This question already has answers here:
A Base Class pointer can point to a derived class object. Why is the vice-versa not true?
(13 answers)
Closed 8 years ago.
I'm a newbie to Java programming, trying to get the hang of OOP.
So I built this abstract class:
public abstract class Vehicle{....}
and 2 subclasses:
public class Car extends Vehicle{....}
public class Boat extends Vehicle{....}
Car and Boat also hold some unique fields and methods that aren't common (don't have the same name, so I can't define an abstract method for them in Vehicle).
Now in mainClass I have setup my new Garage:
Vehicle[] myGarage= new Vehicle[10];
myGarage[0]=new Car(2,true);
myGarage[1]=new Boat(4,600);
I was very happy with polymorphism until I tried to access one of the fields that are unique to Car, such as:
boolean carIsAutomatic = myGarage[0].auto;
The compiler doesn't accept that. I worked around this issue using casting:
boolean carIsAutomatic = ((Car)myGarage[0]).auto;
That works... but it doesn't help with methods, just fields. Meaning I can't do
(Car)myGarage[0].doSomeCarStuff();
So my question is - what do I really have in my garage? I'm trying to get the intuition as well as understand what's going on "behind the scenes".
for the sake of future readers, a short summary of the answers below:
Yes, there's a Car in myGarage[]
Being a static typed language, the Java compiler will not lend access to methods/fields that are non-"Vehicle", if accessing those through a data structure based on the Vehicle super class( such as Vehicle myGarage[])
As for how to solve, there are 2 main approaches below:
Use type casting, which will ease the compiler's concerns and leave any errors in the design to run time
The fact that I need casting says the design is flawed. If I need access to non-Vehicle capabilities then I shouldn't be storing the Cars and Boats in a Vehicle based data structure. Either make all those capabilities belong to Vehicle, or use more specific (derived) type based structures
In many cases, composition and/or interfaces would be a better alternative to inheritance. Probably the subject of my next question...
Plus many other good insights down there, if one does have the time to browse through the answers.
If you need to make the difference between Car and Boat in your garage, then you should store them in distinct structures.
For instance:
public class Garage {
private List<Car> cars;
private List<Boat> boats;
}
Then you can define methods that are specific on boats or specific on cars.
Why have polymorphism then?
Let's say Vehicle is like:
public abstract class Vehicle {
protected int price;
public getPrice() { return price; }
public abstract int getPriceAfterYears(int years);
}
Every Vehicle has a price so it can be put inside the Vehicle abstract class.
Yet, the formula determining the price after n years depends on the vehicle, so it left to the implementing class to define it. For instance:
public Car extends Vehicle {
// car specific
private boolean automatic;
#Override
public getPriceAfterYears(int years) {
// losing 1000$ every year
return Math.max(0, this.price - (years * 1000));
}
}
The Boat class may have an other definition for getPriceAfterYears and specific attributes and methods.
So now back in the Garage class, you can define:
// car specific
public int numberOfAutomaticCars() {
int s = 0;
for(Car car : cars) {
if(car.isAutomatic()) {
s++;
}
}
return s;
}
public List<Vehicle> getVehicles() {
List<Vehicle> v = new ArrayList<>(); // init with sum
v.addAll(cars);
v.addAll(boats);
return v;
}
// all vehicles method
public getAveragePriceAfterYears(int years) {
List<Vehicle> vehicules = getVehicles();
int s = 0;
for(Vehicle v : vehicules) {
// call the implementation of the actual type!
s += v.getPriceAfterYears(years);
}
return s / vehicules.size();
}
The interest of polymorphism is to be able to call getPriceAfterYears on a Vehicle without caring about the implementation.
Usually, downcasting is a sign of a flawed design: do not store your vehicles all together if you need to differenciate their actual type.
Note: of course the design here can be easily improved. It is just an example to demonstrate the points.
To answer your question you can find out what exactly is in your garage you do the following:
Vehicle v = myGarage[0];
if (v instanceof Car) {
// This vehicle is a car
((Car)v).doSomeCarStuff();
} else if(v instanceof Boat){
// This vehicle is a boat
((Boat)v).doSomeBoatStuff();
}
UPDATE: As you can read from the comments below, this method is okay for simple solutions but it is not good practice, particularly if you have a huge number of vehicles in your garage. So use it only if you know the garage will stay small. If that's not the case, search for "Avoiding instanceof" on stack overflow, there are multiple ways to do it.
