The high level goal of this course is to learn how to transition from coding for courses to coding in the real world.
A lot! Teaching is my hobby; writing code is what I actually do for a living. I am teaching you the things I wish I had known when I graduated.
“Candidates used to struggle to get through the technical portion of interviews…now the technical questions are largely useless, because students do so well now…”
“…we started probing the students to find out what had changed and the answer was CSC-430”
A local employer (paraphrased)
Component | Weight |
---|---|
Assignments | 68% |
Exams | 28% |
Misc | 4% |
For this course, we will be using the following textbooks, in addition to free online resources:
You are expected to read everything that is assigned. If you don’t:
Don’t be afraid to ask questions!
I am available during my office hours and online throughout the day via email and the #csc430 slack channel.
Additional office hours can be planned in advance, so contact me!
Don’t be afraid to ask questions!
Don’t be afraid to ask questions!
Canvas will be used to handle the general organization of the course and all critical announcements.
Other reminders, notes, etc. may be distributed via twitter @msupwright4 and slack.
My goal is to push you hard. Easy classes are not worth the money you are paying and a degree with no actual skills is worthless.
Be responsible. Ask questions. Do your work.
To understand why this class exists, we need to first analyze how you code now, and then we can talk about why this does not scale beyond the classroom.
Most of your coding experience probably consists of projects with less than 100 lines of code.
How do you manage this code?
Some line counts for a few projects I work on:
At these sizes, just working with the code (distributing, sharing, etc.) becomes a non trivial task.
We need a way to backup our code, track changes, and avoid conflicts with coworkers–at scale.
How do you test your course projects?
Manually testing large codebases is, literally, not possible without doing a poor job.
The number of paths in your code to test grows exponentially! If the time you spend testing does not, then you are not testing your code.
Even writing tests for large codebases is not possible if the code is written poorly.
Accordingly, we need to automate testing and write our code in a way that makes it feasible to write sufficient tests.
How do you build your code?
Do you even know how you build your code?
As projects grow in size and complexity, even compiling, building, and deploying your code becomes a problem.
We can not rely on manual steps!
Instead, we must automate the build process (including testing!) to ensure that we can deliver code in a reproducible, safe way.
Ideally, we automate deployment as well.
How hard is it to maintain your code after a year?
You don’t know, because you throw it away after a week!
In the real world, your code will live for years (or decades) and will have to be maintainable by the unlucky individual that gets stuck with your legacy code.
Often, you are that unlucky individual.
Also often, you will not even understand your code if you are not careful with how you write it.
We can largely conquer these problems (and more) by simply caring about our code and automating all the things.
Humans suck at coding and we must humbly accept all of the help we can get from tools, processes, etc.
According to its own website…
Apache Maven is a software project management and comprehension tool […] can manage a project’s build, reporting and documentation from a central piece of information
We will boil that down to the following, though:
Maven is a dependency management and build tool.
What is dependency management?
Code you work on for your courses is often completely self contained, in one or two class files.
You will typically only be importing other classes from the standard library.
In a real world project, though, you will typically be relying on a significant amount of code written by others.
This code will be packaged in jar files which you will need to have available when building and distributing your code.
In the bad old days, this meant:
This may not sound to bad, but on large scale projects, this can be a huge source of problems!
A dependency manager will allow you to provide a small amount of configuration, and it will then handle all of these problems for you in an automated manner.
When we talk about a build tool, we are generically referring to any tool that allows you to provide a configuration (or script), which can then handle all build steps that are necessary to produce your end product.
For instance, you could use a build tool to trigger dependency management, compile your code, execute automated tests, package your compiled code, and more!
A keep theme here is automation.
If our build process is too complicated, we will forget steps and make mistakes.
This will lead to inconsistencies and errors.
Complex manual processes also make it difficult to work with collaborators, because it takes significant work just to get the code running the same on all developer machines.
Instead, we use a clear, precise configuration and feed it to a build tool to guarantee that we have reproducibility anywhere our code is built.
This also allows us to reduce our build process to a single command!
There are usually multiple build tools that can be used for any given programming language, but Maven is one of the most commonly used in the Java world.
You may be interested in becoming familiar with Gradle as well, though.
Maven relies on a configuration called a Project Object Model (POM) file.
Our main concern at this point is how to configure dependencies.
For simple projects, the building works out of the box!
To add a dependency, we simply need to provide the group id, artifact id and version of the library you want to use.
We call these the coordinates of the artifact.
For example, we might add a dependency on a course library like:
<dependencies>
<dependency>
<groupId>edu.murraystate</groupId>
<artifactId>BlobAPI</artifactId>
<version>1.0</artifactId>
</dependency>
</dependencies>
Note that, when you add a dependency, it may also need its own dependencies.
