I was wondering how associations, dependencies and such relations inherit in UML (or let's say, in general).
So, in a situation like this:
┌──────────┐ ┌──────────┐
│ ClassA │ │ ClassB │
├──────────┤ ├──────────┤
│ │─────────"One kind of relation"────────>│ │
├──────────┤ ├──────────┤
│ │ │ │
└──────────┘ └──────────┘
^
/┬\
│
│
│
│
┌─────┴────┐
│ ClassC │
├──────────┤
│ │
├──────────┤
│ │
└──────────┘
Note:
ClassA-ClassC is in a generalization relation, the arrow is meant to be solid
ClassA-ClassB is in one of [dependency, association, aggregation, composition]
Unicode is cool, but looked much better with the editor font :)
My question is, how do these relations inherit? For example, when ClassA is dependent on ClassB, will ClassC depend on ClassB? etc.
Thank you.
The simple answer is yes (and you should not need to look beyond that for most practical purpose).
But the matter is more complicated than it looks; quoting The Unified Modeling Language Reference Manual, Second Edition:
The words generalization and
inheritance are often used interchangeably, but there are
actually two related but distinct
concepts. Generalization is a
taxonomic relationship among modeling
elements. It describes what an element
is. Inheritance is a mechanism for combining shared incremental
descriptions to form a full
description of an element. In most
object-oriented systems, inheritance
is based on generalization, but
inheritance can be based on other
concepts, such as the delegation
pointer of the Self language. Basing
the inheritance mechanism on the
generalization relationship enables
factoring and sharing of descriptions
and polymorphic behavior. This is the
approach taken by most object-oriented
languages and by UML. But keep in mind
that there are other approaches that
could have been taken and that are
used by some programming languages.
I remember a pretty long lecture, back in 2003, on the difference between generalization and inheritance. In short, these two concepts belong to different levels of software design, or quoting Martin Fowler in UML Distilled, Third Edition, "various perspectives of modeling":
Conceptually, we can say that a
Corporate Customer is a subtype of
Customer if all instances of Corporate
Customer are also, by definition,
instances of Customer. A Corporate
Customer is then a special kind of
Customer.
The concept of generalization belongs to the conceptual, design level.
But inheritance is a concept that belongs to the implementation perspective:
With a software perspective, the
obvious interpretation is inheritance:
The Corporate Customer is a subclass
of Customer. In mainstream OO
languages, the subclass inherits all
the features of the superclass and may
override any superclass methods.
I remember an example where the difference between generalization and inheritance really made sense:
A square is a rectangle. That comes from their definitions in mathematics:
a rectangle is a quadrilateral with four right angles
a square is a polygon with four equal sides and angles
At design level, there is a generalization relationship between the square and the rectangle.
But at the implementation level, things beg to differ:
a rectangle can be defined by two measures: its width and height
a square can be defined by a single measure, since all sides are equal
It would be counterproductive to inherit the width and height of the Rectangle class in the Square class, which would require extra constraints to ensure that the same value is provided to a setter for the width and a setter for the height.
This is a case (somewhat edgy I admit) where you would not translate a generalization relationship to the use of inheritance at the implementation level.
Well, those were the days.
You're not asking a UML question, you're asking a more general question.
What does inheritance mean?
ClassC is a subclass of ClassA. In every programming language that implements inheritance, ClassC will have all the features of ClassA.
In a language that doesn't support inheritance, you must create the illusion of proper inheritance by assuring that all features of ClassA are also part of ClassC.
That's the definition of inheritance. Always and forever. Even in UML diagrams.
Since inheritance is a "is-a" (no pun intended) relationship, you can read it as "ClassC is a ClassA, which knows a ClassB", so:
For example, when ClassA is dependent on ClassB, will ClassC depend on ClassB?
-- Yes :)
Related
http://tutorials.jenkov.com/ood/understanding-dependencies.html#whatis says that (emphasis mine):
Whenever a class A uses another class or interface B, then A depends on B. A cannot carry out it's work without B, and A cannot be
reused without also reusing B. In such a situation the class A is
called the "dependant" and the class or interface B is called the
"dependency". A dependant depends on its dependencies.
According to Google's definition:
a dependent or subordinate thing, especially a country or province
controlled by another.
