Should concrete implementation provide any public API not present in the interface it implements? - oop

"Code to interfaces" is considered good practice. Such code is easy to unit test and enables loose coupling. Users only know the interfaces and the onus of wiring concrete objects is upon the top-most level (this can be done in some init code or with the help of frameworks).
My question is about following the practice of code to interfaces: does it imply that a concrete class can never declare any public method which is not present in its interface?
Otherwise, it will force users to depend upon the concrete implementation. This will make such methods difficult for unit testing; if the test fails, determining if it failed due to an issue in the caller code or due to the concrete method will require extra effort. This will also break the Dependency Inversion Principle. It will induce type-checking and down-casting, which are considered bad practice.


That is totally acceptable provided that the new methods aren't crucial to the operating of the class, and in particular to how it functions when someone thinks of it as the superclass or interface.
ArrayList provides good examples. It has methods that let you manage its internal memory, like ensureCapacity(int) or trimToSize(). Those are sometimes helpful if you know you're working with an ArrayList and need to be more precise about memory allocation, but they're not required for the basic operation of the ArrayList, and in particular, they're not required for having it operate as a general List.
In fact, interfaces themselves can add new methods in this way. Consider NavigableSet, which extends Set. It adds a whole bunch of methods that rely on the ordering of the set's elements (give me the first, the last, a subtree starting from here, etc). None of those methods are defined on Set, and even the fact that the elements are ordered isn't defined by the Set contract; but the Set methods all work just fine without the additional methods and ordering.
The advice to "code to the interface" is a good start, but it's a bit over-generalized. A refinement of that advice would be, "code to the most general interface that you need." If you don't need ArrayLists's methods (or its contract, such as its random-access performance), code to List; but if you do need them, then by all means use them.


#yshavit's third paragraph hits it right. Implement an extension of the "not enough" base interface, as exampled with public interface NavigableSet<E> extends SortedSet<E> (which, BTW, extends Set<E> extends Collection<E> extends Iterable<E>).
It's his second paragraph that troubles me. Why have "non-crucial" methods of the API that are not surfaced in some interface being implemented? In the ArrayList example, why not have the size management methods declared in an interface? Perhaps ManagedSize which would describe clear behavior for ArrayList (and other) classes to implement, along with the several other interfaces it implements (my JRE source says: public class ArrayList<E> extends AbstractList<E> implements List<E>, RandomAccess, Cloneable, java.io.Serializable).
With such an approach, there is no need to decide which methods are "non-crucial," only to be surprised by some client code that depends on things like ensureSize to help avoid relocation during a time-critical phase, or trimToSize to release excessive overalloaction when it's algorthmically known that further growth will not be needed. Not that I'm promoting such algorthms as best practice, but even non-functional "behavior management" methods deserve their place in the light.
Finally, while I agree with sentiment of "Know Where the Lines Are, and yet Color As You See Fit" it doesn't give practical guidance. Here's attempt at such:
Always start by coding to an interface, ie. all concrete public methods should be declared in an interface:
Use multiple interfaces as needed
Each interface should partition the implemented API into coherent non-overlapping aspects, e.g. List, RandomAccess, Cloneable, Serializable
Tend to start with larger scoped interfaces and break them up as the design develops (before coding ala Waterfall, or as code evolves ala Agile); interfaces are one of the easier design artefacts to refactor.
If a given interface you are implementing is "insufficient":
Extend the base interface and add the methods you need, then implement that one, OR
Create an augmenting interface (like the ManagedSize idea, above) with just the additional methods and then implement them both
Only when you find you can't do that, then relax only as much of the rule as you need to make things work (often, this will be an experimental trial-error "does it work, yet?" cycle).
Reasons for #3's "can't" will vary, but I expect them to be external to the application design, e.g. the ORM I'm using becomes confused, the IDE plug-in doesn't refactor it correctly, the DSL translator I'm forced to use fails when a class implements more than three interfaces...

Related

[SELENIUM]Webdriver driver = new FirefoxDriver() cn anyone say how the constructor name is different from class name here [duplicate]

I have seen this mentioned a few times and I am not clear on what it means. When and why would you do this?
I know what interfaces do, but the fact I am not clear on this makes me think I am missing out on using them correctly.
Is it just so if you were to do:
IInterface classRef = new ObjectWhatever()
You could use any class that implements IInterface? When would you need to do that? The only thing I can think of is if you have a method and you are unsure of what object will be passed except for it implementing IInterface. I cannot think how often you would need to do that.
Also, how could you write a method that takes in an object that implements an interface? Is that possible?

There are some wonderful answers on here to this questions that get into all sorts of great detail about interfaces and loosely coupling code, inversion of control and so on. There are some fairly heady discussions, so I'd like to take the opportunity to break things down a bit for understanding why an interface is useful.
When I first started getting exposed to interfaces, I too was confused about their relevance. I didn't understand why you needed them. If we're using a language like Java or C#, we already have inheritance and I viewed interfaces as a weaker form of inheritance and thought, "why bother?" In a sense I was right, you can think of interfaces as sort of a weak form of inheritance, but beyond that I finally understood their use as a language construct by thinking of them as a means of classifying common traits or behaviors that were exhibited by potentially many non-related classes of objects.
For example -- say you have a SIM game and have the following classes:
class HouseFly inherits Insect {
void FlyAroundYourHead(){}
void LandOnThings(){}
}
class Telemarketer inherits Person {
void CallDuringDinner(){}
void ContinueTalkingWhenYouSayNo(){}
}
Clearly, these two objects have nothing in common in terms of direct inheritance. But, you could say they are both annoying.
Let's say our game needs to have some sort of random thing that annoys the game player when they eat dinner. This could be a HouseFly or a Telemarketer or both -- but how do you allow for both with a single function? And how do you ask each different type of object to "do their annoying thing" in the same way?
The key to realize is that both a Telemarketer and HouseFly share a common loosely interpreted behavior even though they are nothing alike in terms of modeling them. So, let's make an interface that both can implement:
interface IPest {
void BeAnnoying();
}
class HouseFly inherits Insect implements IPest {
void FlyAroundYourHead(){}
void LandOnThings(){}
void BeAnnoying() {
FlyAroundYourHead();
LandOnThings();
}
}
class Telemarketer inherits Person implements IPest {
void CallDuringDinner(){}
void ContinueTalkingWhenYouSayNo(){}
void BeAnnoying() {
CallDuringDinner();
ContinueTalkingWhenYouSayNo();
}
}
We now have two classes that can each be annoying in their own way. And they do not need to derive from the same base class and share common inherent characteristics -- they simply need to satisfy the contract of IPest -- that contract is simple. You just have to BeAnnoying. In this regard, we can model the following:
class DiningRoom {
DiningRoom(Person[] diningPeople, IPest[] pests) { ... }
void ServeDinner() {
when diningPeople are eating,
foreach pest in pests
pest.BeAnnoying();
}
}
Here we have a dining room that accepts a number of diners and a number of pests -- note the use of the interface. This means that in our little world, a member of the pests array could actually be a Telemarketer object or a HouseFly object.
The ServeDinner method is called when dinner is served and our people in the dining room are supposed to eat. In our little game, that's when our pests do their work -- each pest is instructed to be annoying by way of the IPest interface. In this way, we can easily have both Telemarketers and HouseFlys be annoying in each of their own ways -- we care only that we have something in the DiningRoom object that is a pest, we don't really care what it is and they could have nothing in common with other.
This very contrived pseudo-code example (that dragged on a lot longer than I anticipated) is simply meant to illustrate the kind of thing that finally turned the light on for me in terms of when we might use an interface. I apologize in advance for the silliness of the example, but hope that it helps in your understanding. And, to be sure, the other posted answers you've received here really cover the gamut of the use of interfaces today in design patterns and development methodologies.