Fortunately, Maven artifacts are packaged with their own POM file, so Maven will go ahead and download all dependencies transitively.
Maven is configured, by default, to pull artifacts from Maven Central, which is a public, centralized artifact repository.
You may, however, need to use custom, private repositories.
<repositories>
<repository>
<id>BlobAPI-mvn-repo</id>
<url>https://raw.github.com/MSUCSIS/csc430-maven/mvn-repo/</url>
<snapshots>
<enabled>true</enabled>
<updatePolicy>always</updatePolicy>
</snapshots>
</repository>
</repositories>
For this course, being comfortable with object oriented programming concepts and having a good understand of why they are beneficial is required.
Let’s start with interfaces!
First, what are interfaces?
We could say that interfaces are “contracts” between the implementers of some code and the users of that code.
In other words, in an interface, you are stating what methods you guarantee will be provided by any class implementing that interface.
Next, how do interfaces differ from classes?
The old school explanation would be that interfaces can not contain implementations of methods. They may only contain method signatures.
(This isn’t actually true anymore, though)
public interface Transform {
String apply(final String input);
}
This interface represents the abstract concept of code which performs String transformations.
Note that this code does not do anything! It just states that any class implementing this interface will provide a method called apply that will perform a String transformation.
Now how about classes?
We mentioned above that classes contain actual implementation details. In other words, they contain code that actually does stuff.
public class Reverse {
public String transform(final String input){
final StringBuilder sb = new StringBuilder();
for(int i=input.length-1; i>=0; i--){
sb.append(input.charAt(i));
}
return sb.toString();
}
}
Is this a class?
Does it perform a String transformation?
Does it implement the Transform interface?
public class Reverse implements Transform {
public String apply(final String input){
// ...
}
}
Now we have stated that we are agreeing to the “contract” defined by Transform
final Reverse transformer = new Reverse();
System.out.println(transformer.apply("Hello");
final Transform transformer = new Reverse();
System.out.println(transformer.apply("Hello"));
What’s the difference here? Why would we do this?
At this point, there is no real reason to use interfaces at all. What happens, though, when we decide to add another type of transformation to our code.
if(doReverse){
final Reverse reverse = new Reverse();
return reverse.apply("Hello");
}else if(doNoVowels){
final NoVowels noVowels = new NoVowels();
return noVowels.apply("Hello");
}else if ...
While all of these classes do different things, they all, at a high level, are transforming Strings.
So, if we use the Transform interface for all of them, then we can unify them with one variable.
final Transform transform;
if(doReverse){
transform = new Reverse();
}else if(doNoVowels){
transform = new NoVowels();
}else if ...
return transform.apply("Hello");
This may not seem to impressive, but what if the logic choosing the transformation is not in our code?
What if we have to pass a transform into someone elses code?
What if we need a collection of transformations?
final SomeoneElses code = new SomeoneElsesCode();
// We have no idea what this could possibly return!
final Transform transform = code.getTransform();
// ...but we can use it anyway, because we know its interface!
return transform.apply("Hello");
In someone else’s code:
public void setTransform(final Transform t){
transform = t;
}
In our code:
//They have no idea what this is
final Transformer tx = new Reverse();
//...but they can still use it!
code.setTransform(tx);
final List<Transform> txs = new ArrayList<>();
txs.add(new Reverse());
txs.add(new NoVowels());
//...
What are all of the angle brackets?
final List<Transform> txs = new ArrayList<>();
I guess we’re going to need to talk about generics.
We mentioned how an interface can be used to represent different classes by referring to them as instances of the interface instead.
We can also do this with an ancestor classes, if all of the classes have a common ancestor. So, in the bad old days, a List would simply store Objects, because Object is a common ancestor for all other classes.
This led to problems:
final List txs = new ArrayList();
txs.add(new Reverse());
// Not an error, because it's an object
txs.add("Hello");
// Have to cast, because we need a Transform, not an Object
final Transform tx = (Transform)txs.get(1);
To fix this, generics were introduced to allow for more type safe code to be written:
final List<Transform> txs = new ArrayList<>();
txs.add(new Reverse());
//Now this is an error
txs.add("Hello");
//And a cast wouldn't be needed if the above didn't crash!
final Transform tx = txs.get(0);
Let’s improve our Transform interface now:
public interface Transform<T,U> {
U apply(final T input);
}
Now we’re not limited to String transformations!
public class Reverse implements Transform<String,String> {
public String apply(final String input){
//...