Since A depends on B, the dependency is A and the dependant is B. This is contrary to the first quote. Or am I missing something?
Terminology is context dependent. In general, the dependency is a synonym of dependence and describes not one entity, but state. According to Cambridge dictionary:
the situation in which you need something or someone all the time, especially in order to continue existing or operating.
So it's more about a relation to something. For example:
A minor child is dependent of his/her parents. The child is in a state
of dependency. The dependency describes the relation between the child and its parents.
The google definition is only one of the available definitions (see for example the Collins dictionary), the one that is especially applicable in the context of geopolitics and real estate. For example:
Before 1776, America was a dependency of the British Empire. The
dependency America was dependent on the Empire.
He bought a farm with several dependencies
The object oriented terminology is not so different from the more general definition: a dependency is a relationship between objects or classes. Here a more formal definition from the UML 2.5 standard:
7.7.1: A Dependency signifies a supplier/client relationship between model
elements where the modification of a supplier may impact the client
model elements.
So, restating what's in the tutorial, we can say that:
If A has to use B, A depends on B. There is a relation of dependency
from A to B. A is said to be the dependent.
For practical reason and language convenience, B is often called the dependency (which is indeed the contrary of everyday's language). But in reality, it is still the relation of A to B that is the dependency.
The thing that is dependent on something is the dependency. I am dependent on H. I need H injected into me. H is a dependency of me.
Composition: A class can have references to objects of other classes as members. This is called composition and is sometimes referred to as a has-a relationship.
By Deitel P.J., Deitel H.M. - Java How to Program 9th Edition.
This viewpoint is discussed in this topic:
Prefer composition over inheritance?
Composition: Composite aggregation (composition) is a "strong" form of aggregation with the following characteristics:
*it is binary association,
*it is a whole/part relationship,
*a part could be included in at most one composite (whole) at a time, and
*if a composite (whole) is deleted, all of its composite parts are "normally" deleted with it.
Found on http://www.uml-diagrams.org/composition.html
(actually, Deitel presents UML examples following this idea, in the same book, but did not bother to explain the difference).
This viewpoint is discussed in this topic:
What is the difference between association, aggregation and composition?
Fine, BOTH ARE CORRECT. And this introduces the problem of homonym concepts.
For instance: don't draw a UML model with composition arrows to exemplify the first definition: In UML, any association is a composition by Deitels' the first definition.
Here are some aspects of my question that may help in the correct answer:
How I can say (and know) which composition are we talking about?
Where we draw the line between the two definitions (in contextual terms)?
Can I say that the first is object oriented programming and the second is software engineering/modeling?
Is the UML composition a model-only concept/jargon?
Is the UML composition an UML exclusive thing? or is also applied in the programming field?
How to avoid miscommunication of "what composition are we talking about" in a team?
Please, answer with references, evidences, it is not a philosophical/opinion problem, it is a "scope" problem that I´m trying to address.
And it is not "what is composition" question.
Edit: I´m thinking if the distinction is verb x adjective: "to compose" a class (first def.) and "a composite relation" (second def.).
I found it hard to explain the difference between UML association and implementation references without explaining at least a little bit what UML associations actually are, and what they can do, so here we go.
Association & Link
Lets start by looking at what a UML Association and a link (Association's instance) are.
[11.5.3.1] An Association specifies a semantic relationship that can occur between typed instances.
[11.8.1.1] A link is a tuple of values that refer to typed objects. An Association classifies a set of links, each of which is an instance of the Association. Each value in the link refers to an instance of the type of the corresponding end of the Association.
So the following is a valid implementation of a limited association.
class Brain { }
class Head { }
a = new Brain;
b = new Head;
link = (new Array).add(a).add(b);
Ownership
[9.5.3] When a Property is owned by a Classifier other than an Association via ownedAttribute, then it represents an attribute of the Classifier.
(Note: Class is a subclass of a Classifier.)
Navigability
[11.5.3.1] An end Property of an Association that is owned by an end Class or that is a navigableOwnedEnd of the Association indicates that the Association is navigable from the opposite ends; otherwise, the Association is not navigable from the opposite ends. Navigability means that instances participating in links at runtime (instances of an Association) can be accessed efficiently from instances at the other ends of the Association. The precise mechanism by which such efficient access is achieved is implementation specific. If an end is not navigable, access from the other ends may or may not be possible, and if it is, it might not be efficient.