If you operate on the base type, you can only access public methods and fields of it.
If you want to access the extended type, but have a field of the base type where it's stored (as in your case), you first have to cast it and then you can access it:
Car car = (Car)myGarage[0];
car.doSomeCarStuff();
Or shorter without temp field:
((Car)myGarage[0]).doSomeCarStuff();
Since you are using Vehicle objects, you can only call methods from the base class on them without casting. So for your garage it may be advisable to distinguish the objects in different arrays - or better lists - an array is often not a good idea, since it's far less flexible in handling than a Collection-based class.
You defined that your garage will store vehicles, so you do not care what type of vehicles you have. The vehicles have common features like engine, wheel, behavior like moving.
The actual representation of these features might be different, but at abstract layer are the same.
You used abstract class which means that some attributes, behaviors are exactly the same by both vehicle. If you want to express that your vehicles have common abstract features then use interface like moving might mean different by car and boat. Both can get from point A to point B, but in a different way (on wheel or on water - so the implementation will be different)
So you have vehicles in the garage which behave the same way and you do not car about the specific features of them.
To answer the comment:
Interface means a contract which describes how to communicate with the outer world. In the contract you define that your vehicle can move, can be steered, but you do not describe how it will actually work, it is described in the implementation.By abstract class you might have functions where you share some implementation, but you also have function which you do not know how it will be implemented.
One example of using abstract class:
abstract class Vehicle {
protected abstract void identifyWhereIAm();
protected abstract void startEngine();
protected abstract void driveUntilIArriveHome();
protected abstract void stopEngine();
public void navigateToHome() {
identifyWhereIAm();
startEngine();
driveUntilIArriveHome();
stopEngine();
}
}
You will use the same steps by each vehicle, but the implementation of the steps will differ by vehicle type. Car might use GPS, boat might use sonar to identify where it is.
I'm a newbie to Java programming, trying to get the hang of OOP.
Just my 2 cents — I will try to make it short as many interesting things have already been said. But, in fact, there is two questions here. One about "OOP" and one about how it is implemented in Java.
First of all, yes, you have a car in your garage. So your assumptions are right. But, Java is a statically typed language. And the type system in the compiler can only "know" the type of your various object by their corresponding declaration. Not by their usage. If you have an array of Vehicle, the compiler only knows that. So it will check that you only perform operation allowed on any Vehicle. (In other words, methods and attributes visible in the Vehicle declaration).
You can explain to the compiler that "you in fact know this Vehicle is a Car", by using an explicit cast (Car). the compiler will believe you -- even if in Java there is a check at run-time, that might lead to a ClassCastException that prevent further damages if you lied (other language like C++ won't check at run-time - you have to know what you do)
Finally, if you really need, you might rely of run-time type identification (i.e.: instanceof) to check the "real" type of an object before attempting to cast it. But this is mostly considered as a bad practice in Java.
As I said, this is the Java way of implementing OOP. There is whole different class family of languages broadly known as "dynamic languages", that only check at run-time if an operation is allowed on an object or not. With those languages, you don't need to "move up" all the common methods to some (possibly abstract) base class to satisfy the type system. This is called duck typing.
You asked your butler:
Jeeves, remember my garage on the Isle of Java? Go check whether the first vehicle parked there is automatic.
and lazy Jeeves said:
but sir, what if it's a vehicle that can't be automatic or non-automatic?
That's all.
Ok, that's not really all since reality is more duck-typed than statically typed. That's why I said Jeeves is lazy.
Your problem here is at a more fundamental level: you built Vehicle in such a way that Garage needs to know more about its objects than the Vehicle interface gives away. You should try and build the Vehicle class from the Garage perspective (and in general from the perspective of everything that's going to use Vehicle): what kind of things do they need to do with their vehicles? How can I make those things possible with my methods?
For example, from your example:
bool carIsAutomatic = myGarage[0].auto;
Your garage want to know about a vehicle's engine for... reasons? Anyway, there is no need for this to be just exposed by Car. You can still expose an unimplemented isAutomatic() method in Vehicle, then implement it as return True in Boat and return this.auto in Car.
