Which class should the conversion code belong to? - oop

If I want to cast from class Foo to another class Bar, should the code for the conversion belong in the Foo or Bar class?
Foo f
Bar b = f.ToBar()
Or:
Foo f
Bar b = Bar.FromFoo(f)
The latter could be rewritten as a constructor rather than a static method, but that's not the point of this question.
Of course in many cases, not being able to modify one of the classes makes the decision for you. For example if I had a custom class Foo and wanted to convert it to an integer.

You can look at it this way: whose concern is the conversion? Should Foo know or care that some other class called Bar, is using it in some ways? Of course not. Because, what if there are MANY other classes that use Foo, in various ways, then should Foo go ahead and take care of them too? No.
Therefore, first important point is that this functionality has to be implemented on Bar's side. And then, comes the question of secondary importance: how to implement it. Here, implementing it as a constructor, as others have pointed out, is more logical and inline with oop principle.

The operation should typically live on the object it affects, so in this case it would be f.ToBar().

Related

Is it correct to return an object which class is not the expected class?

Well, i hope you understand me. I have two classes, A and B. B is subclass of A. They have the same public methods and means the same thing, but B does some things a little different, so it has additional methods and attributes that only uses itself. Let say, class A implements a method newFromWizard that interactively creates an object. Can I implement logic for, depending on the user input, create an object A or and object B in the newFromWizard method of A. I mean, can i create a B object from that method of A? Or i need to implement that elsewhere? How is the best way to do it? In practice, i can. But, it is correct for OOP?
By the way, if that matters, i'm using Smalltalk.
Yes, this is a well-known pattern in OO. In the Objective-C Cocoa libraries you'll find it applied systematically, and is known as class clusters. The result is that the Cocoa libraries are much easier to understand than the equivalent c# or java ones. It allows the hiding of a inheritance hierarchy behind an abstract class that has class side creation methods that return subclasses.
public class A{
public B method(){
B b = new B();
return b;
}
}
class B extends A{
}
If this is what you're talking about, it's valid.
I would say that it's not intuitive way of doing things. Let's simplify it and say that you just redefine new. Then in some point you do A new and get an instance of B. The thing that they are similar makes it not so bad. But imagine that someone else starts working with your code. And hew knows that message new should result in creation of the instance of the receiver. And then something goes different. I'd say that conceptually it's wrong. Why now to implements some builder class? And have there something like
createInstanceOfAB
|className|
className := "do what you need".
^ className asClass new.
This is a more clear way.
Once again you can make new… method to do whatever you want, even shoot fireworks, but most of the people will expect it to create instance of the same class
I think you shouldn't worry too much about whether it is "clean OO" to return an instance of a subclass. It's being used and done so often because it is very helpful and makes your code very readable compared to somme kind of factories and stuff.
My biggest concern here would be that you should name your class method carefully. Don't use #new is probably the most important rule, but you should always use a name that already says: give me an instance of what is appropriate right now.
I'd say this is not limited to subclasses, such a method can even return objects that do not inherit from the class. In a dynamically typed language, this is okay. Remember, we're in a dynamically typed language, so as long as you have polymorphic interfaces, the class of an object is not really important to its users as long as it responds to your message sends...

Adding conflicting methods in an Objective C class using category

I have added a method foo to a class MYCustomClass in a category Category1 separate from the original definition of the class. Then I added another method also called foo in another category Category2. I then call foo on an instance of MYCustomClass. In my case the foo in Category2 is being called. My question is: Is there any explanation for this? Or, is it one of those "undefined"/"compiler dependent" behaviours. Also, is it possible to handle such situations by qualifying the method call by specifying the category I want to be used in the call.
EDIT: I am aware that what I am doing is not supported. I am just interested in if there is a hack around it.
When a category is loaded, its methods are inserted into the existing method table, and there's no way to distinguish where they came from once that's done. The last category to load wins. Back in the NeXTSTEP days, we would sometimes do this deliberately as a very kludgey way to fix a broken method in code for which we didn't have the source.
It’s undefined behaviour. From the Objective-C Programming Language document:
A category cannot reliably override methods declared in another category of the same class.
This issue is of particular significance because many of the Cocoa classes are implemented using categories. A framework-defined method you try to override may itself have been implemented in a category, and so which implementation takes precedence is not defined.
And no, you cannot specify that you want foo from Category1, or foo from Category2. If you need this, you should give different names to those methods, e.g. foo1 and foo2.

Do I correctly understand what a class is?

