against any gamer who are solely interested in Windows. Let’s be honest there’s not a whole lot going on around Macintosh in the gaming industry.

Microsoft XNA

XNA is a platform developed by Microsoft. It supports both Windows games and Xbox games. It installs itself as an extension to visual studio. It’s completely free and anyone can download the express edition of visual studio as well as the Microsoft XNA installer. The really cool thing is that you don’t have to pay to start developing games. You are however limited to C# Microsoft’s proprietary implementation of Java. There is quite an extensive website for an and it’s very easy to publish to the Xbox live Arcade once you have finished developing a game. As is the custom with Microsoft, XNA almost in its own world ignoring other technologies and platforms out there. It uses exclusively Microsoft directX with no support for OpenGL. Again you can’t really begrudge Microsoft for wanting to support their own platform. It is however ironic when they say that XNA supports multiple platforms. Windows and the Xbox. Ironic.

There are so many other game technologies are available and so many other resources at the game developers disposal you feel spoilt for choice. Much like the Web there are different tools out there that are good at different things and it’s up to the developer to choose the appropriate tool for the job. I’m surprised because I thought more of the games industry was based around building up proprietary libraries and not really sharing a whole lot of code. Whilst the big players undoubtedly do this and so they should they have the budget to do so there is room for the Indy game developer. People in my position who want to get started in the industry but don’t have a lot or even any money to spend on licensing engines and technology. Whilst it’s hard to break into the gaming industry there are certainly plenty of tools around to help you do it. Really encouraging stuff.

Bookmark It