Why are those concepts relevant? Imagine the following example.
We see that brain is an attribute of Head class (the black dot signifies ownership by the opposite Class), and that it is navigable (the arrow).
We also see that head is NOT an attribute of Brain (no black dot ⇒ not owned by the Brain class ⇒ not an attribute of Brain), however it is still navigable. This means that in UML the head Property is held by the association itself.
The implementation could, for example, look like this (the association itself is represented by a tuple of two references (see link description earlier)).
class Head {
public Brain brain;
}
class Brain {
}
h = new Head;
b = new Brain;
h.brain = b;
link = (new Array).add(h).add(b);
So as you hopefully start to see, UML association is not such a simple concept as a has-a relationship.
Composition
Lets add another piece, composition.
[11.5.3.1] A binary Association may represent a composite aggregation (i.e., a whole/part relationship). Composition is represented by the isComposite attribute
[9.9.17] The value of isComposite is true only if aggregation is composite.
With the aggregation being
none - Indicates that the Property has no aggregation semantics.
shared - Indicates that the Property has shared aggregation semantics. Precise semantics of shared aggregation varies by application area and modeler.
composite -- Indicates that the Property is aggregated compositely, i.e., the composite object has responsibility for the existence and storage of the composed objects
Again we see, that a UML association is explicitly specifying concepts that are hard to perceive from implementation (e.g. who is responsible for object management/destruction).
Model Composition vs Object Implementation Composition
So from the description above we can construct a more precise description of what an implementation composition (has-a relationship) would be.
[Deteils] Composition: A class can have references to objects of other classes as members. This is called composition and is sometimes referred to as a has-a relationship.
McConnell [Code Complete 2, 6.3] also refers to has-a relationship as a Containment.
Neither of them however talk about HOW the objects (container-contained, composer-composite) are related to one another, who is responsible for lifecycles, or whether the contained element knows about the container.
So just by saying that objects have a has-a relationship (and call it composition), you could actually mean any of these (and several more)
So if you call something composition in programming, you can mean pretty much any relationship/reference (or rather not an inheritance), so the word by itself is not very useful.
In UML on the other hand you are trying to capture all such information about how the objects are related to one another. Therefore there's a focus on giving terms a more precise meaning. So when you call something composition in UML you have in mind a very specific has-a relationship, where the container is responsible for the lifecycle of the contained items.
Implementation of UML associations
All those extra concepts information mean that there is really no precise way how to even implement associations. This makes sense as the implementation would depend on the target programming language or environment (e.g. executable models, where the UML concepts are used as the final product).
As an example I can recommend a paper describing UML association implementation in Java with enforced concepts such as multiplicity, navigability, and visibility Implementing UML Associations in Java.
More subquestions
How I can say (and know) which composition are we talking about?
By context, or you can just ask (which is always a good thing to do when unsure). Personally I've heard the use of composition as "has-a relationship" only when differentiating from inheritance; and in the rest in terms of UML. But then again I am in academia, so my view is biased.
Where we draw the line between the two definitions (in contextual terms)?
As the "programming" term composition doesn't actually mean anything (only that it is has-a), I'd recommend drawing the line yourself and pushing others to use more precise terminology.
Can I say that the first is object oriented programming and the second is software engineering/modeling?
More or less, with all the nuances mentioned in this answer.
Is the UML composition a model-only concept/jargon?
Is the UML composition an UML exclusive thing? or is also applied in the programming field?
No, you can use it in programming to mean the same thing as it means in UML, but you might need to state it more obviously. E.g. "This class is a composite for those classes, because it manages their lifecycle.".
The point is to teach people to differentiate between regular-old has-a relationships, and relationships that have more precise semantics.
How to avoid miscommunication of "what composition are we talking about" in a team?
This is a very broad question that you could apply to any term to which you want attach special meaning (what even is software engineering?), and there is no best way. Have a team-shared vocabulary (you are probably already having a lots of specific terms in your domain), and guide people to use more precise terminology.
numbered quotes refers to sections in UML 2.5 Specifications.