It would be even better to have a three-valued EngineType enum (HAS_NO_GEARS, HAS_GEARS_AUTO_SHIFT, HAS_GEARS_MANUAL_SHIFT), which would let your code reason on the actual characteristics of a generic Vehicle cleanly and accurately. (You'd need this distinction to handle motorbikes, anyway.)
You garage contains Vehicles, so the compiler static control view that you have a Vehicle and as .auto is a Car field you can't access it, dynamically it is a Car so the cast don't create some problem, if it will be a Boat and you try to make cast to Car will rise an exception on runtime.
This is a good place for application of the Visitor design pattern.
The beauty of this pattern is you can call unrelated code on different subclasses of a superclass without having to do weird casts everywhere or putting tons of unrelated methods into the superclass.
This works by creating a Visitor object and allowing our Vehicle class to accept() the visitor.
You can also create many types of Visitor and call unrelated code using the same methods, just a different Visitor implementation, which makes this design pattern very powerful when creating clean classes.
A demo for example:
public class VisitorDemo {
// We'll use this to mark a class visitable.
public static interface Visitable {
void accept(Visitor visitor);
}
// This is the visitor
public static interface Visitor {
void visit(Boat boat);
void visit(Car car);
}
// Abstract
public static abstract class Vehicle implements Visitable {
// NO OTHER RANDOM ABSTRACT METHODS!
}
// Concrete
public static class Car extends Vehicle {
public void doCarStuff() {
System.out.println("Doing car stuff");
}
#Override
public void accept(Visitor visitor) {
visitor.visit(this);
}
}
// Concrete
public static class Boat extends Vehicle {
public void doBoatStuff() {
System.out.println("Doing boat stuff");
}
#Override
public void accept(Visitor visitor) {
visitor.visit(this);
}
}
// Concrete visitor
public static class StuffVisitor implements Visitor {
#Override
public void visit(Boat boat) {
boat.doBoatStuff();
}
#Override
public void visit(Car car) {
car.doCarStuff();
}
}
public static void main(String[] args) {
// Create our garage
Vehicle[] garage = {
new Boat(),
new Car(),
new Car(),
new Boat(),
new Car()
};
// Create our visitor
Visitor visitor = new StuffVisitor();
// Visit each item in our garage in turn
for (Vehicle v : garage) {
v.accept(visitor);
}
}
}
As you can see, StuffVisitor allows you to call different code on Boat or Car depending on which implementation of visit is called. You can also create other implementations of the Visitor to call different code with the same .visit() pattern.
Also notice that using this method, there is no use of instanceof or any hacky class checking. The only duplicated code between classes is the method void accept(Visitor).
If you want to support 3 types of concrete subclasses for example, you can just add that implementation into the Visitor interface too.
I'm really just pooling the ideas of the others here (and I'm not a Java guy, so this is pseudo rather than actual) but, in this contrived example, I would abstract my car checking approach into a dedicated class, that only knows about cars and only cares about cars when looking at garages:
abstract class Vehicle {
public abstract string getDescription() ;
}
class Transmission {
public Transmission(bool isAutomatic) {
this.isAutomatic = isAutomatic;
}
private bool isAutomatic;
public bool getIsAutomatic() { return isAutomatic; }
}
class Car extends Vehicle {
#Override
public string getDescription() {
return "a car";
}
private Transmission transmission;
public Transmission getTransmission() {
return transmission;
}
}
class Boat extends Vehicle {
#Override
public string getDescription() {
return "a boat";
}
}
public enum InspectionBoolean {
FALSE, TRUE, UNSUPPORTED
}
public class CarInspector {
public bool isCar(Vehicle v) {
return (v instanceof Car);
}
public bool isAutomatic(Car car) {
Transmission t = car.getTransmission();
return t.getIsAutomatic();
}
public bool isAutomatic(Vehicle vehicle) {
if (!isCar(vehicle)) throw new UnsupportedVehicleException();
return isAutomatic((Car)vehicle);
}
public InspectionBoolean isAutomatic(Vehicle[] garage, int bay) {
if (!isCar(garage[bay])) return InspectionBoolean.UNSUPPORTED;
return isAutomatic(garage[bay])
? InspectionBoolean.TRUE
: InspectionBoolean.FALSE;
}
}
Point is, you've already decided you only care about cars when you ask about the car's transmission. So just ask the CarInspector. Thanks to the tri-state Enum, you can now know whether it is automatic or even if it is not a car.