The specific example I used to give to students is that they should write
List myList = new ArrayList(); // programming to the List interface
instead of
ArrayList myList = new ArrayList(); // this is bad
These look exactly the same in a short program, but if you go on to use myList 100 times in your program you can start to see a difference. The first declaration ensures that you only call methods on myList that are defined by the List interface (so no ArrayList specific methods). If you've programmed to the interface this way, later on you can decide that you really need
List myList = new TreeList();
and you only have to change your code in that one spot. You already know that the rest of your code doesn't do anything that will be broken by changing the implementation because you programmed to the interface.
The benefits are even more obvious (I think) when you're talking about method parameters and return values. Take this for example:
public ArrayList doSomething(HashMap map);
That method declaration ties you to two concrete implementations (ArrayList and HashMap). As soon as that method is called from other code, any changes to those types probably mean you're going to have to change the calling code as well. It would be better to program to the interfaces.
public List doSomething(Map map);
Now it doesn't matter what kind of List you return, or what kind of Map is passed in as a parameter. Changes that you make inside the doSomething method won't force you to change the calling code.

Programming to an interface is saying, "I need this functionality and I don't care where it comes from."
Consider (in Java), the List interface versus the ArrayList and LinkedList concrete classes. If all I care about is that I have a data structure containing multiple data items that I should access via iteration, I'd pick a List (and that's 99% of the time). If I know that I need constant-time insert/delete from either end of the list, I might pick the LinkedList concrete implementation (or more likely, use the Queue interface). If I know I need random access by index, I'd pick the ArrayList concrete class.

Programming to an interface has absolutely nothing to do with abstract interfaces like we see in Java or .NET. It isn't even an OOP concept.
What it means is don't go messing around with the internals of an object or data structure. Use the Abstract Program Interface, or API, to interact with your data. In Java or C# that means using public properties and methods instead of raw field access. For C that means using functions instead of raw pointers.
EDIT: And with databases it means using views and stored procedures instead of direct table access.

Using interfaces is a key factor in making your code easily testable in addition to removing unnecessary couplings between your classes. By creating an interface that defines the operations on your class, you allow classes that want to use that functionality the ability to use it without depending on your implementing class directly. If later on you decide to change and use a different implementation, you need only change the part of the code where the implementation is instantiated. The rest of the code need not change because it depends on the interface, not the implementing class.
This is very useful in creating unit tests. In the class under test you have it depend on the interface and inject an instance of the interface into the class (or a factory that allows it to build instances of the interface as needed) via the constructor or a property settor. The class uses the provided (or created) interface in its methods. When you go to write your tests, you can mock or fake the interface and provide an interface that responds with data configured in your unit test. You can do this because your class under test deals only with the interface, not your concrete implementation. Any class implementing the interface, including your mock or fake class, will do.
EDIT: Below is a link to an article where Erich Gamma discusses his quote, "Program to an interface, not an implementation."
http://www.artima.com/lejava/articles/designprinciples.html

You should look into Inversion of Control:
Martin Fowler: Inversion of Control Containers and the Dependency Injection pattern
Wikipedia: Inversion of Control
In such a scenario, you wouldn't write this:
IInterface classRef = new ObjectWhatever();
You would write something like this:
IInterface classRef = container.Resolve<IInterface>();
This would go into a rule-based setup in the container object, and construct the actual object for you, which could be ObjectWhatever. The important thing is that you could replace this rule with something that used another type of object altogether, and your code would still work.
If we leave IoC off the table, you can write code that knows that it can talk to an object that does something specific, but not which type of object or how it does it.
This would come in handy when passing parameters.
As for your parenthesized question "Also, how could you write a method that takes in an object that implements an Interface? Is that possible?", in C# you would simply use the interface type for the parameter type, like this:
public void DoSomethingToAnObject(IInterface whatever) { ... }
This plugs right into the "talk to an object that does something specific." The method defined above knows what to expect from the object, that it implements everything in IInterface, but it doesn't care which type of object it is, only that it adheres to the contract, which is what an interface is.
For instance, you're probably familiar with calculators and have probably used quite a few in your days, but most of the time they're all different. You, on the other hand, knows how a standard calculator should work, so you're able to use them all, even if you can't use the specific features that each calculator has that none of the other has.
This is the beauty of interfaces. You can write a piece of code, that knows that it will get objects passed to it that it can expect certain behavior from. It doesn't care one hoot what kind of object it is, only that it supports the behavior needed.
Let me give you a concrete example.
We have a custom-built translation system for windows forms. This system loops through controls on a form and translate text in each. The system knows how to handle basic controls, like the-type-of-control-that-has-a-Text-property, and similar basic stuff, but for anything basic, it falls short.
Now, since controls inherit from pre-defined classes that we have no control over, we could do one of three things:
Build support for our translation system to detect specifically which type of control it is working with, and translate the correct bits (maintenance nightmare)
Build support into base classes (impossible, since all the controls inherit from different pre-defined classes)
Add interface support
So we did nr. 3. All our controls implement ILocalizable, which is an interface that gives us one method, the ability to translate "itself" into a container of translation text/rules. As such, the form doesn't need to know which kind of control it has found, only that it implements the specific interface, and knows that there is a method where it can call to localize the control.

Code to the Interface Not the Implementation has NOTHING to do with Java, nor its Interface construct.
This concept was brought to prominence in the Patterns / Gang of Four books but was most probably around well before that. The concept certainly existed well before Java ever existed.
The Java Interface construct was created to aid in this idea (among other things), and people have become too focused on the construct as the centre of the meaning rather than the original intent. However, it is the reason we have public and private methods and attributes in Java, C++, C#, etc.
It means just interact with an object or system's public interface. Don't worry or even anticipate how it does what it does internally. Don't worry about how it is implemented. In object-oriented code, it is why we have public vs. private methods/attributes. We are intended to use the public methods because the private methods are there only for use internally, within the class. They make up the implementation of the class and can be changed as required without changing the public interface. Assume that regarding functionality, a method on a class will perform the same operation with the same expected result every time you call it with the same parameters. It allows the author to change how the class works, its implementation, without breaking how people interact with it.
And you can program to the interface, not the implementation without ever using an Interface construct. You can program to the interface not the implementation in C++, which does not have an Interface construct. You can integrate two massive enterprise systems much more robustly as long as they interact through public interfaces (contracts) rather than calling methods on objects internal to the systems. The interfaces are expected to always react the same expected way given the same input parameters; if implemented to the interface and not the implementation. The concept works in many places.
Shake the thought that Java Interfaces have anything what-so-ever to do with the concept of 'Program to the Interface, Not the Implementation'. They can help apply the concept, but they are not the concept.

