Wednesday, September 8, 2010

Beans Should Be Deprecated - The Future of Object Architecture With Properties (Part 2 of 2)

In part 1 we discussed the basics of properties in Sgine, how to listen for events, what a delegate property is, how to filter properties, transactions with properties, and animation. In this part we will conclude our tutorial of properties in Sgine with a discussion of paths in properties, binding support, property containers, dependent property, and advanced property.

Path-based Properties


In Sgine there is a unique path system. However, this is not a typical path system because it represents object paths, not filesystem paths. In the org.sgine.path package there is a class called OPath (object path) that can represent the hierarchical path to an object. So, for example, take the following code:
import org.sgine.property.MutableProperty

class Person {
    val addresses = new MutableProperty[Addresses]
}

class Addresses {
    val billing = new MutableProperty[Address]
    val shipping = new MutableProperty[Address]
}

class Address {
    val line1 = new MutableProperty[String]
}

Now, presume I have a reference to an instance of a person object, and I want to listen for any changes to address line1 on the shipping address. This is difficult as "addresses" is mutable, "shipping" is mutable, and "line1" is mutable. So, if I do the following:
person.addresses().shipping().line1.listeners += shippingLine1Changed _
I may get a NullPointerException because something isn't initialized, or I may connect up, but when any one of the mutable chains in-between break I'm now referencing the wrong data. This is where OPath comes in. I can pass a root reference and a path as a String and it can resolve the underlying object and return Option[T]. So for example, if I want to resolve the value as above with an OPath I can invoke:
val o = OPath.resolve[String](person, "addresses().shipping().line1()")
The value of "o" will be Some[String] representing value of line1 or None if any point in the hierarchy is not able to be resolved.

Now, back to properties. The PathProperty works with OPath instances (calling resolve like above simply resolves the value, but an OPath instance is a long-term reference to that path that receives events when anything changes). So if we wanted to create a property and know when the value of line1 in the shipping address has changed we could do the following:
val shippingLine1 = new PathProperty[String](OPath(person, "addresses().shipping().line1()"))
shippingLine1.listeners += shippingLine1Changed _

Your listener would look like this:
private def shippingLine1Changed(evt: PropertyChangeEvent[String]) = {
    ... shipping line1 has changed ...
}

Binding


I mentioned earlier that events were probably the most used feature in properties, but binding is likely a close second. The premise of binding is fairly simple. You have two values and you want one or both to replicate the value of the other. Newer frameworks like Flex and JavaFX make heavy use of binding and do a decent job of it. However, there's a lot that Sgine provides that they have no ability to do. Alas, I'm getting ahead of myself. Let's first take a look at a simple example. Consider the standard UI usage of binding. We have a CRUD app that manages people, so we have a text field that when updated needs to modify the name of the specific person we are working with. We must first modify our Person class to support binding:
import org.sgine.property.BindingProperty
import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String] with BindingProperty[String]
}
Now, presuming we have a text field with signature like the following:
class TextField {
    val text = new MutableProperty[String] with BindingProperty[String]
}
We can simply take our Person's name and bind it to the text field's text property:
person.name bind nameField.text
Now any changes that occur to the nameField will automatically be applied back to person.name. This is the simple case and pretty much the extent to which most binding systems function. However, as I stated before, in Sgine binding support goes well beyond that.

The first problem you often run into with standard binding functionality is that you can only bind the same type of properties together. For example, in the previous case we were binding the name of a person to the text value of a text field. These both were String representations, but what if we wanted to do something like have an "age" property that is bound to a "birthday" property? We have an Int that wants to link up to a Calendar? This brings us to translation bindings.

Translation Binding


Translation binding is very similar to standard binding except there is a translation step that must occur to get the value ready for use in the target.

Lets look at the age / birthday example:
import java.util.Calendar

import org.sgine.property.BindingProperty
import org.sgine.property.MutableProperty

class Person {
    val age = new MutableProperty[Int] with BindingProperty[Int]
    val birthday = new MutableProperty[Calendar] with BindingProperty[Calendar]

    age bind(birthday, determineAge)

    private def determineAge(c: Calendar) = {
     if (c != null) {
         val now = Calendar.getInstance
         val age = now.get(Calendar.YEAR) - c.get(Calendar.YEAR)
         now.set(Calendar.YEAR, c.get(Calendar.YEAR))
         if (now before c) {
             age - 1
         } else {
             age
         }
     } else {
      -1
     }
    }
}
We simply bind birthday to age, but at the same time provide a function to convert from a Calendar to an Int. Now every time birthday changes, age gets updated.