To cite the UML 2.5 specification on page 110:
Sometimes a Property is used to model circumstances in which one instance is used to group together a set of instances; this is called aggregation. To represent such circumstances, a Property has an aggregation property, of type AggregationKind; the instance representing the whole group is classified by the owner of the Property, and the instances representing the grouped individuals are classified by the type of the Property. AggregationKind is an enumeration with the following literal values:
none: Indicates that the Property has no aggregation semantics.
shared: Indicates that the Property has shared aggregation semantics. Precise semantics of shared aggregation varies by application area and modeler.
composite: Indicates that the Property is aggregated compositely, i.e., the composite object has responsibility for the existence and storage of the composed objects (see the definition of parts in 11.2.3).
Personally I see it the way that notion of a composite aggregation is about object lifetime, not about static relation. A composite aggregation kills aggregate members when their parent dies. None leaves this open. And shared aggregation is a bastard that OMG should not have introduced at all since it's semantics is domain dependent.
I don't understand what "CBO-Coupling between object classes" really means. The definition I found is so short that I think I'm missing something, so it would be great if you help me with an example.
Here is the definition I found:
"The coupling between object classes is a count of the number of other classes to which it is coupled."
Thanks in advance.
Coupling between objects (CBO) is a count of the number of classes that are coupled to a particular class i.e. where the methods of one class call the methods or access the variables of the other. These calls need to be counted in both directions so the CBO of class A is the size of the set of classes that class A references and those classes that reference class A. Since this is a set - each class is counted only once even if the reference operates in both directions i.e. if A references B and B references A, B is only counted once.
This is the definition given here - www.virtualmachinery.com/sidebar3.htm
There is some more detail in the link - as well as an interesting general discussion of the Chidamber and Kemerer metrics - CBO is a part of these metrics.
Here's an example with UML that complements the other answers:
Notes:
CBO doesn't care about the direction of a dependency. D has a CBO of 1 because C depends on it, even though D depends on no other classes. B and C are similar cases.
Coupling can be via attributes (composition), associations, local variables, instanciations or injected dependencies (arguments to methods).
Coupling is when a class (A) depends (knows about, requires, uses) on another specific class(B). This means when you change a public member B that is used by A, you have to change A as well. You want low coupling between types, so that you can change classes without many side effects. Usually, coupling 'comes' together with bad encapsulation so you'll have A knowing information that should be private to B.
Some types are generic enough (like List in C#) and you can use them directly without fearing side effects. But whatever classes you define for your own app, you need to be aware that those might change. So in many situations, you are more interested in some behaviour (or attributes) of B, instead of A using the whole B. In those cases, it's better to extract an interface (to abstract the desired behaviour) and then A will know only about an abstraction, while B will implement it. This allows you to have more than one concrete implementation (useful every time you're dealing with things like databases, network, import/export etc) and A won't know about B.
Thus, A can unknowingly use any of B,C,D etc as long as they implement the interface and you can change things in B,C,D as long as this doesn't break the public contract (the interface).
While we usually want our classes to be decoupled, but cohesive (as in to work together), in many situations coupling won't really hurt you, as decoupling might require more effort than provide value. It's up to the developer to identify those situations and to make a proper decision. However, this comes with experience, so in the mean time, just try not to couple your classes too much.
This is more of a subjective question, so I'm going to preemptively mark it as community wiki.
Basically, I've found that in most of my code, there are many classes, many of which use each other, but few of which are directly related to each other. I look back at my college days, and think of the traditional class Cat : Animal type examples, where you have huge inheritance trees, but I see none of this in my code. My class diagrams look like giant spiderwebs, not like nice pretty trees.
I feel I've done a good job of separating information logically, and recently I've done a good job of isolating dependencies between classes via DI/IoC techniques, but I'm worried I might be missing something. I do tend to clump behavior in interfaces, but I simply don't subclass.
I can easily understand subclassing in terms of the traditional examples such as class Dog : Animal or class Employee : Person, but I simply don't have anything that obvious I'm dealing with. And things are rarely as clear-cut as class Label : Control. But when it comes to actually modeling real entities in my code as a hierarchy, I have no clue where to begin.
So, I guess my questions boil down to this:
Is it ok to simply not subclass or inherit? Should I be concerned at all?
What are some strategies you have to determine objects that could benefit from inheritance?
Is it acceptable to always inherit based on behavior (interfaces) rather than the actual type?