It sounds like you understand how interfaces work but are unsure of when to use them and what advantages they offer. Here are a few examples of when an interface would make sense:
// if I want to add search capabilities to my application and support multiple search
// engines such as Google, Yahoo, Live, etc.
interface ISearchProvider
{
string Search(string keywords);
}
then I could create GoogleSearchProvider, YahooSearchProvider, LiveSearchProvider, etc.
// if I want to support multiple downloads using different protocols
// HTTP, HTTPS, FTP, FTPS, etc.
interface IUrlDownload
{
void Download(string url)
}
// how about an image loader for different kinds of images JPG, GIF, PNG, etc.
interface IImageLoader
{
Bitmap LoadImage(string filename)
}
then create JpegImageLoader, GifImageLoader, PngImageLoader, etc.
Most add-ins and plugin systems work off interfaces.
Another popular use is for the Repository pattern. Say I want to load a list of zip codes from different sources
interface IZipCodeRepository
{
IList<ZipCode> GetZipCodes(string state);
}
then I could create an XMLZipCodeRepository, SQLZipCodeRepository, CSVZipCodeRepository, etc. For my web applications, I often create XML repositories early on so I can get something up and running before the SQL Database is ready. Once the database is ready I write an SQLRepository to replace the XML version. The rest of my code remains unchanged since it runs solely off of interfaces.
Methods can accept interfaces such as:
PrintZipCodes(IZipCodeRepository zipCodeRepository, string state)
{
foreach (ZipCode zipCode in zipCodeRepository.GetZipCodes(state))
{
Console.WriteLine(zipCode.ToString());
}
}

It makes your code a lot more extensible and easier to maintain when you have sets of similar classes. I am a junior programmer, so I am no expert, but I just finished a project that required something similar.
I work on client side software that talks to a server running a medical device. We are developing a new version of this device that has some new components that the customer must configure at times. There are two types of new components, and they are different, but they are also very similar. Basically, I had to create two config forms, two lists classes, two of everything.
I decided that it would be best to create an abstract base class for each control type that would hold almost all of the real logic, and then derived types to take care of the differences between the two components. However, the base classes would not have been able to perform operations on these components if I had to worry about types all of the time (well, they could have, but there would have been an "if" statement or switch in every method).
I defined a simple interface for these components and all of the base classes talk to this interface. Now when I change something, it pretty much 'just works' everywhere and I have no code duplication.

A lot of explanation out there, but to make it even more simpler. Take for instance a List. One can implement a list with as:
An internal array
A linked list
Other implementations
By building to an interface, say a List. You only code as to definition of List or what List means in reality.
You could use any type of implementation internally say an array implementation. But suppose you wish to change the implementation for some reason say a bug or performance. Then you just have to change the declaration List<String> ls = new ArrayList<String>() to List<String> ls = new LinkedList<String>().
Nowhere else in code, will you have to change anything else; Because everything else was built on the definition of List.

If you program in Java, JDBC is a good example. JDBC defines a set of interfaces but says nothing about the implementation. Your applications can be written against this set of interfaces. In theory, you pick some JDBC driver and your application would just work. If you discover there's a faster or "better" or cheaper JDBC driver or for whatever reason, you can again in theory re-configure your property file, and without having to make any change in your application, your application would still work.

I am a late comer to this question, but I want to mention here that the line "Program to an interface, not an implementation" had some good discussion in the GoF (Gang of Four) Design Patterns book.
It stated, on p. 18:
Program to an interface, not an implementation
Don't declare variables to be instances of particular concrete classes. Instead, commit only to an interface defined by an abstract class. You will find this to be a common theme of the design patterns in this book.
and above that, it began with:
There are two benefits to manipulating objects solely in terms of the interface defined by abstract classes:
Clients remain unaware of the specific types of objects they use, as long as the objects adhere to the interface that clients expect.
Clients remain unaware of the classes that implement these objects. Clients only know about the abstract class(es) defining the interface.
So in other words, don't write it your classes so that it has a quack() method for ducks, and then a bark() method for dogs, because they are too specific for a particular implementation of a class (or subclass). Instead, write the method using names that are general enough to be used in the base class, such as giveSound() or move(), so that they can be used for ducks, dogs, or even cars, and then the client of your classes can just say .giveSound() rather than thinking about whether to use quack() or bark() or even determine the type before issuing the correct message to be sent to the object.

Programming to Interfaces is awesome, it promotes loose coupling. As #lassevk mentioned, Inversion of Control is a great use of this.
In addition, look into SOLID principals. here is a video series
It goes through a hard coded (strongly coupled example) then looks at interfaces, finally progressing to a IoC/DI tool (NInject)

To add to the existing posts, sometimes coding to interfaces helps on large projects when developers work on separate components simultaneously. All you need is to define interfaces upfront and write code to them while other developers write code to the interface you are implementing.

It can be advantageous to program to interfaces, even when we are not depending on abstractions.
Programming to interfaces forces us to use a contextually appropriate subset of an object. That helps because it:
prevents us from doing contextually inappropriate things, and
lets us safely change the implementation in the future.
For example, consider a Person class that implements the Friend and the Employee interface.
class Person implements AbstractEmployee, AbstractFriend {
}
In the context of the person's birthday, we program to the Friend interface, to prevent treating the person like an Employee.
function party() {
const friend: Friend = new Person("Kathryn");
friend.HaveFun();
}
In the context of the person's work, we program to the Employee interface, to prevent blurring workplace boundaries.
function workplace() {
const employee: Employee = new Person("Kathryn");
employee.DoWork();
}
Great. We have behaved appropriately in different contexts, and our software is working well.
Far into the future, if our business changes to work with dogs, we can change the software fairly easily. First, we create a Dog class that implements both Friend and Employee. Then, we safely change new Person() to new Dog(). Even if both functions have thousands of lines of code, that simple edit will work because we know the following are true:
Function party uses only the Friend subset of Person.
Function workplace uses only the Employee subset of Person.
Class Dog implements both the Friend and Employee interfaces.
On the other hand, if either party or workplace were to have programmed against Person, there would be a risk of both having Person-specific code. Changing from Person to Dog would require us to comb through the code to extirpate any Person-specific code that Dog does not support.
The moral: programming to interfaces helps our code to behave appropriately and to be ready for change. It also prepares our code to depend on abstractions, which brings even more advantages.

If I'm writing a new class Swimmer to add the functionality swim() and need to use an object of class say Dog, and this Dog class implements interface Animal which declares swim().
At the top of the hierarchy (Animal), it's very abstract while at the bottom (Dog) it's very concrete. The way I think about "programming to interfaces" is that, as I write Swimmer class, I want to write my code against the interface that's as far up that hierarchy which in this case is an Animal object. An interface is free from implementation details and thus makes your code loosely-coupled.
The implementation details can be changed with time, however, it would not affect the remaining code since all you are interacting with is with the interface and not the implementation. You don't care what the implementation is like... all you know is that there will be a class that would implement the interface.