Of course, you'll need different VehicleInspectors for each vehicle you care about. And you have just pushed the problem of which VehicleInspector to instantiate up the chain.
So instead, you might want to look at interfaces.
Abstract getTransmission out to an interface (e.g. HasTransmission). That way, you can check if a vehicle has a transmission, or write an TransmissionInspector:
abstract class Vehicle { }
class Transmission {
public Transmission(bool isAutomatic) {
this.isAutomatic = isAutomatic;
}
private bool isAutomatic;
public bool getIsAutomatic() { return isAutomatic; }
}
interface HasTransmission {
Transmission getTransmission();
}
class Car extends Vehicle, HasTransmission {
private Transmission transmission;
#Override
public Transmission getTransmission() {
return transmission;
}
}
class Bus extends Vehicle, HasTransmission {
private Transmission transmission;
#Override
public Transmission getTransmission() {
return transmission;
}
}
class Boat extends Vehicle { }
enum InspectionBoolean {
FALSE, TRUE, UNSUPPORTED
}
class TransmissionInspector {
public bool hasTransmission(Vehicle v) {
return (v instanceof HasTransmission);
}
public bool isAutomatic(HasTransmission h) {
Transmission t = h.getTransmission();
return t.getIsAutomatic();
}
public bool isAutomatic(Vehicle v) {
if (!hasTranmission(v)) throw new UnsupportedVehicleException();
return isAutomatic((HasTransmission)v);
}
public InspectionBoolean isAutomatic(Vehicle[] garage, int bay) {
if (!hasTranmission(garage[bay])) return InspectionBoolean.UNSUPPORTED;
return isAutomatic(garage[bay])
? InspectionBoolean.TRUE
: InspectionBoolean.FALSE;
}
}
Now you are saying, you only about transmission, regardless of Vehicle, so can ask the TransmissionInspector. Both the bus and the car can be inspected by the TransmissionInspector, but it can only ask about the transmission.
Now, you might decide that boolean values are not all you care about. At that point, you might prefer to use a generic Supported type, that exposes both the supported state and the value:
class Supported<T> {
private bool supported = false;
private T value;
public Supported() { }
public Supported(T value) {
this.isSupported = true;
this.value = value;
}
public bool isSupported() { return supported; }
public T getValue() {
if (!supported) throw new NotSupportedException();
return value;
}
}
Now your Inspector might be defined as:
class TransmissionInspector {
public Supported<bool> isAutomatic(Vehicle[] garage, int bay) {
if (!hasTranmission(garage[bay])) return new Supported<bool>();
return new Supported<bool>(isAutomatic(garage[bay]));
}
public Supported<int> getGearCount(Vehicle[] garage, int bay) {
if (!hasTranmission(garage[bay])) return new Supported<int>();
return new Supported<int>(getGearCount(garage[bay]));
}
}
As I've said, I'm not a Java guy, so some of the syntax above may be wrong, but the concepts should hold. Nevertheless, don't run the above anywhere important without testing it first.
If you are on Java, could use reflections to check if a function is available and execute it, too
Create Vehicle level fields that will help make each individual Vehicle more distinct.
public abstract class Vehicle {
public final boolean isCar;
public final boolean isBoat;
public Vehicle (boolean isCar, boolean isBoat) {
this.isCar = isCar;
this.isBoat = isBoat;
}
}
Set the Vehicle level fields in the inheriting class to the appropriate value.
public class Car extends Vehicle {
public Car (...) {
super(true, false);
...
}
}
public class Boat extends Vehicle {
public Boat (...) {
super(false, true);
...