Path-based Binding


As we've already discussed, OPath in Sgine represents an object-path to a value within a hierarchical structure. We've already seen it used to resolve values directly and with PathProperty. Now we're going to look at one of the most powerful uses of OPath, and that's for binding. Often you want to bind to a hierarchical value that may or may not exist at time of instantiation. The solution to this generally is to add a listener, to wait until it has been defined, and then add the binding. Later though, the reference structure may change or be set back to null and you find yourself with a binding to something that is no longer being used. With path-based bindings you can bind to a hierarchical structure and no longer concern yourself with the details of parts changing in-between. Every time the path changes the binding will be updated.

For example:
import org.sgine.path.OPath

import org.sgine.property.BindingProperty
import org.sgine.property.MutableProperty

class Person {
    val lastName = new MutableProperty[String] with BindingProperty[String]
    val spouse = new MutableProperty[Person] with BindingProperty[Person]
}
Now, consider that this person has stated they want their lastName to be bound to that of their spouse's. We could do this:
person.lastName bind person.spouse().lastName
But it's possible they are not yet married, and it's also possible that the spouse might change. If we do a path-based binding instead, no matter if they are married now or change spouses in the future, the binding will remain pointing to their spouse's lastName:
person.lastName bindPath OPath(person, "spouse().lastName()")
Though a little hard to grasp at first, the benefits of such functionality significantly reduce the amount of handling that must be added in order to deal with multi-level mutability.

Property Container


Properties by themselves are quite useful little beasts, but add PropertyContainer to this mix and the benefits multiply. One of the troubles you have with the standard bean scenario is introspecting the contents of that bean in order to find out what values it contains in order to make use of it. Inevitably Reflection is used to determine the fields or methods that make up that bean. PropertyContainer eliminates the need for manual Reflection.

The ability to know what properties exist for a specific instance can be very useful. Rather than pushing back the necessity to use Reflection onto the developer, PropertyContainer dynamically introspects the instance in order to determine what properties are available and by what name they can be referenced. For example:
import org.sgine.property.MutableProperty
import org.sgine.property.container.PropertyContainer

class Person extends PropertyContainer {
    val name = new MutableProperty[String]
}
Now, I can iterate over all the properties of my "person" with:
for (prop <- person.properties) {
    ...
}
I can also do a lookup of a property by name:
val name = person("name")
Another very useful feature to PropertyContainers is that they are also Listenable instances, so they can receive events. This gives you the ability to listen for events on the container and receive events for the children. See the following example:
import org.sgine.property.ListenableProperty
import org.sgine.property.MutableProperty
import org.sgine.property.container.PropertyContainer

class Person extends PropertyContainer {
    val name = new MutableProperty[String] with ListenableProperty[String] {
        override val parent = Person.this
    }
}
It's important to notice here the only revision that needed take place is that "name" provide a reference to the Person instance via the "parent" property. This is used to propagate the event up. This can allow a complete hierarchical structure to propagate all the way up to the top. Now we can add a listener with the following:
import org.sgine.event.EventHandler
import org.sgine.event.Recursion

import org.sgine.property.event.PropertyChangeEvent

val person = new Person()
person.listeners += EventHandler(propertyChanged, recursion = Recursion.Children)

private def propertyChanged(evt: PropertyChangeEvent[_]) = {
    ... property changed ...
}
The above example will be invoked if any child throws a PropertyChangeEvent. Finally, beyond the static PropertyContainer we also may need to add and remove properties dynamically. This is where MutablePropertyContainer comes in. MutablePropertyContainer, as the name suggests, allows dynamically adding and removing properties on the fly rather than relying on Reflection in order to determine names and references.

Dependent Properties

DependentProperty allows for the default value to point to another property instead of the initial value of that property. For example:
import org.sgine.property.DependentProperty
import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String] with DependentProperty[String] {
        val dependency = Person.defaultName
    }
}

object Person {
    val defaultName = new MutableProperty[String]("Anonymous")
}
The value of person.name is defaulted to point to whatever the value of Person.defaultName is. Even if the value of "defaultName" should change after initialization it will continue to point to it until the value of the "name" property itself has been changed. This is particularly useful for stylization as it allows default styles to exist but without removing the ability to explicitly override.