It is also good for Unit Testing, you can inject your own classes (that meet the requirements of the interface) into a class that depends on it

Short story: A postman is asked to go home after home and receive the covers contains (letters, documents, cheques, gift cards, application, love letter) with the address written on it to deliver.
Suppose there is no cover and ask the postman to go home after home and receive all the things and deliver to other people, the postman can get confused.
So better wrap it with cover (in our story it is the interface) then he will do his job fine.
Now the postman's job is to receive and deliver the covers only (he wouldn't bothered what is inside in the cover).
Create a type of interface not actual type, but implement it with actual type.
To create to interface means your components get Fit into the rest of code easily
I give you an example.
you have the AirPlane interface as below.
interface Airplane{
parkPlane();
servicePlane();
}
Suppose you have methods in your Controller class of Planes like
parkPlane(Airplane plane)
and
servicePlane(Airplane plane)
implemented in your program. It will not BREAK your code.
I mean, it need not to change as long as it accepts arguments as AirPlane.
Because it will accept any Airplane despite actual type, flyer, highflyr, fighter, etc.
Also, in a collection:
List<Airplane> plane; // Will take all your planes.
The following example will clear your understanding.
You have a fighter plane that implements it, so
public class Fighter implements Airplane {
public void parkPlane(){
// Specific implementations for fighter plane to park
}
public void servicePlane(){
// Specific implementatoins for fighter plane to service.
}
}
The same thing for HighFlyer and other clasess:
public class HighFlyer implements Airplane {
public void parkPlane(){
// Specific implementations for HighFlyer plane to park
}
public void servicePlane(){
// specific implementatoins for HighFlyer plane to service.
}
}
Now think your controller classes using AirPlane several times,
Suppose your Controller class is ControlPlane like below,
public Class ControlPlane{
AirPlane plane;
// so much method with AirPlane reference are used here...
}
Here magic comes as you may make your new AirPlane type instances as many as you want and you are not changing the code of ControlPlane class.
You can add an instance...
JumboJetPlane // implementing AirPlane interface.
AirBus // implementing AirPlane interface.
You may remove instances of previously created types too.

So, just to get this right, the advantage of a interface is that I can separate the calling of a method from any particular class. Instead creating a instance of the interface, where the implementation is given from whichever class I choose that implements that interface. Thus allowing me to have many classes, which have similar but slightly different functionality and in some cases (the cases related to the intention of the interface) not care which object it is.
For example, I could have a movement interface. A method which makes something 'move' and any object (Person, Car, Cat) that implements the movement interface could be passed in and told to move. Without the method every knowing the type of class it is.

Imagine you have a product called 'Zebra' that can be extended by plugins. It finds the plugins by searching for DLLs in some directory. It loads all those DLLs and uses reflection to find any classes that implement IZebraPlugin, and then calls the methods of that interface to communicate with the plugins.
This makes it completely independent of any specific plugin class - it doesn't care what the classes are. It only cares that they fulfill the interface specification.
Interfaces are a way of defining points of extensibility like this. Code that talks to an interface is more loosely coupled - in fact it is not coupled at all to any other specific code. It can inter-operate with plugins written years later by people who have never met the original developer.
You could instead use a base class with virtual functions - all plugins would be derived from the base class. But this is much more limiting because a class can only have one base class, whereas it can implement any number of interfaces.

C++ explanation.
Think of an interface as your classes public methods.
You then could create a template that 'depends' on these public methods in order to carry out it's own function (it makes function calls defined in the classes public interface). Lets say this template is a container, like a Vector class, and the interface it depends on is a search algorithm.
Any algorithm class that defines the functions/interface Vector makes calls to will satisfy the 'contract' (as someone explained in the original reply). The algorithms don't even need to be of the same base class; the only requirement is that the functions/methods that the Vector depends on (interface) is defined in your algorithm.
The point of all of this is that you could supply any different search algorithm/class just as long as it supplied the interface that Vector depends on (bubble search, sequential search, quick search).
You might also want to design other containers (lists, queues) that would harness the same search algorithm as Vector by having them fulfill the interface/contract that your search algorithms depends on.
This saves time (OOP principle 'code reuse') as you are able to write an algorithm once instead of again and again and again specific to every new object you create without over-complicating the issue with an overgrown inheritance tree.
As for 'missing out' on how things operate; big-time (at least in C++), as this is how most of the Standard TEMPLATE Library's framework operates.
Of course when using inheritance and abstract classes the methodology of programming to an interface changes; but the principle is the same, your public functions/methods are your classes interface.
This is a huge topic and one of the the cornerstone principles of Design Patterns.

In Java these concrete classes all implement the CharSequence interface:
CharBuffer, String, StringBuffer, StringBuilder
These concrete classes do not have a common parent class other than Object, so there is nothing that relates them, other than the fact they each have something to do with arrays of characters, representing such, or manipulating such. For instance, the characters of String cannot be changed once a String object is instantiated, whereas the characters of StringBuffer or StringBuilder can be edited.
Yet each one of these classes is capable of suitably implementing the CharSequence interface methods:
char charAt(int index)
int length()
CharSequence subSequence(int start, int end)
String toString()
In some cases, Java class library classes that used to accept String have been revised to now accept the CharSequence interface. So if you have an instance of StringBuilder, instead of extracting a String object (which means instantiating a new object instance), it can instead just pass the StringBuilder itself as it implements the CharSequence interface.
The Appendable interface that some classes implement has much the same kind of benefit for any situation where characters can be appended to an instance of the underlying concrete class object instance. All of these concrete classes implement the Appendable interface:
BufferedWriter, CharArrayWriter, CharBuffer, FileWriter, FilterWriter, LogStream, OutputStreamWriter, PipedWriter, PrintStream, PrintWriter, StringBuffer, StringBuilder, StringWriter, Writer

Previous answers focus on programming to an abstraction for the sake of extensibility and loose coupling. While these are very important points,
readability is equally important. Readability allows others (and your future self) to understand the code with minimal effort. This is why readability leverages abstractions.
An abstraction is, by definition, simpler than its implementation. An abstraction omits detail in order to convey the essence or purpose of a thing, but nothing more.
Because abstractions are simpler, I can fit a lot more of them in my head at one time, compared to implementations.
As a programmer (in any language) I walk around with a general idea of a List in my head at all times. In particular, a List allows random access, duplicate elements, and maintains order. When I see a declaration like this: List myList = new ArrayList() I think, cool, this is a List that's being used in the (basic) way that I understand; and I don't have to think any more about it.
On the other hand, I do not carry around the specific implementation details of ArrayList in my head. So when I see, ArrayList myList = new ArrayList(). I think, uh-oh, this ArrayList must be used in a way that isn't covered by the List interface. Now I have to track down all the usages of this ArrayList to understand why, because otherwise I won't be able to fully understand this code. It gets even more confusing when I discover that 100% of the usages of this ArrayList do conform to the List interface. Then I'm left wondering... was there some code relying on ArrayList implementation details that got deleted? Was the programmer who instantiated it just incompetent? Is this application locked into that specific implementation in some way at runtime? A way that I don't understand?
I'm now confused and uncertain about this application, and all we're talking about is a simple List. What if this was a complex business object ignoring its interface? Then my knowledge of the business domain is insufficient to understand the purpose of the code.
So even when I need a List strictly within a private method (nothing that would break other applications if it changed, and I could easily find/replace every usage in my IDE) it still benefits readability to program to an abstraction. Because abstractions are simpler than implementation details. You could say that programming to abstractions is one way of adhering to the KISS principle.

An interface is like a contract, where you want your implementation class to implement methods written in the contract (interface). Since Java does not provide multiple inheritance, "programming to interface" is a good way to achieve multiple inheritance.
If you have a class A that is already extending some other class B, but you want that class A to also follow certain guidelines or implement a certain contract, then you can do so by the "programming to interface" strategy.