}
}
Implement using the Vehicle level fields to properly decipher the vehicle type.
boolean carIsAutomatic = false;
if (myGarage[0].isCar) {
Car car = (Car) myGarage[0];
car.carMethod();
carIsAutomatic = car.auto;
}
else if (myGarage[0].isBoat) {
Boat boat = (Boat) myGarage[0];
boat.boatMethod();
}
Since your telling your compiler that everything in your garage is a Vehicle, your stuck with the Vehicle class level methods and fields. If you want to properly decipher the Vehicle type, then you should set some class level fields e.g. isCar and isBoat that will give you the programmer a better understanding of what type of Vehicle you are using.
Java is a type safe language so its best to always type check before handling data that has been casted like your Boats and Cars.
Modeling objects you want to present in a program (in order to solve some problem) is one thing, coding is another story. In your code, I think essentially it's inappropriate to model a garage using array. Arrays shouldn't be often considered as objects, although they do appear to be, usually for the sake of self-contained-ness sort of integrity of a language and providing some familiarity, but array as a type is really just a computer-specific thing, IMHO, especially in Java, where you can't extend arrays.
I understand that correctly modeling a class to represent a garage won't help answer your "cars in a garage" question; just a piece of advice.
Head back to the code. Other than getting some hang to OOP, a few questions would be helpful creating a scene hence to better understand the problem you want to resolve (assuming there is one, not just "getting some hang"):
Who or what wants to understand carIsAutomatic?
Given carIsAutomatic, who or what would perform doSomeCarStuff?
It might be some inspector, or someone who knows only how to drive auto-transmission cars, etc., but from the garage's perspective, all it knows is it holds some vehicle, therefore (in this model) it is the responsibility of this inspector or driver to tell if it's a car or a boat; at this moment, you may want to start creating another bunch of classes to represent similar types of *actor*s in the scene. Depends on the problem to be resolved, if you really have to, you can model the garage to be a super intelligent system so it behaves like a vending machine, instead of a regular garage, that has a button says "Car" and another says "Boat", so that people can push the button to get a car or a boat as they want, which in turn makes this super intelligent garage responsible for telling what (a car or a boat) should be presented to its users; to follow this improvisation, the garage may require some bookkeeping when it accepts a vehicle, someone may have to provide the information, etc., all these responsibilities go beyond a simple Main class.
Having said this much, certainly I understand all the troubles, along with the boilerplates, to code an OO program, especially when the problem it tries to resolve is very simple, but OO is indeed a feasible way to resolve many other problems. From my experience, with some input providing use cases, people start to design scenes how objects would interact with each other, categorize them into classes (as well as interfaces in Java), then use something like your Main class to bootstrap the world.
Related
I have the interface
interface Car {
enum Type { MITSUBISHI, FORD }
Car.Type getType();
And two of classes implementing the interface
class Mitsubishi implements Car {
#Override
Car.Type getType() { return Car.Type.MITSUBISHI; }
}
class Ford implements Car {
#Override
Car.Type getType() { return Car.Type.FORD; }
}
And use it field like
List<Car> cars = this.cars;
List<Mitsubishi> mitsubishiCars = cars.stream().filter(c -> c.getType().equals(Car.Type.MITSUBISHI)).collect(Collectors.toList());
Question, is this a bad pattern or not? If it is, then why?
Generally speaking, it seems like a bad pattern because the interface Car is not supposed to know who implements it, while containing this enum Type automatically acquires the knowledges of its potential implementations.
This said, it all depends how you think to use the Car.Type.
Most probably, you will want to use the enum values to determine specified actions to perform. For example:
//park the car depending on the type
switch (type) {
case MITSUBISHI:
//park in a certain way
case FORD:
//park in some other way
//...
}
If that's the case, I see at least 2 further problems except for the interface knowing its implementations:
Each time you have a new Car's implementation, you'll need to enrich the enum Type inside the interface Car. It may be annoying
When the values of Type become a lot, whatever flow structure (switch, if blocks etc.) will become huge and hard to maintain.
If that's the case, I'd rather create methods in the interface that handle the action according to the car's type. For example:
interface Car {
void park();
}
class Mitsubishi implements Car {
#Override
public void park() {
//way to park a Mitsubishi
}
}
If instead you simply wanted to use the enum value just to print the type of car at some point (like in a log), then just add a String getType() method in the interface that returns the proper brand:
interface Car {
String getType();
}
class Mitsubishi implements Car {
#Override
public String getType() {
return "Mitsubishi";
}
}
But overall, I wouldn't suggest such approach unless there is not a use you have in mind for Car.Type that you're not sharing (even though I really don't see how else you may want to use an enum except for the 2 cases I listed above).