Inheritance should always represent an "is-a" relationship. You should be able to say "A is a B" if A derives from B. If not, prefer composition. It's perfectly fine to not subclass when it is not necessary.
For example, saying that FileOpenDialog "is-a" Window makes sense, but saying that an Engine "is-a" Car is nonsense. In that case, an instance of Engine inside a Car instance is more appropriate (It can be said that Car "is-implemented-in-terms-of" Engine).
For a good discussion of inheritance, see Part 1 and Part 2 of "Uses and Abuses of Inheritance" on gotw.ca.
As long as you do not miss the clear cut 'is a' relationships, it's ok and in fact, it's best not to inherit, but to use composition.
is-a is the litmus test. if (Is X a Y?) then class X : Y { } else class X { Y myY; } or class Y { X myX; }
Using interfaces, that is, inheriting behavior, is a very neat way to structure the code via adding only the needed behavior and no other. The tricky part is defining those interfaces well.
No technology or pattern should be used for its own sake. You obviously work in a domain where classes tend to not benefit from inheritance, so you shouldn't use inheritance.
You've used DI to keep things neat and clean. You separated the concerns of your classes. Those are all good things. Don't try and force inheritance if you don't really need it.
An interesting follow-up to this question would be: Which programming domains do tend to make good use of inheritance? (UI and db frameworks have already been mentioned and are great examples. Any others?)
I also hate the Dog -> Mammal -> Animal examples, precisely because they do not occur in real life.
I use very little subclassing, because it tightly couples the subclass to the superclass and makes your code really hard to read. Sometimes implementation inheritance is useful (e.g. PostgreSQLDatabaseImpl and MySQLDatabaseImpl extend AbstractSQLDatabase), but most of the time it just makes a mess of things. Most of the time I see subclasses the concept has been misused and either interfaces or a property should be used.
Interfaces, however, are great and you should use those.
Generally, favour composition over inheritance. Inheritance tends to break encapsulation. e.g. If a class depends on a method of a super class and the super class changes the implementation of that method in some release, the subclass may break.
At times when you are designing a framework, you will have to design classes to be inherited. If you want to use inheritance, you will have to document and design for it carefully. e.g. Not calling any instance methods (that could be overridden by your subclasses) in the constructor. Also if its a genuine 'is-a' relationship, inheritance is useful but is more robust if used within a package.
See Effective Java (Item 14, and 15). It gives a great argument for why you should favour composition over inheritance. It talks about inheritance and encapsulation in general (with java examples). So its a good resource even if you are not using java.
So to answer your 3 questions:
Is it ok to simply not subclass or inherit? Should I be concerned at all?
Ans: Ask yourself the question is it a truly "is-a" relationship? Is decoration possible? Go for decoration
// A collection decorator that is-a collection with
public class MyCustomCollection implements java.util.Collection {
private Collection delegate;
// decorate methods with custom code
}
What are some strategies you have to determine objects that could benefit from inheritance?
Ans: Usually when you are writing a framework, you may want to provide certain interfaces and "base" classes specifically designed for inheritance.
Is it acceptable to always inherit based on behavior (interfaces) rather than the actual type?
Ans: Mostly yes, but you'd be better off if the super class is designed for inheritance and/or under your control. Or else go for composition.
IMHO, you should never do #3, unless you're building an abstract base class specifically for that purpose, and its name makes it clear what its purpose is:
class DataProviderBase {...}
class SqlDataProvider : DataProviderBase {...}
class DB2DataProvider : DataProviderBase {...}
class AccountDataProvider : SqlDataProvider {...}
class OrderDataProvider : SqlDataProvider {...}
class ShippingDataProvider : DB2DataProvider {...}
etc.
Also following this type of model, sometimes if you provide an interface (IDataProvider) it's good to also provide a base class (DataProviderBase) that future consumers can use to conveniently access logic that's common to all/most DataProviders in your application model.
As a general rule, though, I only use inheritance if I have a true "is-a" relationship, or if it will improve the overall design for me to create an "is-a" relationship (provider model, for instance.)
Where you have shared functionality, programming to the interface is more important than inheritance.
Essentially, inheritance is more about relating objects together.
Most of the time we are concerned with what an object can DO, as opposed to what it is.
class Product
class Article
class NewsItem
Are the NewsItem and Article both Content items? Perhaps, and you may find it useful to be able to have a list of content which contains both Article items and NewsItem items.