Q: - ... "Could you use any class that implements an interface?"
A: - Yes.
Q: - ... "When would you need to do that?"
A: - Each time you need a class(es) that implements interface(s).
Note: We couldn't instantiate an interface not implemented by a class - True.
Why?
Because the interface has only method prototypes, not definitions (just functions names, not their logic)
AnIntf anInst = new Aclass();
// we could do this only if Aclass implements AnIntf.
// anInst will have Aclass reference.
Note: Now we could understand what happened if Bclass and Cclass implemented same Dintf.
Dintf bInst = new Bclass();
// now we could call all Dintf functions implemented (defined) in Bclass.
Dintf cInst = new Cclass();
// now we could call all Dintf functions implemented (defined) in Cclass.
What we have: Same interface prototypes (functions names in interface), and call different implementations.
Bibliography:
Prototypes - wikipedia

program to an interface is a term from the GOF book. i would not directly say it has to do with java interface but rather real interfaces. to achieve clean layer separation, you need to create some separation between systems for example: Let's say you had a concrete database you want to use, you would never "program to the database" , instead you would "program to the storage interface". Likewise you would never "program to a Web Service" but rather you would program to a "client interface". this is so you can easily swap things out.
i find these rules help me:
1. we use a java interface when we have multiple types of an object. if i just have single object, i dont see the point. if there are at least two concrete implementations of some idea, then i would use a java interface.
2. if as i stated above, you want to bring decoupling from an external system (storage system) to your own system (local DB) then also use a interface.
notice how there are two ways to consider when to use them.

Coding to an interface is a philosophy, rather than specific language constructs or design patterns - it instructs you what is the correct order of steps to follow in order to create better software systems (e.g. more resilient, more testable, more scalable, more extendible, and other nice traits).
What it actually means is:
===
Before jumping to implementations and coding (the HOW) - think of the WHAT:
What black boxes should make up your system,
What is each box' responsibility,
What are the ways each "client" (that is, one of those other boxes, 3rd party "boxes", or even humans) should communicate with it (the API of each box).
After you figure the above, go ahead and implement those boxes (the HOW).
Thinking first of what a box' is and what its API, leads the developer to distil the box' responsibility, and to mark for himself and future developers the difference between what is its exposed details ("API") and it's hidden details ("implementation details"), which is a very important differentiation to have.
One immediate and easily noticeable gain is the team can then change and improve implementations without affecting the general architecture. It also makes the system MUCH more testable (it goes well with the TDD approach).
===
Beyond the traits I've mentioned above, you also save A LOT OF TIME going this direction.
Micro Services and DDD, when done right, are great examples of "Coding to an interface", however the concept wins in every pattern from monoliths to "serverless", from BE to FE, from OOP to functional, etc....
I strongly recommend this approach for Software Engineering (and I basically believe it makes total sense in other fields as well).

Program to an interface allows to change implementation of contract defined by interface seamlessly. It allows loose coupling between contract and specific implementations.
IInterface classRef = new ObjectWhatever()
You could use any class that implements IInterface? When would you need to do that?
Have a look at this SE question for good example.
Why should the interface for a Java class be preferred?
does using an Interface hit performance?
if so how much?
Yes. It will have slight performance overhead in sub-seconds. But if your application has requirement to change the implementation of interface dynamically, don't worry about performance impact.
how can you avoid it without having to maintain two bits of code?
Don't try to avoid multiple implementations of interface if your application need them. In absence of tight coupling of interface with one specific implementation, you may have to deploy the patch to change one implementation to other implementation.
One good use case: Implementation of Strategy pattern:
Real World Example of the Strategy Pattern

"Program to interface" means don't provide hard code right the way, meaning your code should be extended without breaking the previous functionality. Just extensions, not editing the previous code.

Also I see a lot of good and explanatory answers here, so I want to give my point of view here, including some extra information what I noticed when using this method.
Unit testing
For the last two years, I have written a hobby project and I did not write unit tests for it. After writing about 50K lines I found out it would be really necessary to write unit tests.
I did not use interfaces (or very sparingly) ... and when I made my first unit test, I found out it was complicated. Why?
Because I had to make a lot of class instances, used for input as class variables and/or parameters. So the tests look more like integration tests (having to make a complete 'framework' of classes since all was tied together).
Fear of interfaces
So I decided to use interfaces. My fear was that I had to implement all functionality everywhere (in all used classes) multiple times. In some way this is true, however, by using inheritance it can be reduced a lot.
Combination of interfaces and inheritance
I found out the combination is very good to be used. I give a very simple example.
public interface IPricable
{
int Price { get; }
}
public interface ICar : IPricable
public abstract class Article
{
public int Price { get { return ... } }
}
public class Car : Article, ICar
{
// Price does not need to be defined here
}
This way copying code is not necessary, while still having the benefit of using a car as interface (ICar).

Is this the right understanding of SOLID Object Oriented principles?

After reading about SOLID in a few places, I was having trouble mapping between explanations with different vocabularies and code. To generalize a bit, I created the diagrams below, and I was hoping that people could point out any 'bugs' in my understanding.
Of course, feel free to reuse/remix/redistribute as you'd like!

I think your diagrams look quite nice, but I'm afraid that I couldn't understand them (particularly the interface one), so I'll comment on the text.
It's not really clear to me what you mean by layer, in the Open/closed I thought you might mean interface, but the interface and dependency items suggest you don't mean that.
Open/closed : actually your text from the Liskov item is closer to describing open/closed. If the code is open for extension, we can make use of it (by extending it) to implement new requirements, but by not modifying the existing code (it's closed for modification) we know we wont break any existing code that made use of it.
"Only depend on outer layer" - if this means only depend on an interface not the implementation, then yes, that's an important principle for SOLID code even though it doesn't map directly to any of the 5 letters.
Dependency inversion uses that but goes beyond it. A piece of code can make use of another via its interface and this is has great maintainability benefits over relying on the implementation, but if the calling code still has the responsibility for creating the object (and therefore choosing the class) that implements the interface then it still has a dependency. If we create the concrete object outside the class or method and pass it in as an interface, then the called code no longer depends on the concrete class, just the interface
void SomeFunction()
{
IThing myIthing* = new ConcreteThing();
// code below can use the interface but this function depends on the Concrete class
}
void SomeFunctionDependencyInjectedVersion(IThing myIthing*)
{
// this version should be able to work with any class that implements the IThing interface,
// whether it really can might depend on some of the other SOLID principles
}
Single responsibility : this isn't about classes intersecting, this is about not giving a code more than one responsibility. If you have a function where you can't think of a better name than doSomethingAndSomethingElse this might be a sign its got more than one responsibility and could be better if it was split (the point I'm making is about the "and" in the name even when the "somethings" are better named).
You should try to define that responsibility so that the class can perform it entirely, (although it make may use of other classes that perform sub-responsibilities for it) but at each level of responsibility that a class is defined it should have one clear reason to exist. When it has more than one it can make code harder to understand, and changes to code related to one responsibility can have unwanted side-effects on other responsibilities.
Iterface segregation: Consider a class implementing a collection. The class will implement code to add to the collection or to read from it. We could put all this in one interface, but if we separate it then when we have consuming code that only needs to read and doesn't need to add to the collection then it can use the interface made of the reading methods. This can make the code clearer in that it shows quickly that the code only needs those methods, and, if we've injected the collection by interface we could also use that code with a different source of items that doesn't have the ability to add items
(consider IEnumerable vs ICollection vs IList)
Liskov substitution is all about making sure that objects that inherit from an interface/base class behave in the way that the interface/base class promised to behave. In the strictest interpretation of the original definition they'd need to behave exactly the same, but that's not all that useful. More generally its about behaving in a consistent and expected way, the derived classes may add functionality, but they should be able to do the job of the base objects (they can be substituted for them)

Is it good practice for every public method to be covered by an interface?