I've had trouble finding a clear, concise laymans definition of a class. Usually, they give general ideas without specifically spelling it out, and I'm wondering if I'm understanding this correctly. As I understand it, a class is the set of code that controls an object. For example, in an app that has a button for 'Yes' and a button for 'No', and a text box for output, the code that tells the computer what to do when the user uses the Yes button is one class, the code for hitting No is another class, and an object is the two buttons and what they do together to influence the output box. Am I right, or am I confusing terms here?
Thanks
A class is a kind of thing, an object is an actual thing. So, in the real world, "person" is a class, you and I are objects (or instances) of that class. "Car" is a class, my 1996 beater Volvo station wagon is an object.
Objects all have certain similarities in form and function because of the class they belong to. When I say my station wagon is a "car", you have a pretty good idea of what it looks like, what it's used for, and what it can do. But objects of a class can also differ from each other. Just because something's a car doesn't tell you exactly what shape it is or how many seats it has or what color it is.
In the example you gave, it's likely that the yes and no buttons are both objects of the class "button" (or some similar name) and that the differences in their behavior are due to changes added by the programmer without his or her bothering to create a new class. However, it is possible that the programmer made the decision to make each type of button a subclass of the original class "button".
What's a subclass? Well, if you think of "car" as a class, it is obvious that there are several intermediate "kinds" of things between "car" and "Larry's 1996 beater Volvo station wagon". These could be "station wagon" and "Volvo station wagon". So my car would be an instance of "Volvo station wagon" which itself would be subclass of "station wagon" which would be a subclass of "car". From the "car" part, we know a good deal about my object, from the "station wagon" part we learn a little more, and from the "Volvo station wagon" a little more still.
The way in which classes and subclasses are arranged is a decision made by the programmer. In my example above, another programmer might have made the classes "car", "Volvos", "pre-Ford", and "Wagons". It depends on the problem you're trying to solve.
This is going to be a very simplified explanation. A class is a set of functions and variables and is used to create objects. I think it's good to use real examples instead of dog / bark / talk etc.
Class Email
Subject (string)
Message (string)
ToAddress (string)
FromAddress (string)
Send (function)
When you call 'new Email()' it creates a new object with those variables and functions. Then you can populate the variables and send it.
In object-oriented programming, a class is a type for objects. An object is a bundle of data together with functionality that can operate in the context of that data; the definition of what the object is and does when it is first created is determined by its class.
Like a type for data, the class of an object specifies what is common to all instances of that class. Instances, which are the objects themselves, then get to override that common baseline (otherwise there's not much point having distinct instances). In many OO systems, instances may or may not have new members that are not part of the class definition.
What that means in the context of a specific object-oriented language is going to differ from language to language. But if you think of classes as types, and build on that, you won't go far wrong.
A class is basically a way to organize your code.
It allows you to put all of the code related to one abstraction (think "concept" or "idea") in one place.
As an example - in your example of an app, the Window with the two buttons, a text box, and some code for handling what happens when the user types in the information may be organized into a single class: something like "PromptWindow". This class would be made up of multiple classes internally (two buttons, a textbox, etc) This would probably be used by some separate class, which would create an instance of the PromptWindow class, call a method on the class to show the window, then use the results.
At the very basis, there's procedural code:
var a = 4
var b = 5;
print a + b;
… and so on, statements following statements…
To make such pieces of code reusable, you make a function out of them:
function a_plus_b() {
var a = 4
var b = 5;
print a + b;
}
Now you can use that piece of code as often as you want, but you only had to write it once.
Usually an application depends on a certain way of how pieces of code and variables have to work together. This data needs to be processed by that function, but cannot be processed by that other function.
function foo(data) {
…do something…
return data;
}
function bar(data) {
…do something else…
return data;
}
a = "some data";
b = 123456;
a = foo(a);
b = bar(b);
c = bar(a); // ERROR, WRONG DATA FOR FUNCTION
To help group these related parts together, there are classes.
class A {
var data = 'some data';
function foo() {
…do something…
return data;
}
}
The function foo now has a variable data that is "bundled" with it in the same class. It can operate on that variable without having to worry about that it may be the wrong kind of data. Also there's no way that data can accidentally end up in function bar, which is part of another class.
