Hacker News Reader

Comment by Programmatic | Hacker News Reader

I haven't yet replaced Python, but I really enjoyed Dart using JetBrains' WebStorm IDE for a side project.

Python maps really well to a lot of the problems that I'm tackling right now professionally, but I'm considering trying to use Java for a number of them. C# may wind up being viable for me as well; it has a chance of being a little more cross-platform with Xamarin's purchase by MS and overtures toward OSS and Linux by integrating Ubuntu.

View on HN

I used to be a huge Python fan until I found Scala. It can be as expressive as Python (slightly more so even), but as safe as Java (again, slightly more so).

Unfortunately Java is far from safe, NullPointerException and ClassCastException abound. Which isn't surprising given that Java was originally inspired by ObjectiveC, a mostly dynamic language. Scala has a sort of ML subset which mostly avoids these errors, but unless you need the JVM, why not try OCaml or Haskell?

Default-nullable types and thereby having to check for null in every non-primitive argument is definitely a thing in Java, but I don't think I've seen a ClassCastException in the wild for a very long time... Generics pretty much solved the problems that led to bad casts, such as a integer sneaking into your list of strings.

Yes true, Generics helped a lot (thanks to Odersky and Wadler). But the language does seem to encourage more dynamic frameworks like Spring, that IIRC do perform a lot of casting.

There are some Java libraries based on reflection, sure, including Spring. They're not widely used from Scala.

NullPointerException can usually be prevented by using FindBugs annotations. This is a work around for a fundamental design flaw in the Java language, but I find it's pretty effective.

http://findbugs.sourceforge.net/manual/annotations.html

I agree that those are good languages - my reasons for using Scala rather than OCaml or Haskell are mostly boring practical ones (interop with Java, employment, tooling). That said I do think higher-kinded types are very valuable (which rules out OCaml until they finish the modular implicits feature), and that eager evaluation is very useful if there's a chance performance might become important at some point.

Reply on news.ycombinator.com