}
}
final List<Transform<String,String>> stringTransformers;
final List<Transform<Integer,Integer>> intTransformers;
final List<Transform<String,Integer>> stringToIntTransformers;
Can you feel the power??? Programming is so awesome.
Ok, let’s dial things back a bit and get back to some basics.
What is inheritance? How does it work?
Previously, we implemented an interface.
Sometimes, though, we want to extend a class.
When a class B extends a class A, then the class B will inherit the methods and fields of class A.
public class ReReverse extends Reverse {
}
What will the following do? Will it compile?
final Reverse r1 = new ReReverse();
System.out.println(r1.apply("hello");
public class ReReverse extends Reverse {
public String apply(final String input){
final String reversed = super.apply(input);
final String reReversed = super.apply(reveresed);
return reversed;
}
}
Super?
What will the following do?
final Reverse r1 = new ReReverse();
System.out.println(r1.apply("hello");
final ReReverse r2 = new Reverse();
System.out.println(r2.apply("hello");
final Transform t = new ReReverse();
System.out.println("hello");
When we declare a variable’s type, that tells us what fields and methods must exist, but it might not be the precise type of the value bound to that variable.
(The actual value may have more fields and methods!)
When we call a method declared in the parent class, the JVM will determine, at run time, what the actual type of the value is, and will use this information to select the correct implementation of the called method.
What is the correct implementation?
If the child class has not overridden the method, then the parent class will be checked for an implementation.
If the child class has overriden the method, then the implementation in the child class is chosen.
Remember, though, that you can have several levels of inheritance!
Don’t forget that access modifiers affect what fields and methods are visible to children!
If, for example, a parent class declares a private field, the child can not access it!
none | private | protected | public | |
---|---|---|---|---|
class | Y | Y | Y | Y |
package | Y | N | Y | Y |
subclass | N | N | Y | Y |
world | N | N | N | Y |
A common mistake students (and professionals!) make is to assume that the access modifiers apply to instances of a class. This is not true!
For instance, a private field in class A is visible ot every instance of A!
For example, this code is perfectly valid!
public class A {
private final int x;
//...
public void compareXs(final A anotherA){
// works even though x is private!
return x==anotherA.x;
}
}
Let’s circle back around to some of the basics:
An interface (sort of) only provides method signatures.
A class must be completely implemented.
An abstract class is when you want something in between.
public abstract class Censor
implements Transform<String,String> {
public abstract List<String> getBadWords();
public String apply(String input){
for(final String word : badWords){
input = input.replace(word, "!!!")
}
return input;
}
}
Now we can create different variations by subclassing and implementing getBadWords.
What if we want to inherit from two classes?
You can’t!
You can implement several interfaces, but you can only extend one class (concrete or abstract).
However…
You can now, actually, provide “default” implementations of methods in interfaces.
Which means that you can often accomplish what you were trying to do with two parent classes by instead implementing two interfaces with default methods.
There are, however, limitations. First, remember that you can only store static, constant fields in an interface. If you need instance data for your default methods, you are out of luck.
You can, though, add a getter to the interface for the instance data, and use that in your default method.
Then implementers will declare their own instance fields and implements the getter.
public interface Censor extends Transform<String,String> {
List<String> getBadWords();
default String apply(String input){
for(final String bad : getBadWords()){
input = input.replace(bad, "!!!");
}
return input;
}
}
Another common mistake is to try and implement an interface in another interface. This makes no sense, because you aren’t (generally) providing an implementation!
So, interfaces extend other interfaces!
An interesting use for multiple interfaces is to create an interface for each kind of behavior that might be needed, and mixing them together.
public interface Labeled {
public String getLabel();
}
public interface Logged {
public List<String> getLog();
}
public class Reverse
implements Transform<String,String>, Labeled, Logged {
private final List<String> log = new ArrayList<>();
public String getLabel(){return "Reverse (Transform)";}
public List<String> getLog(){return log;}
public String apply(final input){
log.add("Starting transform of input: " + input);
//...
log.add("Transform complete:" + result);
return result;
}
}
So, in addition to being able to use this class wherever we need a Transform, we can use it anywhere we expect, for example, instances with a label.
For example, imagine a UI where we can select operations to perform. Some might be transformers, other might not be, but we can put them all in a common UI widget with a nicely displayed label if they all implement the Labeled interface!
At runtime, though, we may not know what interfaces are implemented, so we will need to check using the instanceof operator.
public void doTransform(final Transform<String,String> t){
final String input = getInput();
final String output = t.apply(input);
if(t instanceof Logged){
final Logged logged = (Logged) t;
printLog(logged.getLog());
}
writeOutput(output);
}