AdvancedProperty

In the examples above there is a lot of mixing of traits to achieve the desired functionality, but in most cases there is no reason not to make all of this functionality available. For simplification the AdvancedProperty combines all the standard functionality that might be desired in a property without sacrificing performance. The traits that AdvancedProperty mixes in by default are:
  • MutableProperty
  • DepedentProperty
  • ListenableProperty
  • NamedProperty
  • BindingProperty
  • AnimatingProperty
  • EventDelegationProperty
  • CombiningProperty
Simply instantiating AdvancedProperty is much easier than concerning yourself with every possibly trait that might be needed on a given property and in general there's nothing lost by doing so. AdvancedProperty does not contain all the available traits for properties, but additional traits can be mixed in as needed. To use AdvancedProperty is quite simple:
import org.sgine.property.AdvancedProperty

class Person {
    val name = new AdvancedProperty[String]("")
}
This gives you all the functionality you are likely to need in "name". If you are taking advantage of PropertyContainer you can use the following signature instead:
import org.sgine.property.AdvancedProperty
import org.sgine.property.container.PropertyContainer

class Person extends PropertyContainer {
    val name = new AdvancedProperty[String]("", this)
}

Conclusion

This has been a lengthy tutorial about the functionality that is provided in Sgine's Property system, but hopefully provides the foundation necessary for you to start taking advantage of Properties in your projects. The focus of Sgine is very much toward 3d engine development, but much of the foundation is geared toward general usage. To that end it is my desire that functionality such as the Property system may be leveraged in diverse systems that may have nothing to do with graphical programming at all. Though this was a lengthy examination of properties it did not go over every feature and trait available in Sgine. For additional information please refer to the ScalaDocs for Sgine.

Beans Should Be Deprecated - The Future of Object Architecture With Properties (Part 1 of 2)

In my previous post I discussed the case for Properties and the basics of how they work. In this post I will dive into the specific workings of Properties in Sgine and how to use the various features it provides.

This post is all about Properties in Sgine for use primarily with Scala. However, I know many people, for whatever reason, are still using Java. For those people I would like to point you to this post I made a year previous regarding Properties in Java and the implementation of Properties in the open-source project xjava.

Basics


In Sgine Property is a trait that extends Function0[T] (getter) and Function1[T, Property[T]] (setter). This basic foundation allows better compliance of use along with functional delegation. Here's a simple example class implementation of a Property:

import org.sgine.property.Property

import scala.reflect.Manifest

class SimpleProperty[T](protected implicit val manifest: Manifest[T]) extends Property[T] {
    private var _value: T = _

    def apply() = _value

    def apply(value: T): Property[T] = {
        _value = value

        this
    }
}

This is relatively straight-forward to the way properties were discussed in the previous article with one glaring difference. The use of Manifest is a very cool feature in Scala that Sgine takes extensive use of. If you're not familiar with Manifest I suggest reading more here. If you are familiar with erasure Generics in Java you'll know that Generic type information is not available at runtime, it is purely a compile-time feature to maintain static-typing. However, there are many cases where having that information at runtime is extremely useful and that's exactly what scala.reflect.Manifest gives you. Fortunately though, the only case in which this knowledge will be necessary is if you need to extend Properties, which will never be necessary for the majority of developers.

Now that we have our SimpleProperty class we want to use it:

class Person {
    val name = new SimpleProperty[String]()
}

val person = new Person()
person.name("John Doe")
val name = person.name()

Now that we understand how easy it is to create properties, lets put that knowledge aside for now. As I stated, you'll probably never need to do this, but hopefully this helps you understand how easy it is to do if you should desire to do so.

Moving on, Sgine already supplies all the functionality demonstrated in SimpleProperty in org.sgine.property.MutableProperty. This class is the base for most of the standard mutable Property functionality in Sgine.

The above example requires the creation of your own Property class, but you can use the built-in MutableProperty:

import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String]()
}

val person = new Person()
person.name("John Doe")
val name = person.name()

Because setting name via apply:
person.name("John Doe")
Is not the most intuitive to realize it is setting the value of the Property, there are a couple additional convenience methods in Property that invoke the same functionality but may be more readable in code.

The first is ":=" so you can write:
person.name := "John Doe"
This is what I personally prefer when setting the value of a Property as it is quite clear that I am intentionally modifying the value of person.name.

Additionally you can use "value" for both getting and setting:
person.name.value = "John Doe"    // Setter
val name = person.name.value    // Getter

Usage aside, Properties are quite simple in concept and though every implementation has its own details you must understand, you hopefully have a good foundation to start with. In the next sections we will be building upon this foundation with specific functionality that Sgine provides to make Properties that much more powerful and effective.

Events


Events in properties are arguably the most useful feature in any properties framework and definitely the most requested feature to any properties system. Sgine provides a complete Event system that is general purpose apart from properties, but is utilized in the properties system to provide a simple, yet complete event system.

The relevant trait we'll use for event handling is ListenableProperty. Here's a sample of setup:
import org.sgine.property.ListenableProperty
import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String] with ListenableProperty[String]
}

Notice we simply had to mix-in the ListenableProperty to our MutableProperty and now we've got event support in our "name" property.