However, it's probably more likely you'll have them implement similar interfaces. For example, IRssFeedable could be an interface that they both implement. In fact, Product could also implement this interface.
Then they can all be thrown to an RSS Feed easily to provide lists of things on your web page. This is a great example when the interface is important whereas the inheritance model is perhaps less useful.
Inheritance is all about identifying the nature of Objects
Interfaces are all about identifying what Objects can DO.
My class hierarchies tend to be fairly flat as well, with interfaces and composition providing the necessary coupling. Inheritance seems to pop up mostly when I'm storing collections of things, where the different kinds of things will have data/properties in common. Inheritance often feels more natural to me when there is common data, whereas interfaces are a very natural way to express common behavior.
The answer to each of your 3 questions is "it depends". Ultimately it will all depend on your domain and what your program does with it. A lot of times, I find the design patterns I choose to use actually help with finding points where inheritance works well.
For example, consider a 'transformer' used to massage data into a desired form. If you get 3 data sources as CSV files, and want to put them into three different object models (and maybe persist them into a database), you could create a 'csv transformer' base and then override some methods when you inherit from it in order to handle the different specific objects.
'Casting' the development process into the pattern language will help you find objects/methods that behave similarly and help in reducing redundant code (maybe through inheritance, maybe through the use of shared libraries - whichever suits the situation best).
Also, if you keep your layers separate (business, data, presentation, etc.), your class diagram will be simpler, and you could then 'visualize' those objects that aught to be inherited.
I wouldn't get too worried about how your class diagram looks, things are rarely like the classroom...
Rather ask yourself two questions:
Does your code work?
Is it extremely time consuming to maintain? Does a change sometimes require changing the 'same' code in many places?
If the answer to (2) is yes, you might want to look at how you have structured your code to see if there is a more sensible fashion, but always bearing in mind that at the end of the day, you need to be able to answer yes to question (1)... Pretty code that doesn't work is of no use to anybody, and hard to explain to the management.
IMHO, the primary reason to use inheritance is to allow code which was written to operate upon a base-class object to operate upon a derived-class object instead.
As it currently stands, this question is not a good fit for our Q&A format. We expect answers to be supported by facts, references, or expertise, but this question will likely solicit debate, arguments, polling, or extended discussion. If you feel that this question can be improved and possibly reopened, visit the help center for guidance.
Closed 10 years ago.
There are two schools of thought on how to best extend, enhance, and reuse code in an object-oriented system:
Inheritance: extend the functionality of a class by creating a subclass. Override superclass members in the subclasses to provide new functionality. Make methods abstract/virtual to force subclasses to "fill-in-the-blanks" when the superclass wants a particular interface but is agnostic about its implementation.
Aggregation: create new functionality by taking other classes and combining them into a new class. Attach an common interface to this new class for interoperability with other code.
What are the benefits, costs, and consequences of each? Are there other alternatives?
I see this debate come up on a regular basis, but I don't think it's been asked on
Stack Overflow yet (though there is some related discussion). There's also a surprising lack of good Google results for it.
It's not a matter of which is the best, but of when to use what.
In the 'normal' cases a simple question is enough to find out if we need inheritance or aggregation.
If The new class is more or less as the original class. Use inheritance. The new class is now a subclass of the original class.
If the new class must have the original class. Use aggregation. The new class has now the original class as a member.
However, there is a big gray area. So we need several other tricks.
If we have used inheritance (or we plan to use it) but we only use part of the interface, or we are forced to override a lot of functionality to keep the correlation logical. Then we have a big nasty smell that indicates that we had to use aggregation.
If we have used aggregation (or we plan to use it) but we find out we need to copy almost all of the functionality. Then we have a smell that points in the direction of inheritance.
To cut it short. We should use aggregation if part of the interface is not used or has to be changed to avoid an illogical situation. We only need to use inheritance, if we need almost all of the functionality without major changes. And when in doubt, use Aggregation.
An other possibility for, the case that we have an class that needs part of the functionality of the original class, is to split the original class in a root class and a sub class. And let the new class inherit from the root class. But you should take care with this, not to create an illogical separation.
