Why do we need "? extends" in Java | Hacker News Reader

Why do we need "? extends" in Java

stackoverflow.com

32 pointsrocky_ex25 comments

There is a good free tutorial on generics: https://java.sun.com/j2se/1.5/pdf/generics-tutorial.pdf. It explains the design of generics, and tells a bit about why Sun engineers came up with this implementation.

When we hire developers with proficiency in Java that's one of our common interview questions. We also seek ability to figure out what "? super" means.

I have a feeling you're going to get a bunch of type theorists instead of programmers who get things done if you ask about Java's syntax for use-site contravariance.

Not true. You'll get, for instance, people who read Effective Java [1], by Joshua Bloch (who is, IMHO, THE man when it comes to java), and remembered the PECS mnemonics. PECS stands for producer-extends, consumer-super.

So, to cite the textbook example, say you are implementing a Stack<E>. It will probably have the methods:

  public void push(E element);
  public E pop();

and, for convenience:

  public void pushAll(Iterable<? extends E> elements);
  public void popAll(Collection<? super E> destination);

The pushAll has "? extends" because the elements Iterable will "produce" elements for the stack. The popAll has super because the destination Collection will "consume" elements from the stack. It is not that hard, is it? Let's note that guard-of-terra is talking about proficiency, not mere familiarity. I believe reading "Effective Java" is a nice way to get closer to the proficient level.

Let it be noted that this whole mess exists because generics in Java were implemented with type erasure so their introduction wouldn't break legacy code. I personally think this was a bad idea, but it does show that when a language is evolving, there are a bunch of constraints the designers must be aware of.

[1] http://www.amazon.com/Effective-Java-Edition-Joshua-Bloch/dp...

This doesn't seem to have anything to do with type erasure, does it? It's needed for type-safe generics. Even without type erasure, you still need type safety. If your generics are always covariant (like Dart's are), you break type safety.

I think the mistake the Java designers made here was going with use-site variance instead of definition-site variance.

Yes, you're right. We're talking about variance when it comes to the "? extends, ? super" situation, not erasure. And you're also right the things would be better with definition-site variance.

On the other hand, sometimes I think an unsound type system with List<String> being a subtype of List<Object> would be better than what we have today, for pragmatic reasons. Of course, I think this makes me a non-type-theorist, as what I'm saying is considered heresy in some circles [1].

[1] http://lambda-the-ultimate.org/node/4377 (search for unsound).

But than you can never get anything from List<String>! Because you never know what type are you getting!

On the other hand, if your structure is immutable it would probably work.

"? extends" is a cool question because even if you know nothing theoretical about generics, you can figure it out with a few hints because it does have an actual underlying reason. Or you can't - even with massive hinting some people are just afraid to guess and reason. So it's a nice test.

We had to learn it when doing an oo programming course at university. And I immediately forgot all about it.

I hope when writing Java code sometimes at a job, I can refer to this again.

This is all done in the name of type safe programming (I don't know the exact english terminology).

<? extends B> is called covariance. It helps to access attributes of B and derived classes with get() but prevents to use set().

ArrayList<? extends Number> list;

list = new ArrayList<Integer>();

list = new ArrayList<Double>();

list = new ArrayList<Long>();

list.set(index, myInteger); // throws error

Compiler has to know the exact type to set the value, but Long, Double and Integer are all from type Number, so you can read it into a Number variable.

Is that correct?

You describe how it works, but not what it means in simple terms.

List<Shape> is not a "list of Shapes", it's a "list parametrized with Shape"

List<? extends Shape> is a "list parametrized with something that extends Shape"

e.g. you can assign it with List<Circle> (you can't assign it to List<Shape> since Shape != Circle) and then you can do operations on it that would not break List<Circle> (can't put Square in it)

Great, thank you. It's interesting to learn, when you don't know how to describe something in simple terms. I blame my lacking college professor and me, not taking enough time to learn it properly.

Why do we need Java?

ava's design allows app development and performance to be decoupled (to a good extent, when compared to C or C++). Combined with the lower learning curve (again, compared to that which ruled before - aka C++), this commoditizes app development and has hence found much adoption in industry, leaving a small team of specialists to work on performance (& the jvm). Adoption leads to funds leads to research and/or improvements until Java expanded to include much of what needed to be done via programming, in other words, you can get almost everything you want done in java. In other words, Java is an industry standard, or if you will, a household name. So why do we need Java ? same reason as we need division of labour.

Currently, the JVM usually outperforms everything other than C or C++, and those are only fast until they blow up randomly because humans can't write 100% flawless code.

I don't know about that. Some of the Java libraries that were written in C were long ago converted to Java and they run faster.

It's hard to say that the JVM is faster or slower than X because it usually comes down to the way in which the application is developed.

If C++ was ALWAYS faster then I wouldn't be using Java for our trading systems. Some of our competitors use C++ and some use Java. We have one of the fastest trading platforms in the world, the only firm to have faster than us are also using Java.

Cool. I've always expected JIT could eventually beat AOT compilation, though I haven't heard of that being commonly seen. I'm a little surprised that C++ still tends to use crude vtable dispatch just because that's what Stroustrup demonstrated and it's easy.

Many companies depend on Java for their mission critical infrastructure and applications.

I used to work at a world leading investment bank who's entire low latency, high scalable trading platforms were built in Java.

Just about every large financial firm that I've worked at plus both telco firms and one management consultancy used Java to build large, fast and scalable applications.

You may have no use for Java in your small bubble of a world where you most probably work on insignificant, crappy pay and waste of time projects, but people who work on important stuff in the real world do. I'm sure other people can tell you why Java is important in ways that I cannot. I just get annoyed when I see people slag off a piece of technology just because they don't use it or like it.

Because it powers tens of thousands of huge systems in use?

How about that?

Have you tried programming BEFORE Java, when the same kind of systems were done in C++ with CORBA or similar monstrosities?

All about Java Generics in one big FAQ

Wow, I'm so glad that I don't need to worry about such a mess when using duck-typing in Python.

I'm glad that when I refactor code in Java, the type checker catches many of my mistakes. Not as many as in Haskell, but still.

In Python, the deploy/run unit tests/spot typo cycle greatly slows me down.

Yeah this... I've found out that I subconsciously actively avoid refactoring our (old and in parts terrible) codebase because the tools for it are so bad.

Yeah, it's soooooo complicated!

I thought the whole point about Java is that it's supposed to be extra verbose (which is also the reason it sucks a bit). wtf is this shit

Let me tell you why everyone is downvoting you:

The "? extends" is not an "extra verbosity". It's actually a sound CS concept, one that you need to make compiler understand how some generic code turns out to be correct while working on partially specified types.

IIRC C# didn't have that "extra verbose" thing in its early versions and the developers weren't very happy because you do need it.

Not being able to separate "extra verbosity" from "solid CS concepts" makes you lousy at reasoning about programming languages. Might as well stop bothering.

Reply on news.ycombinator.com