Here's an example to listen for change of "name":
val person = new Person()
person.name.listeners += nameChanged _

Notice here that we simply reference the "listeners" instance of "name" and invoke "+=" to add our listener.

The definition of our listener "nameChanged":
private def nameChanged(evt: PropertyChangeEvent[String]) = {
    println("Name Changed!")
}

Like I stated previously, the event is extremely powerful and can do some really awesome things, but for the purposes of this tutorial we'll keep it to simply mentioning that we can reference as our "listener" any method/function that takes an Event and returns Unit. In the above code we are explicitly wanting to receive PropertyChangeEvents for Strings and the internal event system will filter all other events from getting to our listener method.

Now every time we change "name" an event will be thrown to "nameChanged".

Another tutorial needs to be written in the future to explain the features and functionality provided in the event system, but you now have the basic knowledge to listen for events in properties.

Delegate Properties


DelegateProperty is an alternative base property to MutableProperty. Instead of maintaining a value internally within the property itself, it delegates to outside functions for getting and setting values. This is most useful when dealing with existing code. Consider the following example:
import org.sgine.property.DelegateProperty

class Window {
    private val frame = new javax.swing.JFrame()

    val title = new DelegateProperty[String](frame.getTitle _, frame.setTitle _)
}
This example is a simple wrapper around Swing's JFrame, and though we could create a title property, listen to changes, and then apply them to the underlying JFrame it would be quite a bit of effort (and a lot more code) to do so. Instead, since all we want to do is replicate the functionality from JFrame's getTitle/setTitle methods we can just delegate to and from those methods. This allows us to utilize all the features of properties while relying on existing functions.

This can also be useful for more complex scenarios where determining the value requires some internal processing. You simply specify your getter to point to a private method/function that does the work and you're finished. The setter function is an optional parameter and if unspecified will throw an UnsupportedOperationException if invoked.

Filtering


FilteredProperty provides the ability to modify an incoming value before it gets applied to the property itself. See the following example usage:
import org.sgine.property.FilteredProperty
import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String] with FilteredProperty[String] {
        val filter = (s: String) => if (s == null) "" else s
    }
}
This simple example converts nulls to a blank String. This allows you to work with "name" without ever being worried about its value being null.

Another usage of this would be to disallow specific values by throwing an Exception:
import org.sgine.property.FilteredProperty
import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String] with FilteredProperty[String] {
        val filter = (s: String) => if (s == null) throw new NullPointerException("This property does not allow null!") else s
    }
}
Now, this is creating boiler-plate code, and we don't like boiler-plate code. Lets do this in a more reusable way:
import org.sgine.property.FilteredProperty

trait NullsAreBadFilteredProperty[T] extends FilteredProperty[T] {
    val filter = (value: T) => {
        if (value == null) {
            throw new NullPointerException("This property does not allow null!")
        } else {
            value
        }
    }
}
Now I can simply mix-in my new trait wherever I want to disallow nulls:
import org.sgine.property.MutableProperty

class Person {
    val name = new MutableProperty[String] with NullsAreBadFilteredProperty[String]
}

Transactional Properties


TransactionalProperty is the property equivalent of a database transaction. When a the property is created it maintains a reference to its initial value and the current value. At any time the value can be committed in order to roll forward the initial value to the current value or it may be reverted in order to roll back the current value to the initial value. Finally, when a TransactionalProperty is committed a PropertyTransactionEvent is thrown for the property that represents the original value and the new value.
import org.sgine.property.MutableProperty
import org.sgine.property.TransactionalProperty

class Person {
    val name = new MutableProperty[String]("") with TransactionalProperty[String]
}

Simple usage:
val person = new Person()
person.name := "John Doe"
person.name.revert() // Reverts the value back to ""
person.name := "Jane Doe"
person.name.commit() // Commits "Jane Doe"

Animation


Animation may be a little confusing of a name here since we're dealing with properties and not images, but AnimatingProperty works with a PropertyAnimator to allow "animation" from the current value to the newly applied value over time. Obviously this is of primary benefit in the case of a user-interface, but can be applicable to other uses as well.