It's good practice for a class' implementation to be defined by interfaces. If a class has any public methods that aren't covered by any interfaces then they have the potential to leak their implementation.
E.g. if class Foo has methods bar() and baz() but only bar() is covered by an interface then any use of baz() doesn't use an interface.
It feels like to get cleaner code it would make sense to either:
create extra interfaces if the class has to have those methods (eg a separate interface to cover the behavior of baz() above)
or ideally refactor (eg using more composition) so the class doesn't need to have so many methods (put baz() in another class)
Having methods not covered by an interface feels like a code smell. Or am I being unrealistic?

I consider it as "overusing" the interface.
Interface can give you access only to limited functionality, therefore it is good for gathering more classes with similar functionality into one List<Interface> and using them, for example.
Or if you want to keep loose coupling principle, you rather give another component some interface than the whole class(es).
Also some classes should have restricted access to another classes, which can be done with interfaces too.
However high cohesion principle (which is usually connected to loose coupling) does not prevent you from using class itself, if two classes are and should be "strong" connected to each other.

I don't think that's the purpose of interfaces. If you actually talk about the 'is-a' and 'has-a' relationship between classes, not necessarily a class needs to cover all public methods in interfaces. That's like taking the concept too far.
A class can have methods which describe it's behavior but then, there are some methods that do not exactly describe the classes' behavior but rather describe what else the class can do.
In case if a question arises about SRP regarding the 'can-do' behaviors, it is possible that the class can use a component to execute those behaviors rather than implementing within itself.
For e.g., I have a class DataGrid, why would I need to have an interface called IDataGrid which exposes all the public methods. But may be there is an additional functionality that the DataGrid can do, which is export the data. In that case I can have it implement IExportData, and implement the ExportData method, which in turn does not export the data but uses a component, say DataExportHelper, that actually does the job.
The DataGrid only passes the data to the component.
I don't think SRP will be violated in the above example.
EDIT:
I am a .Net developer, so would like to give you and example from MS library classes. For e.g., the class System.Windows.Window does not implemnt any interface that has Close() method. And I don't see why it should be a part of any presenter.
Also, it is possible that something might look seem like a code smell but not necessarily it might be wrong. Code smell itself does not mean there is a problem but that there is a possibility of problem.
I have never come across any principle or guideline in software design which mentions that all the public members of a class need to be exposed in some or the other interface. May be doing that just for the sake of it might be a bad design.

No, I would definitely not consider methods not covered by an interface a code smell.
It seems like this might be dependent on the object infrastructure you are building in, but in the infrastructures I'm familiar with, the real point of interfaces is to provide a manageable form of multiple inheritance. I consider the overuse of multiple inheritance a notable smell.
In .NET at least, abstract classes are explicitly the preferred construct for exposing abstraction (not interfaces). The .NET design guidelines say: Do favor defining classes over interfaces., with rationale described here http://msdn.microsoft.com/en-us/library/vstudio/ms229013(v=vs.100).aspx.
Even in COM (where any externally visible functionality had to be defined in an interface) there are perfectly good reasons to have non-exposed functions: limiting the visibility of implementation details. COM was originally defined in C (not C++) which lacked the richer set of access modifiers that newer languages have, but the concepts were there: published interface members were public, everything else was internal.

Composition and Inversion of Control

I just came across Inversion of Control approach (implemented using Dependency Injection) of designing loosely coupled software architecture. As per my understanding the IOC approach aims to solve problem related to tight coupling between classes by instantiating an object of a class inside another class which should ideally not happen (as per the pattern). Is my understanding correct here?
If above is true than what about composition or has-a relationship (the very basic important aspect of OO). For an example I write my stack class using a linked list class already defined so I instantiate a linked list class inside my stack class. But as per IOC this will result in tight coupling and hence a bad design. Is this true? I am bit confused here between composition or has-a relationship and IOC.

As per my understanding the IOC approach aims to solve problem related
to tight coupling between classes by instantiating an object of a
class inside another class which should ideally not happen (as per the
pattern). Is my understanding correct here?
Close, but you are slightly off. The problem of tight coupling is addressed when you define contracts between classes (interfaces in Java). Since you need implementations of your contracts(interfaces), at some point those implementations must be provided. IoC is one way of providing an implementation, but not the only way. So tight coupling is really orthogonal to Inversion of Control (meaning it's not directly related).
More specifically, you can have loose coupling but no IoC. The IoC part is that the implementations are coming from outside of the components. Consider the case where you define a class that uses an interface implementation. When you test that class, you might provide a mock. When you pass the mock to the class under test, you are not using IoC. However when you start your app, and the IoC container decides what to pass to your class, that's the IoC.
For an example I write my stack class using a linked list class
already defined so I instantiate a linked list class inside my stack
class. But as per IOC this will result in tight coupling and hence a
bad design. Is this true? I am bit confused here between composition
or has-a relationship and IOC.
Yes and No. In the general sense, you don't need to completely abstract every bit of functionality in your app. You can, and purists probably would, but it can be tedious and over-done.
In this case, you could treat your stack as a black box, and not manage it with IoC. Remember, the Stack itself is loosely couple because the Stack's behavior can be abstracted away. Also, consider the following two definitions
class StackImpl implements Stack {
private List backingList
vs
class StackImpl implements Stack {
private LinkedList backingList
The first is vastly superior to the second, precisely because it's easier to change List implementations; i.e. you have already provided a loose coupling.
That's as far as I would take it. Besides, if you are using composition, you can certainly configure most IoC containers (if not all) to pass things to the constructor or invoke setters, so you can still have a has-A relationship.

Good implementations of IoC can fulfill the "has a" pattern, but just abstract the implementation of the child.
For example, every business layer class may, by your design, "have a" exception handler; with IoC you can define it so that the exception handler that actually gets instantiated at runtime be different in different environments.
The most value in IoC is if you are doing lots of automated testing; in these scenarios you can instantiate mock data access components in your test environment, but have real data access components instantiated in production, which keeps your tests clean. The downside of IoC is that it's harder to debug, since everything is more abstract.