The only problem is, there's only one variable data here. The function is reusable, but it can only operate on one set of data. To solve this, you create objects (also called instances) from the class:
instance1 = new A();
instance2 = new A();
instance1 and instance2 both behave exactly like class A, they both know how to perform function foo (now called an instance method) and they both hold a variable data (now called an instance variable or attribute), but that variable data can hold different data for both instances.
That's the basics of classes and objects. How your particular "OK", "Cancel" dialog box is implemented is a different story. Both buttons could be linked to different methods of different classes, or just to different methods of the same class, or even to the same method of the same class. Classes are just a way to logically group code and data, how that's going to be used is up to you.
In a language agnostic fasion, I would describe a class as being an object that contains related information.
A person class would have methods like Talk(), Walk(), Eat(); and attributes like Height, Color, Language, etc.
A class is like a blueprint of things that you can instantiate, and also procedures that are related to each other.
If you have a Database class, you might have many methods related to databases that do all sorts of voodoo with a database.
A class is a bunch of code that defines an entity in your application. There may be many such entities, but the behaviour of each is the same, as defined by the class. The class will typically define fields, whose contents are local to the instances (or objects) of that class. It is these fields that provide the objects with state and make them distinguishable from one another.
To use your example, there might be a Button class that defines what a button is in your application. This class would then be instantiated twice to provide two objects: one for the "No" button and another for the "Yes" button. The Button class could have a Text field/property that defines what text it contains – this could be set to "No" and "Yes" on the appropriate Button instances to give them their different appearances.
As for the click behaviour of the buttons, this would typically be implemented via the observer pattern, in which the subject class (Button in this case) maintains a list of separate "observer" objects which it notifies whenever some event occurs (i.e. when the button is clicked).
You should look at some sample code, in your language of choice. Just reading about the concept of classes will not answer many questions.
For example, I could tell you that a class is a "blueprint" for an object. Using this class, you can instantiate multiple such objects, each one of them (potentially) having unique attributes.
But you didn't understand a thing, now, did you? Your example with the buttons is very limited. Classes have nothing to do with user interfaces or actions or whatever. They define a way of programming, just like functions/methods/whatever you want to call them do.
So, to give a concrete example, here's a class that defines a ball, written in Python:
class Ball:
color = ''
def __init__(self, color):
self.color = color
def bounce(self):
print "%s ball bounces" % self.color
blueBall = Ball("blue")
redBall = Ball("red")
blueBall.bounce()
redBall.bounce()
Running this produces the expected output:
blue ball bounces
red ball bounces
However, there is much more to classes than I described here. Once you understand what a class is, you go on to learn about constructors, destructors, inheritance and a lot of other good stuff. Break a leg :)
From the definition of Class at Wikipedia:
In object-oriented programming, a
class is a construct that is used as
a blueprint (or template) to create
objects of that class. This blueprint
describes the state and behavior that
the objects of the class all share. An
object of a given class is called an
instance of the class. The class that
contains (and was used to create) that
instance can be considered as the type
of that object, e.g. an object
instance of the "Fruit" class would be
of the type "Fruit".
A class usually represents a noun,
such as a person, place or (possibly
quite abstract) thing - it is a model
of a concept within a computer
program. Fundamentally, it
encapsulates the state and behavior of
the concept it represents. It
encapsulates state through data
placeholders called attributes (or
member variables or instance
variables); it encapsulates behavior
through reusable sections of code
called methods.
Your understanding of a Class isn't at all incorrect but to make things clear consider the following...
The Yes and No buttons plus the TextBox are usually specified within a class taking for example code written in C# (Microsoft .NET Framework). Let's name this class MyClass.
The actions the buttons cause are handled by what are called handlers (methods). You could write your code in such a way that when you click the Yes button something gets written in the TextBox.
To instantiate MyClass you'd do the following:
MyClass myClass = new MyClass();
myClass.ButtonYes += new EventHandler(YourMethodForYes);
myClass.ButtonNo += new EventHandler(YourMethodForNo);
myClass.TextBox.Text = "Yes button was clicked";
Hope you get the idea.
I wrote usually above because this cenario you described could be implemented in a number of ways. OOP gives you plenty of ways to accomplish the same task.
Besides the definition of Class I think that reading about Object Oriented Programming (OOP) can help you a lot to understand it even more. Take a look at Fundamental Concepts.