Here is a simple example to animate linearly from 0.0 to 5.0 over five seconds:
import org.sgine.property.AnimatingProperty
import org.sgine.property.MutableProperty
import org.sgine.property.animate.LinearNumericAnimator

class Location {
    val x = new MutableProperty[Double](0.0) with AnimatingProperty[Double]
    x.animator = new LinearNumericAnimator(1.0)    // Move 1.0 per second
}

val location = new Location()
location.x := 5.0    // Begin animation from 0.0 to 5.0
location.x.waitForTarget()    // Convenience method to wait for target value to be reached
We're using LinearNumericAnimator, a built-in animator to work with Double properties, to animate 1.0 per second. If you are familiar with Flex or JavaFX this is similar in purpose to what they call "Effects". However, in those frameworks an Effect is defined apart from the property being modified and is simply activated upon it. Though this means writing explicit code for every movement you wish to make it is perfectly acceptable most of the time. However, this does create the potential problem of multiple effects being enacted on the same value at the same time. For example, a box is displayed in the center of the screen with two buttons. The first button moves the box to the left side of the screen and the second button moves the box to the right side of the screen. This is perfectly fine until you try hitting one right after another. If you don't add explicit support to cancel the first effect you will run into what I call "effect fighting". This is the scenario where two effects are trying to modify the same property to two different targets. The visual result of the box example is the box jumping to the left, then the right, then left, and so on with each effect attempting to update the position. Particularly in the case of Sgine, being a fully multi-threaded and thread-safe 3d engine this was of great concern. PropertyAnimator simplifies and abstracts the process, preventing effect fighting from occurring.

Consider another scenario. In the Location class we created above, if it were being managed in a layout manager of some sort, the layout manager wouldn't necessarily know anything about animation. By virtue of the fact that the layout manager modifies the "x" property, instead of immediately jumping to the new position, animation would occur.

LinearNumericAnimator is just one built-in PropertyAnimator. EasingNumericAnimator similarly works with Double properties, but uses easing functions instead of linear motion. Numbers are not the only things that can be animated though. The EasingColorAnimator works similarly to EasingNumericAnimator, but works on Colors. At the time of this writing those are the only built-in PropertyAnimators, but writing your own PropertyAnimator is a simple matter of providing your own implementation of the following method signature:
def apply(current: T, target: T, elapsed: Double): T

Easings with animation is generally more elegant than simple linear animation. Sgine has a full listing of easing functions supported:
  • BackIn / BackOut / BackInOut
  • BounceIn / BounceOut / BounceInOut
  • CircularIn / CircularOut / CircularInOut
  • CubicIn / CubicOut / CubicInOut
  • ElasticIn / ElasticOut / ElasticInOut
  • ExponentialIn / ExponentialOut / ExponentialInOut
  • LinearIn / LinearOut / LinearInOut
  • QuadraticIn / QuadraticOut / QuadraticInOut
  • QuarticIn / QuarticOut / QuarticInOut
  • QuinticIn / QuinticOut / QuinticInOut
  • SineIn / SineOut / SineInOut

This concludes part 1 of this tutorial. Please continue on to part 2 for the conclusion.

Tuesday, August 31, 2010

The Advantages of Using Properties Instead of Getters/Setters in Java and Scala

There have been a few concepts I've been pushing for most of my programming career, and one of the top items on that list is Properties. This is a topic I've been meaning to delve into for a while, and have previously, but as this is an incredibly important part of the core of Sgine I would like to take some time to state the case for Properties. This post is all about why Properties are useful and ultimately why Sgine uses them. The next post, Properties - Tutorial, will go into great detail about how to use Properties within the context of Sgine. Properties as a term is bandied about in many different forms, so first a definition is in order.

Definition


In the context of this discussion a Property is the concept of a wrapper class around a value as an alternative to the standard private field, public getter/setter. For example in Java:
public class Person {
    private String name;

    public String getName() {
        return name;
    }

    public void setName(String name) {
        this.name = name;
    }
}

person.setName("John Doe");
String name = person.getName();

The property alternative:
public class Person {
    public final name = new Property<String>(null);
}

person.name.set("John Doe");
String name = person.name.get();

A basic Property class in Java might look something like:
public class Property<T> {
    private T value;

    public void set(T value) {
        this.value = value;
    }

    public T get() {
        return value;
    }
}

Or in Scala:
class Person {
    private var _name: String = _

    def name = _name

    def name_=(name: String) = _name = name
}

person.name = "John Doe"
val name = person.name

The property alternative:
class Person {
    val name = new Property[String](null)
}

person.name := "John Doe"
val name = person.name()

A basic Property class in Scala might look something like:
class Property[T] {
    private var value: T = _

    def :=(value: T) = this.value = value

    def apply() = value
}

Since this blog is about Sgine, a 3d engine in Scala, I will be showing all further example code only in Scala though much of this still applies directly to Java.

Advantages


This is all well and good, but for all practical appearances it's about the same as classic getter/setter pattern. Why would you want to break convention and use Properties instead?

In the examples above there isn't much practical difference between the two methodologies. Performance-wise they are identical. Properties reduce the amount of code in your class by a pretty large margin if you consider one line:
val name = new Property[String](null)

Instead of three (or seven in the case of Java):
private var _name: String = _
def name = _name
def name_=(name: String) = _name = name

However, the typical response here is that with today's IDEs they will generate the getter/setters for you automagically, so what does that matter? To some degree I agree with this, but it is still clutter added to your classes that make managing your source code that much more complicated down the road. I will agree though that this in and of itself is not a convincing reason to break convention so lets get on to the real reason.