I have my doubts as to my understanding of Inversion of Control too. (It seems like an application of good OO design principles given a fancy name) So, let me assume you are a beginner, analyse your example and clarify my thoughts on the path.
We should start by defining an interface IStack.
interface IStack<T>
{
bool IsEmpty();
T Pop();
void Push(T item);
}
In a way we are already finished; the rest of the code probably will not care whether we implemented it with linked lists, or arrays, or whatever. StackWithLinkedList : IStack and StackWithArray : IStack will behave the same.
class StackWithLinkedList<T> : IStack<T>
{
private LinkedList<T> list;
public StackWithLinkedList<T>()
{
list = new LinkedList<T>();
}
}
So StackWithLinkedList totally owns the list; it does not need any help from outside to construct it, it does not need any flexibility (that line will never change) and the clients of StackWithLinkedList couldn't care less (they have no access to the list). In short, this is not a good example to discuss Inversion of Control: we don't need any.
Let's discuss a similar example, PriorityQueue<T> :
interface IPriorityQueue<T>
{
bool IsEmpty();
T Dequeue();
void Enqueue(T item);
}
Now we have a problem: we need to compare items of type T to provide an implementation of a IPriorityQueue. Clients still do not care whether we use an array, or a heap or whatever inside, but they do care about how we compare items. We could require T to implement IComparable<T> but that would be an unnecessary restriction. What we need is some piece of functionality that will compare T items by our request:
class PriorityQueue<T> : IPriorityQueue<T>
{
private Func<T,T,int> CompareTo;
private LinkedList<T> list;
//bla bla.
}
Such that:
if CompareTo(left,right) < 0 then left < right (in some sense)
if CompareTo(left,right) > 0 then left > right (in some sense)
if CompareTo(left,right) = 0 then left = right (in some sense)
(We would also require CompareTo to be consistent, etc. but that's another topic)
The problem is how to initialize CompareTo.
One option might be, -let's suppose there is a generic comparison creator somewhere- use the comparison creator. (I agree, the example is becoming a little silly)
public PriorityQueue()
{
this.CompareTo = ComparisonCreator<T>.CreateComparison();
this.list = new LinkedList<T>();
}
Or, perhaps even something like: ServiceLocator.Instance.ComparisonCreator<T>.CreateComparison();
This is not an ideal solution for the following reasons:
PriorityQueue is now (very unnecessarily) dependant on ComparisonCreator. If it is on a different assembly, it has to reference it. If someone changes ComparisonCreator he has to make sure PriorityQueue is not affected.
The clients will have a difficult time to use the PriorityQueue. They will first need to make sure that the ComparisonCreator is constructed and initialized.
The clients will have a difficult time to change the default behaviour. Suppose somewhere a client needs a different CompareTo function. There is no easy solution. For example, if it changes the ComparisonCreator<T>'s behaviour, it may affect other clients. What if there are other threads. Even in a single thread environment the client will probably need to undo the change on construction. It's too much effort just to make it work.
For the same reasons, it is difficult to unit test the PriorityQueue. One needs to set up the whole environment.
Of course, - and of course you knew this all along - there is a much easier way in this specific problem. Just provide the CompareTo function in the constructor:
public PriorityQueue(Func<T,T,int> CompareTo)
{
this.CompareTo = CompareTo;
this.list = new LinkedList<T>();
}
Let's check:
PriorityQueue is independent of ComparisonCreator.
For the clients, probably it is much easier to use PriorityQueue. They may need to provide a CompareTo function, but at the worst case they can always ask the ServiceLocator, so al least it is never more difficult.
Changing the default behaviour is very easy. Just give a different CompareTo function. What one client does, does not affect other clients.
It is very easy to unit test PriorityQueue. There is no complex environment to set up. We can easily test it with different CompareTo functions, etc.
What we did is called "constructor injection" because we injected a dependency in the constructor. By giving the needed dependency at the construction, we were able to change the PriorityQueue into a "self sufficient" class. We still create a LinkedList<T>, a concrete class in the construction for the same reasons in Stack example: it is not a real dependency.

The tight coupling in your stack example comes from the stack intantiating a specific list type. The IOC allows the creator of the stack type to provide which exact list implementation to use (e.g. for performance or testing purposes), realizing that the stack does not (at least should not) care what the exact type of the list is as long as it has a specific interface (the methods that stack wants to use) and the concetere implementation provides the required semantics (e.g. iterating through the list will give access to all elements added to the list in the order they were added).

As per my understanding the IOC approach aims to solve problem related
to tight coupling between classes by instantiating an object of a
class inside another class which should ideally not happen (as per the
pattern). Is my understanding correct here?
IoC is actually quite a broad concept, so let's restrict the field to the Dependency Injection approach that you are referring to. Yes, Dependency Injection does what you said.
I think the reason why hvgotcodes thinks that you are slightly off is that the concept of tight coupling can be thought as of having multiple levels. Programming to interfaces is the way to abstract from a particular implementation, which keeps the usage of some piece of code some client code interacts with and its implementation loosely coupled.
The implementation has to be created (instantiated) somewhere though: even if you program to an interface, if the implementation is created inside the client code you are bound to that particular implementation.
So we can abstract the implementation from the interface, but we can also abstract the choice of which implementation to use.
As soon as this detail is clear, you have to ask yourself when it makes sense to abstract the choice of the implementation, which is basically one of the fundamental questions of software engineering: when should you abstract what? The answer to the question is of course context dependent.
But as per IOC this will result in tight coupling and hence a bad
design. Is this true?
If tight coupling is bad design, why are you still relying on standard Java classes? We actually need to distinguish between stable and volatile dependencies.
Citing your example, if you are using the standard implementation of a list, you probably may not want to inject this dependency into your class. What would you achieve by doing this? Do you expect the standard implementation of the list to change any time soon, or do you want to be able to inject a different implementation of a standard list?
On the other hand, suppose you have a custom list with some sort of change tracking mechanism, so that you can perform undo and redo operations on it. Now it could make sense to inject it, because you may want to be able to unit test the client class in isolation, without incurring in potential bugs of your custom list implementation.
As you see, tight coupling is not always bad, sometimes it makes sense, sometimes it is to be avoided: in the end it comes down to the type of dependency.

Why should you prevent a class from being subclassed?

What can be reasons to prevent a class from being inherited? (e.g. using sealed on a c# class)
Right now I can't think of any.

Because writing classes to be substitutably extended is damn hard and requires you to make accurate predictions of how future users will want to extend what you've written.
Sealing your class forces them to use composition, which is much more robust.

How about if you are not sure about the interface yet and don't want any other code depending on the present interface? [That's off the top of my head, but I'd be interested in other reasons as well!]
Edit:
A bit of googling gave the following:
http://codebetter.com/blogs/patricksmacchia/archive/2008/01/05/rambling-on-the-sealed-keyword.aspx
Quoting:
There are three reasons why a sealed class is better than an unsealed class:
Versioning: When a class is originally sealed, it can change to unsealed in the future without breaking compatibility. (…)
Performance: (…) if the JIT compiler sees a call to a virtual method using a sealed types, the JIT compiler can produce more efficient code by calling the method non-virtually.(…)
Security and Predictability: A class must protect its own state and not allow itself to ever become corrupted. When a class is unsealed, a derived class can access and manipulate the base class’s state if any data fields or methods that internally manipulate fields are accessible and not private.(…)

I want to give you this message from "Code Complete":
Inheritance - subclasses - tends to
work against the primary technical
imperative you have as a programmer,
which is to manage complexity.For the sake of controlling complexity, you should maintain a heavy bias against inheritance.

The only legitimate use of inheritance is to define a particular case of a base class like, for example, when inherit from Shape to derive Circle. To check this look at the relation in opposite direction: is a Shape a generalization of Circle? If the answer is yes then it is ok to use inheritance.
So if you have a class for which there can not be any particular cases that specialize its behavior it should be sealed.
Also due to LSP (Liskov Substitution Principle) one can use derived class where base class is expected and this is actually imposes the greatest impact from use of inheritance: code using base class may be given an inherited class and it still has to work as expected. In order to protect external code when there is no obvious need for subclasses you seal the class and its clients can rely that its behavior will not be changed. Otherwise external code needs to be explicitly designed to expect possible changes in behavior in subclasses.
A more concrete example would be Singleton pattern. You need to seal singleton to ensure one can not break the "singletonness".

This may not apply to your code, but a lot of classes within the .NET framework are sealed purposely so that no one tries to create a sub-class.
There are certain situations where the internals are complex and require certain things to be controlled very specifically so the designer decided no one should inherit the class so that no one accidentally breaks functionality by using something in the wrong way.

#jjnguy
Another user may want to re-use your code by sub-classing your class. I don't see a reason to stop this.
If they want to use the functionality of my class they can achieve that with containment, and they will have much less brittle code as a result.
Composition seems to be often overlooked; all too often people want to jump on the inheritance bandwagon. They should not! Substitutability is difficult. Default to composition; you'll thank me in the long run.

