Hacker News Reader

Comment by the_af | Hacker News Reader

And yet people here on HN have told me in recent threads, when discussing Java, that:

- Java is not suitable for enterprise software (yes! Some people here actually believe that).

- Java is not a good general purpose language (but Minecraft was written in Java. Or even better, the list of very varied apps from this article shows it's indeed a good general purpose language).

What these people actually mean is "I don't like Java". Because Java hasn't been "hip" for years, this makes people claim all sorts of absurd things about it. Things that directly contradict known history.

View on HN

I work in Java every day, and I agree it's been a huge success. But I also don't think it's a good general purpose language. It's absolutely full of warts and gotchas and ergonomic problems and questionable design decisions that even the people who made them have come to regret.

But known history tells us that there is very little correlation between a language's success and how well-designed it is. Outside of software, too - I think the archetype for this phenomenon is probably VHS and Betamax.

> It's absolutely full of warts and gotchas and ergonomic problems and questionable design decisions that even the people who made them have come to regret.

And which general purpose language isn't?

> questionable design decisions that even the people who made them have come to regret.

Such as?

Serialization is one of them http://cr.openjdk.java.net/~briangoetz/amber/serialization.h...

Wild mixes of consumer and provider API's for example in Swing. The later designs have been much better in that regard.

The built-in http client library is borderline horrible.

I think the most interesting one to me was an interview with Joshua Bloch in the book Coders at Work. It was to do with generics, of course, but distinct from the usual surface-level debate over type erasure. He felt that generics were rushed, and that they could have done a better job of sorting out the details of how things were going to work at the language level if they had given themselves more time.

This is a good read[1], and they're "fixing" them with specialization.

[1] https://cr.openjdk.java.net/~briangoetz/valhalla/erasure.htm...

As someone who's not a big fan of Java, I don't get those arguments.

If anything my argument for Java would be that it tries to be good at -everything-, and because that attempt started from day one (versus, say, C# starting primarily on desktop, then Web, and worrying about smaller targets and big servers later).

And I think it's pretty good at all of that from a 'capability' standpoint.

But because it tries to be all things for all people, it's pretty easy to wind up down rabbit holes of large frameworks. And that leads to the warts Java has from a usability standpoint.

I would say its suitable, but I wouldn't say it's a good general purpose language for a variety of reasons. The first being that "good" is totally subjective. It feels like it has lagged behind its competitors both in terms of language features and API design. Its generally crappy for GUI apps. It doesn't really excel at anything in particular unless you're already invested in it. Even it's original premise of "write once run anywhere" is mostly irrelevant now, and there doesnt seem to be a future for client side Java apps outside of Android.

Overall I'd prefer to use just about anything else in pretty much any domain where Java would be a consideration. It may not be an inherent problem with the language itself, but its ecosystem in my experience leaves it as one of the worst general purpose languages in terms of productivity while not offering a sufficient advantage to make up for that.

All that being said, Java is still good enough most of the time.

My experience with the ecosystem is the opposite. The ecosystem is so good, and so mature, it's hard to argue for other technologies at my current job.

For any fancy new language/framework, you must show you understand its performance issues, pitfalls, metrics and monitoring tools, etc. Few languages beat Java at this for serious work.

The performance of the Java version of Minecraft is so terrible that the entire game had to be rewritten to be played on consoles.

Minecraft was not popular because of Java, Minecraft was popular despite Java.

I strongly disagree. A big part of minecraft's popularity is due to it's mods scene, and it's mod scene is so successful because modders could decompile minecraft's classes and build a deep understanding of how minecraft works under the hood.

That's highly debatable, not a point of fact.

And the rest of the apps from the article? Are they also successful "despite Java"? At which point are you willing to admit that maybe Java was suitable for general purpose programming? Which amount of evidence would convince you?

What's debatable, that the Java version of Minecraft had terrible performance or that the console ports are not written in Java?

About your other question: I only mentioned Minecraft.

The point that's debatable is the point you led with, that Minecraft "had" to be re-written because of poor performance.

Not that it was.

The game had to be rewritten because Microsoft is a C++ shop and game consoles SDKs don't support Java, and it was definitely easier to port the game to C++ than to create Java AOT compilers for game consoles.

Reply on news.ycombinator.com