Ultimately it comes down to one major reason to use the Property concept instead of methods, and that is encapsulation. Given the above Person example with a "name", how would you handle a check upon assignment to make sure the name is no more than 255 characters, and if it is, to chop off the extra characters?

Perhaps something like:
class Person {
    private var _name: String = _

    def name = _name

    def name_=(name: String) = {
        val filtered = ...filter name...
        _name = filtered
    }
}

That's simple enough, but what if you want to add an "address" field as well and limit it to 255 characters? Sure, you can simplify by creating a "filter" method you can invoke, but no matter how you swing it, you end up with boiler-plate code to accomplish your task. The reason this happens is because methods don't have the ability to encapsulate and extend functionality the way that an object does. So for the above example I could simply create a FilteredProperty trait that can be mixed in:
trait FilteredProperty extends Property[String] {
    abstract override :=(value: String) = {
        val filtered = ...filter value...
        super.:=(filtered)
    }
}

Now you simply make a very slight change to your person class to get the benefits of filtering:
class Person {
    val name = new Property[String](null) with FilteredProperty
}

Hopefully you are beginning to see the advantages here. In order to add functionality to your simply extend Property functionality to provide what you want and then mix it together in your resulting Property instance. Not only does this remove any need for boilerplate code, but it makes reading the source code for classes extremely easy as a single line of code describes the make-up of that property.

Here are a few examples of things you can do well with Properties:
  • Events: Adding event listeners to a property is one of the most common uses for a Property system as event handling in a standard Bean scenario involves a lot of boiler-plate code. Here's a simple example of usage:
    person.name.listeners += nameChanged _
    
    private def nameChanged(evt: PropertyChangeEvent[String]) = {
        ... property has changed ...
    }

    As is hopefully obvious, any changes to person.name will result in the invocation of the nameChanged method.
  • Binding: Synchronizing two properties together via binding is probably the second most common reasons Properties are used. This is something that takes quite a bit of effort to support, but by leveraging Properties the functionality can be written once and then re-used everywhere desired. Here's a simple example of usage:
    person1.name bind person2.name

    Sure, that's incredibly simple, but now any changes to person2's name will result in person1's name being updated as well.
  • Translation: The ability to translate incoming values is a very broad topic, whether you are converting Strings to Ints, translating to another language upon set, or simply modifying all applied data to follow a specific standard Properties can provide all of the functionality by simple extension and mix-in.
  • Filtering: Applying a filter may represent limiting of a String to 255 characters as discussed above, or it may include throwing an Exception if invalid data is attempted to be passed (ex. Attempting to set null on a Property that should not be null).
  • Transactions: Simple transaction support may consist of an uncommitted value along with commit and rollback methods. This would allow simple management of objects in a user-interface and the ability to know if the object has uncommitted values and what they are so you are only applying changes to the database instead of sweeping updates.
  • Animation: Also called "adjusters", this is the ability to move from one value to another over time. This is most commonly applied to numbers. For example, a AnimatingProperty trait may start at 0.0 but when the value is set to 100.0 instead of applying it directly, it is applied over time and possibly with an easing function.
  • Delegation: This allows the ability delegate getting/setting functionality to another method, function, or value.

There is a great deal more that can be done with Properties, but this is hopefully a nice glimpse to what is possible.

History


As much as I'd love to claim responsibility for the idea of properties, alas, it is not my own and here are a few URLs to other posts about properties (including some older ones from me):

http://joda-beans.sourceforge.net/
http://www.javalobby.org/java/forums/t88090.html?start=50#92123353
http://www.javalobby.org/java/forums/t90756.html
http://www.javalobby.org/java/forums/m92123242.html
http://www.matthicks.com/2009/06/have-beans-been-holding-us-back.html
http://www.matthicks.com/2009/07/death-of-beans.html

Saturday, August 28, 2010

Stressed - Asynchronously

In a previous post I discussed the improvement in performance to be able to display 1600 cubes on the screen at a time at 32fps. It has been quite a while since that post, and until recently I've been busy with infrastructure cleanup and changes and just recently cycled back around to take another look at performance.

One of the main goals of Sgine from the beginning has been to be a multi-threaded 3d engine. However, early on there were some problems with updating content asynchronously that were causing visual issues so I modified all non-renderer updates to occur in the rendering thread (like every other 3d engine out there). After remembering that I had never switched this back over (yeah, just took a single line of code: Updatable.useWorkManager = true), I re-ran the stress test:



Yes, you read that right, I went from 32fps all the way up to 296fps. That's a pretty substantial jump if I do say so myself. There are additional tweaks I can make to make it even faster I think, but I'll save that for another day.