Lets add an example. We have a class 'Dog' with methods: 'Eat', 'Walk', 'Bark', 'Play'.
class Dog
Eat;
Walk;
Bark;
Play;
end;
We now need a class 'Cat', that needs 'Eat', 'Walk', 'Purr', and 'Play'. So first try to extend it from a Dog.
class Cat is Dog
Purr;
end;
Looks, alright, but wait. This cat can Bark (Cat lovers will kill me for that). And a barking cat violates the principles of the universe. So we need to override the Bark method so that it does nothing.
class Cat is Dog
Purr;
Bark = null;
end;
Ok, this works, but it smells bad. So lets try an aggregation:
class Cat
has Dog;
Eat = Dog.Eat;
Walk = Dog.Walk;
Play = Dog.Play;
Purr;
end;
Ok, this is nice. This cat does not bark anymore, not even silent. But still it has an internal dog that wants out. So lets try solution number three:
class Pet
Eat;
Walk;
Play;
end;
class Dog is Pet
Bark;
end;
class Cat is Pet
Purr;
end;
This is much cleaner. No internal dogs. And cats and dogs are at the same level. We can even introduce other pets to extend the model. Unless it is a fish, or something that does not walk. In that case we again need to refactor. But that is something for an other time.
At the beginning of GOF they state
Favor object composition over class inheritance.
This is further discussed here
The difference is typically expressed as the difference between "is a" and "has a". Inheritance, the "is a" relationship, is summed up nicely in the Liskov Substitution Principle. Aggregation, the "has a" relationship, is just that - it shows that the aggregating object has one of the aggregated objects.
Further distinctions exist as well - private inheritance in C++ indicates a "is implemented in terms of" relationship, which can also be modeled by the aggregation of (non-exposed) member objects as well.
Here's my most common argument:
In any object-oriented system, there are two parts to any class:
Its interface: the "public face" of the object. This is the set of capabilities it announces to the rest of the world. In a lot of languages, the set is well defined into a "class". Usually these are the method signatures of the object, though it varies a bit by language.
Its implementation: the "behind the scenes" work that the object does to satisfy its interface and provide functionality. This is typically the code and member data of the object.
One of the fundamental principles of OOP is that the implementation is encapsulated (ie:hidden) within the class; the only thing that outsiders should see is the interface.
When a subclass inherits from a subclass, it typically inherits both the implementation and the interface. This, in turn, means that you're forced to accept both as constraints on your class.
With aggregation, you get to choose either implementation or interface, or both -- but you're not forced into either. The functionality of an object is left up to the object itself. It can defer to other objects as it likes, but it's ultimately responsible for itself. In my experience, this leads to a more flexible system: one that's easier to modify.
So, whenever I'm developing object-oriented software, I almost always prefer aggregation over inheritance.
I gave an answer to "Is a" vs "Has a" : which one is better?.
Basically I agree with other folks: use inheritance only if your derived class truly is the type you're extending, not merely because it contains the same data. Remember that inheritance means the subclass gains the methods as well as the data.
Does it make sense for your derived class to have all the methods of the superclass? Or do you just quietly promise yourself that those methods should be ignored in the derived class? Or do you find yourself overriding methods from the superclass, making them no-ops so no one calls them inadvertently? Or giving hints to your API doc generation tool to omit the method from the doc?
Those are strong clues that aggregation is the better choice in that case.
I see a lot of "is-a vs. has-a; they're conceptually different" responses on this and the related questions.
The one thing I've found in my experience is that trying to determine whether a relationship is "is-a" or "has-a" is bound to fail. Even if you can correctly make that determination for the objects now, changing requirements mean that you'll probably be wrong at some point in the future.
Another thing I've found is that it's very hard to convert from inheritance to aggregation once there's a lot of code written around an inheritance hierarchy. Just switching from a superclass to an interface means changing nearly every subclass in the system.
And, as I mentioned elsewhere in this post, aggregation tends to be less flexible than inheritance.
So, you have a perfect storm of arguments against inheritance whenever you have to choose one or the other:
Your choice will likely be the wrong one at some point
Changing that choice is difficult once you've made it.
Inheritance tends to be a worse choice as it's more constraining.
Thus, I tend to choose aggregation -- even when there appears to be a strong is-a relationship.
The question is normally phrased as Composition vs. Inheritance, and it has been asked here before.