Act on base or subclass without RTTI or base class modification

I asked a similar question yesterday that was specific to a technology, but now I find myself wondering about the topic in the broad sense.
For simplicity's sake, we have two classes, A and B, where B is derived from A. B truly "is a" A, and all of the routines defined in A have the same meaning in B.
Let's say we want to display a list of As, some of which are actually Bs. As we traverse our list of As, if the current object is actually a B, we want to display some of Bs additional properties....or maybe we just want to color the Bs differently, but neither A nor B have any notion of "color" or "display stuff".
Solutions:
Make the A class semi-aware of B by basically including a method called isB() in A that returns false. B will override the method and return true. Display code would have a check like: if (currentA.isB()) B b = currentA;
Provide a display() method in A that B can override.... but then we start merging the UI and the model. I won't consider this unless there is some cool trick I'm not seeing.
Use instanceof to check if the current A object to be displayed is really a B.
Just add all the junk from B to A, even though it doesn't apply to A. Basically just contain a B (that does not inherit from A) in A and set it to null until it applies. This is somewhat attractive. This is similar to #1 I guess w/ composition over inheritance.
It seems like this particular problem should come up from time to time and have an obvious solution.
So I guess the question maybe really boils down to:
If I have a subclass that extends a base class by adding additional functionality (not just changing the existing behavior of the base class), am I doing something tragically wrong? It all seems to instantly fall apart as soon as we try to act on a collection of objects that may be A or B.
A variant of option 2 (or hybrid of 1 and 2) may make sense: after all, polymorphism is the standard solution to "Bs are As but need to behave differently in situation X." Agreed, a display() method would probably tie the model to the UI too closely, but presumably the different renderings you want at the UI level reflect semantic or behavioural differences at the model level. Could those be captured in a method? For example, instead of an outright getDisplayColour() method, could it be a getPriority() (for example) method, to which A and B return different values but it is still up to the UI to decide how to translate that into a colour?
Given your more general question, however, of "how can we handle additional behaviour that we can't or won't allow to be accessed polymorphically via the base class," for example if the base class isn't under our control, your options are probably option 3, the Visitor pattern or a helper class. In both cases you are effectively farming out the polymorphism to an external entity -- in option 3, the UI (e.g. the presenter or controller), which performs an instanceOf check and does different things depending on whether it's a B or not; in Visitor or the helper case, the new class. Given your example, Visitor is probably overkill (also, if you were not able/willing to change the base class to accommodate it, it wouldn't be possible to implement it I think), so I'd suggest a simple class called something like "renderer":
public abstract class Renderer {
public static Renderer Create(A obj) {
if (obj instanceOf B)
return new BRenderer();
else
return new ARenderer();
}
public abstract Color getColor();
}
// implementations of ARenderer and BRenderer per your UI logic
This encapsulates the run-time type checking and bundles the code up into reasonably well-defined classes with clear responsibilities, without the conceptual overhead of Visitor. (Per GrizzlyNyo's answer, though, if your hierarchy or function set is more complex than what you've shown here, Visitor could well be more appropriate, but many people find Visitor hard to get their heads around and I would tend to avoid it for simple situations -- but your mileage may vary.)
The answer given by itowlson covers pretty well most part of the question. I will now deal with the very last paragraph as simply as I can.
Inheritance should be implemented for reuse, for your derived class to be reused in old code, not for your class reusing parts of the base class (you can use aggregation for that).
From that standpoint, if you have a class that is to be used on new code with some new functionality, but should be used transparently as a former class, then inheritance is your solution. New code can use the new functionality and old code will seamlessly use your new objects.
While this is the general intention, there are some common pitfals, the line here is subtle and your question is about precisely that line. If you have a collection of objects of type base, that should be because those objects are meant to be used only with base's methods. They are 'bases', behave like bases.
Using techniques as 'instanceof' or downcasts (dynamic_cast<>() in C++) to detect the real runtime type is something that I would flag in a code review and only accept after having the programmer explain to great detail why any other option is worse than that solution. I would accept it, for example, in itowlson's answer under the premises that the information is not available with the given operations in base. That is, the base type does not have any method that would offer enough information for the caller to determine the color. And if it does not make sense to include such operation: besides the prepresentation color, are you going to perform any operation on the objects based on that same information? If logic depends on the real type, then the operation should be in base class to be overriden in derived classes. If that is not possible (the operation is new and only for some given subtypes) there should at least be an operation in the base to allow the caller to determine that a downcast will not fail. And then again, I would really require a sound reason for the caller code to require knowledge of the real type. Why does the user want to see it in different colors? Will the user perform different operations on each one of the types?
If you endup requiring to use code to bypass the type system, your design has a strange smell to it. Of course, never say never, but you can surely say: avoid depending on instanceof or downcasts for logic.
This looks like text book case for the Visitor design pattern (also known as "Double Dispatch").
See this answer for link to a thorough explanation on the Visitor and Composite patterns.