While we're on the topic of multi-threading 3d engines I'd like to share another recent development. As anyone ever developing multi-threaded support in a 3d engine is aware, there are many times you find yourself outside of the rendering thread, but you need to invoke some rendering logic. This was a big problem in jME and Renanse (Joshua Slack) and I came up with the GameTaskQueue that would allow you to inject "work" to be done into a queue to be done later in the rendering thread.

To some degree I make use of the same concept in Sgine, but have created a trait "org.sgine.work.Worker" to represent this functionality. Either a Function0[Unit] or curried function can be passed to invokeLater or invokeAndWait like so:

renderer.invokeLater {
    glPolygonMode(GL_FRONT, GL_POINT)
}

This is on par, but slightly easier to read than passing a Runnable to a queue, but still not too far off. However, I remember being incredibly frustrated in jME having to do this all the time when I needed to inject some logic into the renderer. Now, in Sgine that shouldn't usually be necessary as its purpose is to abstract the rendering layer, but still I wanted to take this a step further (primarily for internal use).

So here's a real-world example of a problem that was created by asynchronous updates. In the Debug trait it adds a keyboard listener to do specific things upon key press. For example, hitting Escape will shutdown the renderer. This shouldn't be a problem as the Renderer.shutdown() method can be invoked asynchronously with any trouble, but unfortunately I have no reference to the current Renderer, so I have to use Renderer() to get the ThreadLocal Renderer. Obviously if keyboard events are being thrown outside of the renderer thread this causes a problem.

So, in an effort to simplify this process, a new feature has been added to EventHandler to allow a "worker" to be specified. If a worker is specified then all calls to invoke the wrapped listener will be "invokeAndWait" through that worker.

In the case of Debug:

Keyboard.listeners += EventHandler(handleKey, ProcessingMode.Blocking)

Now becomes:

Keyboard.listeners += EventHandler(handleKey, ProcessingMode.Blocking, worker = renderer)

From now on I don't have to worry about finding the right renderer and then invokeAndWait within that renderer for anything within the handleKey function, I can rely on the fact that handleKey is always going to be run within the rendering thread.

Saturday, July 31, 2010

Writing Tetris

I'm currently in the process of writing a couple small games in an effort to use Sgine for some real-world game development in order to fill in any wholes and evaluate how easy it is to use in a complete application. The first simple game on my agenda is Tetris. I'd say I'm about half-way finished writing it, and I think it's coming along pretty well:



As you can see I'm using 3d cubes to represent pieces and though the game itself is still played on two dimensions the appearance is in 3d slightly translucent and rotated on the x-axis in order to give it a little more depth. I'll post a link to play the game once it is finished, but I know many people have asked if there are any practical examples of Sgine and I just wanted to send out this note to let everyone know that there will be soon. :)

Also, you can't see it from the screenshot obviously, but I have the original Tetris theme playing in the background on loop. Oh, how it takes me back to playing on the Gameboy as a kid. :)

Sunday, July 25, 2010

Performance Update

I mentioned previously that I was working on performance tweaks to Sgine and I referenced a performance test of 400 cubes rotating and animating back-and-forth on the screen at 38fps.

Since that post I have made some pretty major updates to the backing renderer to increase performance. There's still quite a bit left that can be tweaked, but at this point I'm much more confident in the overall performance of the engine but wanted to outline a few of the changes made to the system and the current results.

First, I was previously using render.RenderImage, that basically just displays a textured quad on the screen with immediate mode rendering. This has been changed to rely on render.shape.Shape class that allows any OpenGL shape to be represented with vertices with triangles, quads, etc. The new Shape class also automatically detects whether VBO is supported and will use that if available over immediate mode. I plan to add additional support for GLSL in the future, but at the moment my focus is the best compatibility and working upwards.

Next, I created a new ui.Box class that represents a modifiable, texturable box on the screen. Previously the cubes being displayed were represented by six distinct RenderImages backed against the same texture.

Finally, I've added the ComponentInstance class I mentioned in my last post to allow instancing rather than creating completely new cubes each time.

The results of these changes were a pretty good bump in performance:



Now, keep in mind that framerates are not linear, so this is a pretty substantial improvement over the 38fps I was previously getting.