I wanted to make this a comment on the original question, but 300 characters bites [;<).
I think we need to be careful. First, there are more flavors than the two rather specific examples made in the question.
Also, I suggest that it is valuable not to confuse the objective with the instrument. One wants to make sure that the chosen technique or methodology supports achievement of the primary objective, but I don't thing out-of-context which-technique-is-best discussion is very useful. It does help to know the pitfalls of the different approaches along with their clear sweet spots.
For example, what are you out to accomplish, what do you have available to start with, and what are the constraints?
Are you creating a component framework, even a special purpose one? Are interfaces separable from implementations in the programming system or is it accomplished by a practice using a different sort of technology? Can you separate the inheritance structure of interfaces (if any) from the inheritance structure of classes that implement them? Is it important to hide the class structure of an implementation from the code that relies on the interfaces the implementation delivers? Are there multiple implementations to be usable at the same time or is the variation more over-time as a consequence of maintenance and enhancememt? This and more needs to be considered before you fixate on a tool or a methodology.
Finally, is it that important to lock distinctions in the abstraction and how you think of it (as in is-a versus has-a) to different features of the OO technology? Perhaps so, if it keeps the conceptual structure consistent and manageable for you and others. But it is wise not to be enslaved by that and the contortions you might end up making. Maybe it is best to stand back a level and not be so rigid (but leave good narration so others can tell what's up). [I look for what makes a particular portion of a program explainable, but some times I go for elegance when there is a bigger win. Not always the best idea.]
I'm an interface purist, and I am drawn to the kinds of problems and approaches where interface purism is appropriate, whether building a Java framework or organizing some COM implementations. That doesn't make it appropriate for everything, not even close to everything, even though I swear by it. (I have a couple of projects that appear to provide serious counter-examples against interface purism, so it will be interesting to see how I manage to cope.)
I'll cover the where-these-might-apply part. Here's an example of both, in a game scenario. Suppose, there's a game which has different types of soldiers. Each soldier can have a knapsack which can hold different things.
Inheritance here?
There's a marine, green beret & a sniper. These are types of soldiers. So, there's a base class Soldier with Marine, Green Beret & Sniper as derived classes
Aggregation here?
The knapsack can contain grenades, guns (different types), knife, medikit, etc. A soldier can be equipped with any of these at any given point in time, plus he can also have a bulletproof vest which acts as armor when attacked and his injury decreases to a certain percentage. The soldier class contains an object of bulletproof vest class and the knapsack class which contains references to these items.
I think it's not an either/or debate. It's just that:
is-a (inheritance) relationships occur less often than has-a (composition) relationships.
Inheritance is harder to get right, even when it's appropriate to use it, so due diligence has to be taken because it can break encapsulation, encourage tight coupling by exposing implementation and so forth.
Both have their place, but inheritance is riskier.
Although of course it wouldn't make sense to have a class Shape 'having-a' Point and a Square classes. Here inheritance is due.
People tend to think about inheritance first when trying to design something extensible, that is what's wrong.
Favour happens when both candidate qualifies. A and B are options and you favour A. The reason is that composition offers more extension/flexiblity possiblities than generalization. This extension/flexiblity refers mostly to runtime/dynamic flexibility.
The benefit is not immediately visible. To see the benefit you need to wait for the next unexpected change request. So in most cases those sticked to generlalization fails when compared to those who embraced composition(except one obvious case mentioned later). Hence the rule. From a learning point of view if you can implement a dependency injection successfully then you should know which one to favour and when. The rule helps you in making a decision as well; if you are not sure then select composition.
Summary: Composition :The coupling is reduced by just having some smaller things you plug into something bigger, and the bigger object just calls the smaller object back. Generlization: From an API point of view defining that a method can be overridden is a stronger commitment than defining that a method can be called. (very few occassions when Generalization wins). And never forget that with composition you are using inheritance too, from a interface instead of a big class
Both approaches are used to solve different problems. You don't always need to aggregate over two or more classes when inheriting from one class.
Sometimes you do have to aggregate a single class because that class is sealed or has otherwise non-virtual members you need to intercept so you create a proxy layer that obviously isn't valid in terms of inheritance but so long as the class you are proxying has an interface you can subscribe to this can work out fairly well.