Post-edit
After seeing how you want to use the enum:
List<Car> cars = this.cars;
List<Mitsubishi> mitsubishiCars = cars.stream()
.filter(c -> c.getType().equals(Car.Type.MITSUBISHI))
.collect(Collectors.toList());
I think you can just define your enum outside the interface, then creating a Type getType() method in the interface and then using it like this:
cars.stream()
.filter(c -> c.getType() == Type.MITSUBISHI)
.collect(Collectors.toList());
(assuming of course that Mitsubishi implementation will return Type.MITSUBISHI on Type getType()).
I am trying to make an enum list, and have an abstract method defined in the enum, which each enum value implements. The problem I am facing is that the abstract class has a generic return type but I want each enum value to return a concrete type.
I'll give an example:
public enum Attributes {
name {
#Override
public void createAttribute(Person person) {
//Do some validations
//Save in some storage
}
#Override
public Name getAttribute(Person person) {
// Validations
// Retreive from storage
return new Name("test");
}
},
address {
#Override
public void createAttribute(Person person) {
//Do some validations
//Save in some storage
}
#Override
public Address getAttribute(Person person) {
// Validations
// Retreive from storage
return new Name("test");
}
}
public abstract Object getAttribute(Person person);
public abstract void createAttribute(Person person);
}
Here the issue is that I would need to do typecasting to get the concrete object which is not recommended and I don't get any type of safety. How Should I go about so that using the enum value I can get my concrete object instead of the generic one.
Now I wanna call this as,
Arrays.stream(Attributes.values()).forEach(r -> {
r.createAttribute(person);
}
final Address address = Attributes.address.getAttribute(person);
final Name name = Attributes.name.getAttribute(person);
So now whenever I need to add a new attribute I don't want to write create methods for it in the Person class every time. I just add it to enum and it gets created. But now since I have the create method in the enum, I also want the getAttribute to be present here.
Here the issue is that I would need to do typecasting to get the concrete object which is not recommended and I don't get any type of safety.
You're right. Given an enum type E with an associated enum constant C, the type of the expression E.C is E. Java provides no mechanism for naming or representing a narrower type for that expression. One of the implications is that although an enum instance can implement methods with covariant return types, the covariance is not visible outside the instance. If you depend for some purpose on the narrower return type of one of those instances' methods, then casting is your only alternative.
And you're right that such casts are not type safe. They cannot be checked by the compiler, and in practice, you as programmer can get them wrong. But the information to perform a compile-time check is not expressed by the language, so there is no scope for a workaround in the language as it is defined today.
How Should I go about so that using the enum value I can get my concrete object instead of the generic one.
You should choose an altogether different approach, not involving an enum.
If you stuck with the enum then you would have to adopt an approach that relies on the enum instances to perform any tasks that depend on their own particular characteristics. Because you ask so persistently, one possibility would be to implement a variation on double dispatch. Instead of a getObject() method, you would have something like
void acceptReceiver(AttributeReceiver r, Person p);
paired with
public interface AttributeReceiver {
default void receiveName(Name name) { /* empty */ }
default void receiveAddress(Address addr) { /* empty */ }
}
Of course, the enum instances would have to implement acceptReceiver appropriately.
You would probably want to use that a little more directly than just to retrieve attributes, but you could use it to retrieve attributes something like this:
class Example {
Name name;
Address address;
void retrieveAttributes(Person person) {
AttributeReceiver receiver = new AttributeReceiver() {
public void receiveName(Name n) { name = n; }
public void receiveAddress(Address a) { addr = a; }
};
Attributes.name.acceptReceiver(receiver, person);
Attributes.address.acceptReceiver(receiver, person);
}
}
But that's awfully roundabout when you have the alternative of using (just) methods, whether on Person or even on some non-enum utility class. I continue not to see any advantage to involving an enum here. I think your code overall would be more complex and harder to understand and maintain with enums than without.