I am in agreement with jjnguy... I think the reasons to seal a class are few and far between. Quite the contrary, I have been in the situation more than once where I want to extend a class, but couldn't because it was sealed.
As a perfect example, I was recently creating a small package (Java, not C#, but same principles) to wrap functionality around the memcached tool. I wanted an interface so in tests I could mock away the memcached client API I was using, and also so we could switch clients if the need arose (there are 2 clients listed on the memcached homepage). Additionally, I wanted to have the opportunity to replace the functionality altogether if the need or desire arose (such as if the memcached servers are down for some reason, we could potentially hot swap with a local cache implementation instead).
I exposed a minimal interface to interact with the client API, and it would have been awesome to extend the client API class and then just add an implements clause with my new interface. The methods that I had in the interface that matched the actual interface would then need no further details and so I wouldn't have to explicitly implement them. However, the class was sealed, so I had to instead proxy calls to an internal reference to this class. The result: more work and a lot more code for no real good reason.
That said, I think there are potential times when you might want to make a class sealed... and the best thing I can think of is an API that you will invoke directly, but allow clients to implement. For example, a game where you can program against the game... if your classes were not sealed, then the players who are adding features could potentially exploit the API to their advantage. This is a very narrow case though, and I think any time you have full control over the codebase, there really is little if any reason to make a class sealed.
This is one reason I really like the Ruby programming language... even the core classes are open, not just to extend but to ADD AND CHANGE functionality dynamically, TO THE CLASS ITSELF! It's called monkeypatching and can be a nightmare if abused, but it's damn fun to play with!

From an object-oriented perspective, sealing a class clearly documents the author's intent without the need for comments. When I seal a class I am trying to say that this class was designed to encapsulate some specific piece of knowledge or some specific service. It was not meant to be enhanced or subclassed further.
This goes well with the Template Method design pattern. I have an interface that says "I perform this service." I then have a class that implements that interface. But, what if performing that service relies on context that the base class doesn't know about (and shouldn't know about)? What happens is that the base class provides virtual methods, which are either protected or private, and these virtual methods are the hooks for subclasses to provide the piece of information or action that the base class does not know and cannot know. Meanwhile, the base class can contain code that is common for all the child classes. These subclasses would be sealed because they are meant to accomplish that one and only one concrete implementation of the service.
Can you make the argument that these subclasses should be further subclassed to enhance them? I would say no because if that subclass couldn't get the job done in the first place then it should never have derived from the base class. If you don't like it then you have the original interface, go write your own implementation class.
Sealing these subclasses also discourages deep levels of inheritence, which works well for GUI frameworks but works poorly for business logic layers.

Because you always want to be handed a reference to the class and not to a derived one for various reasons:
i. invariants that you have in some other part of your code
ii. security
etc
Also, because it's a safe bet with regards to backward compatibility - you'll never be able to close that class for inheritance if it's release unsealed.
Or maybe you didn't have enough time to test the interface that the class exposes to be sure that you can allow others to inherit from it.
Or maybe there's no point (that you see now) in having a subclass.
Or you don't want bug reports when people try to subclass and don't manage to get all the nitty-gritty details - cut support costs.

Sometimes your class interface just isn't meant to be inheirited. The public interface just isn't virtual and while someone could override the functionality that's in place it would just be wrong. Yes in general they shouldn't override the public interface, but you can insure that they don't by making the class non-inheritable.
The example I can think of right now are customized contained classes with deep clones in .Net. If you inherit from them you lose the deep clone ability.[I'm kind of fuzzy on this example, it's been a while since I worked with IClonable] If you have a true singelton class, you probably don't want inherited forms of it around, and a data persistence layer is not normally place you want a lot of inheritance.

Not everything that's important in a class is asserted easily in code. There can be semantics and relationships present that are easily broken by inheriting and overriding methods. Overriding one method at a time is an easy way to do this. You design a class/object as a single meaningful entity and then someone comes along and thinks if a method or two were 'better' it would do no harm. That may or may not be true. Maybe you can correctly separate all methods between private and not private or virtual and not virtual but that still may not be enough. Demanding inheritance of all classes also puts a huge additional burden on the original developer to foresee all the ways an inheriting class could screw things up.
I don't know of a perfect solution. I'm sympathetic to preventing inheritance but that's also a problem because it hinders unit testing.

I exposed a minimal interface to interact with the client API, and it would have been awesome to extend the client API class and then just add an implements clause with my new interface. The methods that I had in the interface that matched the actual interface would then need no further details and so I wouldn't have to explicitly implement them. However, the class was sealed, so I had to instead proxy calls to an internal reference to this class. The result: more work and a lot more code for no real good reason.
Well, there is a reason: your code is now somewhat insulated from changes to the memcached interface.

Performance: (…) if the JIT compiler sees a call to a virtual method using a sealed types, the JIT compiler can produce more efficient code by calling the method non-virtually.(…)
That's a great reason indeed. Thus, for performance-critical classes, sealed and friends make sense.
All the other reasons I've seen mentioned so far boil down to "nobody touches my class!". If you're worried someone might misunderstand its internals, you did a poor job documenting it. You can't possibly know that there's nothing useful to add to your class, or that you already know every imaginable use case for it. Even if you're right and the other developer shouldn't have used your class to solve their problem, using a keyword isn't a great way of preventing such a mistake. Documentation is. If they ignore the documentation, their loss.

Most of answers (when abstracted) state that sealed/finalized classes are tool to protect other programmers against potential mistakes. There is a blurry line between meaningful protection and pointless restriction. But as long as programmer is the one who is expected to understand the program, I see no hardly any reasons to restrict him from reusing parts of a class. Most of you talk about classes. But it's all about objects!
In his first post, DrPizza claims that designing inheritable class means anticipating possible extensions. Do I get it right that you think that class should be inheritable only if it's likely to be extended well? Looks as if you were used to design software from the most abstract classes. Allow me a brief explanation of how do I think when designing:
Starting from the very concrete objects, I find characteristics and [thus] functionality that they have in common and I abstract it to superclass of those particular objects. This is a way to reduce code duplicity.
Unless developing some specific product such as a framework, I should care about my code, not others (virtual) code. The fact that others might find it useful to reuse my code is a nice bonus, not my primary goal. If they decide to do so, it's their responsibility to ensure validity of extensions. This applies team-wide. Up-front design is crucial to productivity.
Getting back to my idea: Your objects should primarily serve your purposes, not some possible shoulda/woulda/coulda functionality of their subtypes. Your goal is to solve given problem. Object oriented languages uses fact that many problems (or more likely their subproblems) are similar and therefore existing code can be used to accelerate further development.
Sealing a class forces people who could possibly take advantage of existing code WITHOUT ACTUALLY MODIFYING YOUR PRODUCT to reinvent the wheel. (This is a crucial idea of my thesis: Inheriting a class doesn't modify it! Which seems quite pedestrian and obvious, but it's being commonly ignored).
People are often scared that their "open" classes will be twisted to something that can not substitute its ascendants. So what? Why should you care? No tool can prevent bad programmer from creating bad software!
I'm not trying to denote inheritable classes as the ultimately correct way of designing, consider this more like an explanation of my inclination to inheritable classes. That's the beauty of programming - virtually infinite set of correct solutions, each with its own cons and pros. Your comments and arguments are welcome.
And finally, my answer to the original question: I'd finalize a class to let others know that I consider the class a leaf of the hierarchical class tree and I see absolutely no possibility that it could become a parent node. (And if anyone thinks that it actually could, then either I was wrong or they don't get me).