"Has a" vs "Is a" - code smells for deciding

I wrote this yesterday, in a class Foo inheriting from Bar:
public override void AddItem(double a, int b)
{
//Code smell?
throw new NotImplementedException("This method not usable for Foo items");
}
Wondered subsequently if this is a possible indication that I should be using a Bar, rather than inheriting from it.
What other 'code smells' could help to chose between inheritance and composition?
EDIT I should add that this is a snippet, there are other methods which are in common, I just didn't want to go into too much detail. I have to analyse the implications of switching to composition, and wondered if there might be other 'code smells' which could help to tip the balance.
The example you gave above is clearly a code smell. The AddItem method is a behaviour of the base class Bar. If Foo doesn't support the AddItem behaviour, it shouldn't inherit from Bar.
Let's think of a more realistic (C++) example. Let's say you had the following class:
class Animal
{
void Breathe() const=0;
}
class Dog : public Animal
{
// Code smell
void Breathe() { throw new NotSupportedException(); }
}
The base abstract class Animal provides a pure virtual Breathe() method, because an animal must breathe to survive. If it doesn't breathe, then by definition it is not an animal.
By creating a new class Dog which inherits from Animal but doesn't support the Breathe() behaviour, you are breaking the contract stipulated by the Animal class. The poor dog won't survive!
The simple rule for public inheritance is that you should only do it if the derived class object truly "is a" base class object.
In your particular example:
Foo doesn't support the AddItem() behaviour stipulated by the Bar contract.
Therefore, by definition, Foo is "not a" Bar, and shouldn't inherit from it.
So why would you inherit from Bar, if Foo without extensions doesn't behave like Bar?
Come to think of it, I wouldn't even declare a method like 'AddItem' as virtual in the base class.
No! A square might be a rectangle, but
a Square object is definitely not a
Rectangle object. Why? Because the
behavior of a Square object is not
consistent with the behavior of a
Rectangle object. Behaviorally, a
Square is not a Rectangle! And it is
behavior that software is really all
about.
From The Liskov Substitution Principle in Object Mentor.
Yes, that you have to "un-implement" methods is an indication that perhaps you shouldn't use an "is-a" relationship. Your Foos don't seem to really be Bars.
But start by thinking about your Foos and Bars. Are Foos Bars? Can you draw sets and subsets on a paper, and will every Foo (that is, every memeber of the Foo class) also be a Bar (that is, a member of the Bar class)? If not, you probably don't want to use inheritance.
Another code smell that indicates that Foos aren't really Bars, and that Foo shouldn't inherit Bar, is that you can't use polymorphism. Lets say you have a method that takes a Bar as an argument, but it won't be able to handle a Foo. (Perhaps because it calls the AddItem method in its argument!) You'll have to add a check, or handle the NotImplementedException, making the code complicated and hard to understand. (And smelling!)
Of course there's the direct converse, if your Foo implements a lot of interface that just forwards messages to the Bar that it owns, this could indicate that it should be a Bar. Only could, though, smells aren't always correct.
While the example above probably indicates that something goes wrong, NotImplementedException itself is not wrong always. It's all about the contract of superclass and about subclass implementing this contract. If your superclass has such method
// This method is optional and may be not supported
// If not supported it should throw NotImplementedException
// To find out if it is supported or not, use isAddItemSupported()
public void AddItem(double a, int b){...}
Then, not supporting this method is still ok with the contract. If it is not supported, you probably should disable corresponding action in UI. So if you are ok with such contract, then subclass implements it correctly.
Another option when client explicitly declares that it does not use all class methods and is never going to. Like this
// This method never modifies orders, it just iterates over
// them and gets elements by index.
// We decided to be not very bureaucratic and not define ReadonlyList interface,
// and this is closed-source 3rd party library so you can not modify it yourself.
// ha-ha-ha
public void saveOrders(List<Order> orders){...}
Then it is ok to pass there implementaion of List that does not support add,remove and other mutators. Just document it.
// Use with care - this implementation does not implement entire List contract
// It does not support methods that modify the content of the list.
public ReadonlyListImpl implements List{...}
While having your code to define all your contract is good, as it let's your compiler to check if you violate the contract, sometimes it is not reasonable and you have to resort to weakly defined contracts, like comments.
In short words it comes to the question, if you really can safely use your subclass as superclass, taking into account that superclass is defined by its contract which is not only code.
The .Net framework has examples like this, particularly in the System.IO namespace - some readers don't implements all of their base class properties / methods and will throw exceptions if you try to use them.
E.g. Stream has a position property, but some streams don't support this.
Composition is preferable to inheritance as it reduces complexity. Better approach would be to use constructor injection and keep a reference to Bar in your Foo class.