The root issue is that you are abstracting away details that you actually care about. That's a deep design flaw. You can program your way around it, but it would be better to choose a more appropriate level of abstraction in the first place.
I have multiple services (in Spring MVC) that are children of a global Service. So I need to know about the best practice (or your opinions) with multiple methods with this example:
//Domain classes
public class MyParentObject{}
public class MyObj extends MyParentObject{}
//Services
public class MyParentObjectServiceImpl implements MyParentObjectService{
#Override
public MyParentObject findObjectByProp(String prop, String objectType){
//myCode (not abstract class)
}
}
public class MyObjServiceImpl extends MyParentObjectServiceImpl implements MyObjectService{
private myObjType = "MyObj";
#Override
public MyObj findMyObjByProp(String prop){
return (MyObj) super.findObjectByProp(prop, this.myObjType);
}
}
And in this approach, I use calls like this:
MyObj foo = myObjService.findMyObjByProp(prop);
So I need to know if this approach is "better" or more apropiate that calling directly the parent method with the second parameter. E.g:
MyObj foo = (MyObj)myParentObjectService.findObjectByProp(prop, "MyObj");
..and avoiding the creation of second methods, more specific. It is important to know that the children services will be created anyway, because we have lot of code that is specific of a domain objects.
I have the idea that the first approach is better, because is more readable, but I need to support that decision with some documents, blog, or opinions to discuss this designs with my colleagues.
This looks like a tagged class hierarchy. It's difficult to comment on the value of this design in general without knowing the details. However, a slightly different approach that I would recommend is to generify your base class to gain a little bit of type safety.
In particular:
public /* abstract */ class MyParentObjectServiceImpl<T extends MyParentObject>
implements MyParentObjectService{
MyParentObjectServiceImpl(Class<T> type) { this.type = type; }
private final Class<T> type; // subclasses provide this
#Override
public T findObjectByProp(String prop){
//you can use type for object specific stuff
}
}
public class MyObjServiceImpl extends MyParentObjectServiceImpl<MyObj>
// You might not need this interface anymore
// if the only method defined is findMyObjByProp
/* implements MyObjectService */ {
MyObjServiceImpl() {
super(MyObj.class);
}
#Override
public /* final */ MyObj findMyObjByProp(String prop) {
return (MyObj) super.findObjectByProp(prop, this.myObjType);
}
}
You definitely gain in type safety (casting will only appear in the base class), you get rid of the "tags" (the strings that identify the different objects) and possibly reduce the number of classes/interfaces required to implement the whole hierarchy. I successfully used this approach several times. Note that this works best if the base class is abstract. Food for thoughts.
I'm searching a clean pattern to solve the following problem:
I have several type of vehicles (let's say Car, Bike and Truck, all these classes expand the abstract class Vehicle). Each type of vehicle has several specific properties.
I have somewhere a list of vehicle, and i need to build a GUI which makes the user able to select a vehicle, and edit its properties. The result I'd like to achieve is to display a vehicle list (using JList) on the left, and a panel on the right containing the fields needed to edit the selected Vehicle.
I know how to do it when the list contains only one type of items (I use a Jlist on the left, and a custom JPanel on the right). I made one custom panel for each subclass : CarPanel, BikePanel...
My problem is the link between the type of the selected Vehicle and the according panel. Although it would work, I'd like to avoid something like that :
if (selectedVehicle instanceof Car) {
useThisPanel(new CarPanel((Car)selectedVehicle));
} ...
because it doesnt seems very maintainable to me.
I also want to avoid something like useThisPanel(selectedVehicle.getPanel()) because there is no reason any vehicle class should know anything about the way it is displayed
I'm currently working on switching my code to a MVC pattern, but it seems to me it's not enought to solve my problem.
I'm pretty sure I'm not the first one to face this, but I couldn't find any answer or advice to manage this kind of situations.