To prove this I added a few more boxes to the scene (1600 to be exact) to get my framerate back into the 30s:



The new source code looks a bit different from the previous stress test, but pretty easy to read I think:

package org.sgine.ui

import org.sgine.core.Resource

import org.sgine.easing.Elastic

import org.sgine.effect._

import org.sgine.property.animate.EasingNumericAnimator
import org.sgine.property.animate.LinearNumericAnimator

import org.sgine.render.Debug
import org.sgine.render.RenderSettings
import org.sgine.render.StandardDisplay

import org.sgine.ui.ext.AdvancedComponent

import scala.math._

object TestStressBoxes extends StandardDisplay with Debug {
 override val settings = RenderSettings.High
 
 def setup() = {
  var box: Box = null
  for (row <- 0 until 40) {
   for (column <- 0 until 40) {
    val y = (row * 30.0) - 350.0
    val z = (column * 90.0) - 1800.0
    if ((row == 0) && (column == 0)) {
     box = createBox(y, z, 0.05)
    } else {
     createInstance(box, y, z, 0.05)
    }
   }
  }
 }
 
 def createBox(y: Double, z: Double, scale: Double) = {
  val box = new Box()
  box.scale.set(scale)
  box.location.set(0.0, y, z)
  box.source := Resource("sgine_256.png")
  
  animate(box)
  
  box.update(0.0)
  scene += box
  
  box
 }
 
 def createInstance(component: Component, y: Double, z: Double, scale: Double) = {
  val instance = ComponentInstance(component)
  instance.scale.set(scale)
  instance.location.set(0.0, y, z)
  
  animate(instance)
  
  instance.update(0.0)
  scene += instance
  
  instance
 }
 
 private def animate(component: AdvancedComponent) = {
  component.rotation.x.animator = new LinearNumericAnimator(2.0)
  component.rotation.y.animator = new LinearNumericAnimator(2.0)
  component.rotation.z.animator = new LinearNumericAnimator(2.0)
  component.location.x.animator = new EasingNumericAnimator(Elastic.easeInOut, 3.0)
  
  component.rotation.x := Double.MaxValue
  component.rotation.y := Double.MaxValue
  component.rotation.z := Double.MaxValue

  // Move the cube back and forth perpetually on the x-axis
  val me0 = new PauseEffect(random * 2.0 + 0.5)
  val me1 = new PropertyChangeEffect(component.location.x, -400.0)
  val me2 = new PropertyChangeEffect(component.location.x, 400.0)
  val move = new CompositeEffect(me0, me1, me2)
  move.repeat = -1
  move.start()
 }
}

This could be cleaned up more, but should give you an idea of how simple it is to fill a scene up with boxes. :) I'd like to make one last note on this test that you can't see unless you actually run it. The previous test took around 20 seconds to asynchronously fully load all cubes on the screen. Since I'm using ComponentInstance that time has dropped to about one second to fully load on the screen. This is a drastic improvement in scene load-time. Now it's time to get back to work on functionality.

Saturday, July 24, 2010

Instancing

I've mostly been working on performance and internal tweaks to Sgine lately, but one of the things I've recently added that is both a performance benefit and a really "nice-to-have" for game development particularly is what I call "instancing". In a game you often have many instances of an object on the screen at any given time whether it's a wall, a character, or something else it can be really performance intensive and a waste of memory to have each instance completely represented independently of the others, particularly for complex objects. To help resolve this I've created ComponentInstance to allow creation of a Component and then re-use without all the baggage.

For example, see the following code snippet:

package org.sgine.ui

import org.sgine.core.Color
import org.sgine.core.Resource

import org.sgine.render.Debug
import org.sgine.render.StandardDisplay

object TestInstancing extends StandardDisplay with Debug {
 def setup() = {
  val image = new Image(Resource("puppies.jpg"))
  scene += image
  
  val i1 = ComponentInstance(image)
  i1.location.set(-300.0, 200.0, 5.0)
  i1.scale.set(0.4)
  i1.alpha := 0.5
  i1.color := Color.Red
  scene += i1
  
  val i2 = ComponentInstance(image)
  i2.location.set(300.0, 200.0, 5.0)
  i2.scale.set(0.4)
  i2.alpha := 0.5
  i2.color := Color.Green
  scene += i2
  
  val i3 = ComponentInstance(image)
  i3.location.set(-300.0, -200.0, 5.0)
  i3.scale.set(0.4)
  i3.alpha := 0.5
  i3.color := Color.Blue
  scene += i3
  
  val i4 = ComponentInstance(image)
  i4.location.set(300.0, -200.0, 5.0)
  i4.scale.set(0.4)
  i4.alpha := 0.5
  scene += i4
 }
}

The result of the above code is this:



This also makes it easy to modify one component and reflect it in all instances. There's a lot more functionality in the pipeline that will help increase the performance of Sgine, but unfortunately most of it is internal and not as cool to look at in a blog post. :)
Scala Engine for high-performance interactive applications.