The usual way of solving this problem in an OO way is to use the visitor pattern:
public interface VehicleVisitor<T> {
T visitCar(Car car);
T visitTruck(Truck truck);
T visitBike(Bike bike);
}
public abstract class Vehicle {
public abstract <T> T accept(VehicleVisitor<T> visitor);
}
public class Car extends Vehicle {
#Override
public <T> T accept(VehicleVisitor<T> visitor) {
return visitor.visitCar(this);
}
}
// same for Bike and Truck
And finally
public class SomeClass {
private JPanel createPanelFor(Vehicle vehicle) {
return vehicle.accept(new VehicleVisitor<JPanel>() {
#Override
public JPanel visitCar(Car car) {
return new CarPanel(car);
}
// same for Truck and Bike
}
}
}
This is clean and everything, but it's a bit more difficult to understand than a simple instanceof based solution. It has an advantage though: you can't forget handling a new Vehicle type: it will force you to implement the abstract accept() method, which will force you to add another method to the Visitor interface, which will force you to implement it in every visitor.
Let's say that I have abstract class Car and there are x concrete classes (RedCar, BlueCar ...) that extend class Car. Then I have class Garage that has property Car[] cars.
I want to write a boolean method to class Garage that will loop through the cars and returns true if searched car (one of concrete type (RedCar, BlueCar..) passed as parameter) was found, otherwise returns false.
What is the best way to do it?
For now, I have something like this :
public boolean hasCar(Class<? extends Car> c) {
for (int i = 0; i < this.cars.length; i++) {
if (c.isInstance(this.cars[i])) {
return true;
}
}
return false;
}
but wouldn't it be better if I create an enum with all possible Car types (subclasses of Car), add a property to class Car, which will hold the constant from enum and compare it based on it?
Something like this:
public enum CarType{
RED_CAR, BLUE_CAR
}
public abstract class Car{
public CarType type;
Car(CarType type){
this.type = type;
}
}
public class RedCar extends Car{
public RedCar(){
super(CarType.RED_CAR);
}
}
public class BlueCar extends Car{
public BlueCar(){
super(CarType.BLUE_CAR);
}
}
public class Garage{
private Car[] cars;
public boolean hasCar(CarType type) {
for (int i = 0; i < this.cars.length; i++) {
if(this.cars[i].type == type){
return true;
}
}
return false;
}
}
Which one is the better method?
If all the subclasses don't add any state to the base class and don't change or add behavior by adding or overriding methods to/of the base class, then you shouldn't have subclasses. Just having a Car class with a type attribute is sufficient.
If the subclasses have a good reason to exist, then you could indeed use an enum, but that would prevent adding any other type of car (without also changing the enum).
Having a method which searches for a specific type of car looks like a design smell to me. That probably means that the Car class is not sufficiently polymorphic: instanceof and casts are not very OO.
If you're definitely sticking with having the different colours of car as subtypes, I would say your first version is better - your second will involve updating the enumeration every time you add a new type of car and involves a lot of copy-and-paste style coding for the constructors.
However, I'd recommend that if you want to search the garage for certain colours of car it would make sense to have the colour as a property of the car rather than having to check the car's type. In your second example, the code looks 'cleaner' as you're not using reflection, but you've basically got duplicated information as a car's colour is represented both by its type and by its CarType property. Simply having a single Car type with a Colour value would make this code a lot shorter and neater. Though with this said, I don't know what you ultimately want to do with this so your approach of having subtypes may still be best - take what I say with a pinch of salt.
The enum solution you posted will only work if you want to confirm that a car of a certain type is found. Not the specific car object itself. Also, if your Car subclasses don't have any difference other than a type variable, you probably don't need subclasses.
The first method looks fine, and may be shortened (a little) like this:
public boolean hasCar(Class<? extends Car> c) {
for (Car car : this.cars) {
if (c.isInstance(car) {
return true;
}
}
return false;
}
Another approach you could use is passing a filter object and asking the car itself if it matches the filter. the implemntation details of the filter may depend on your real life example (and may even compare by using instanceof()), but in the simplified example you give, it may be a string or an enum specifying the color you are looking for. something like
public class CarFilter()
{
String Color;
}
In the base class of car you have a method:
public boolean matchesFilter(CarFilter filter){} //maybe abstract?
In the derived class RedCar the method maybe implemnted like this:
public boolean matchesFilter(CarFilter filter)
{
return filter.Color == "red"; //MAYBE case insensitive?
}
To answer this, I have to ask my self which offers the easiest extensibility? I would say the first implementation.