Scala daddy wraps his Java baby in Red Hat-ness
Is it Typesafe? Yes it is
Martin Odersky – the man who created Scala, the Java-based programming language that now drives such big name web services as Twitter, Foursquare, and LinkedIn – has launched a company that provides service and support around an extensive open source application stack for the language.
Both the company and the stack are known as Typesafe. Officially unveiled on Thursday, the Typesafe stack includes the most recent releases of the Scala programming language, the open source Akka middleware, and various open source tools designed to facilitate the development of Scala applications. "We want to provide stable versions of the stack, stable supported versions where you can get backports of fixes and improvements", Odersky tells The Register. "It's very much the open source support model you see with Red Hat".
The company will also offer consulting services, but this is not a primary aim.
Odersky, a professor at the École Polytechnique Fédérale de Lausanne in Switzerland, is known not only for creating Scala, but for building javac, the Java compiler included with the Java Developer Kit. First released in 2003, four years after Odersky built javac, Scala runs atop the Java Virtual Machine, and it's meant to be an improved version of the Java language.
"[My javac] work was a worthy undertaking, but it led me to realize how difficult programming was," Odersky says. "I decided to rethink the way we program, but at the same time, I wanted to stay really practical. So I wanted to build something that was completely interoperable with Java, and that's what Scala became."
First and foremost, Scala was built to provide concurrency. "The idea was to address the next big challenge after internet programming, which Java addressed pretty well," he says. "At the time, you could already see that the next big challenge would be concurrency and parallelism." The best way to address concurrency, he says, was through functional programming. For Odersky, mutable variables cause huge problems when you're dealing with concurrent programs.
"Any sort of mutation is really something very dangerous. Once you have mutable states, it's a big headache to cache it, to replicate it – all the sort of things that are your daily bread if you want to have large, horizontally scalable systems. State is really very hard to manage for all those systems."
But Scala would retain Java's object oriented techniques. It's unique in that it blends the object oriented and the functional. For concurrency, it takes a page from Erlang, passing messages between computational entities known as "actors". These actors are the cornerstone of the Akka middleware included with the Typesafe stack.
Because the language runs atop the existing Java Virtual Machine, it provides performance akin to Java and C and, for that matter, Go – the Google designed language that also takes direct aim at concurrency.
The inaugural Typesafe stack includes Scala version 2.9, which offers tools specifically designed for building parallel applications – i.e. concurrent applications that run across multiple processor cores. Version 2.9 marks the debut of a Parallel Collections library that lets you do all transformation and bulk operations in parallel.
The new Typesafe company combines two existing operations: Odesky's own Scala Solutions, and Scalable Solutions, a company run by Jonas Bonér, the Stockholm-based creator of Akka. The third founder is the San Francisco-based Paul Phillips, the largest contributor to the Scala codebase over the last few years.
Typesafe is backed by $3 million in Series A funding led by the Silicon Valley-based venture capital firm Greylock Partners. The company's board of advisors includes Java creator James Gosling and Doug Lea, a professor of computer science at State University of New York at Oswego, who specializes in concurrency.
In addition to the Scala runtime, the core Scala libraries, and version 1.1 of the Akka middleware, the open source Typesafe offers a new Scala Eclipse plugin and a simple build tool for Scala known as SBT. The stack runs on Mac OS X, Linux, and Windows, and it requires the Java runtime version 1.6 or later. ®
Languages and Languages
You mixed a lot of different things up there.
F# is .Net's version of OCaml, an old and traditional functional programming language.
Go targets mainly system programming. It is an alternative mostly to C.
Scala and Ceylon are JVM languages, though Scala released its first version 8 years ago, and it is now on version 2.9.0, with a proven record of success while Ceylon has yet to release its first version.
Ruby is another language that is as old as Java, and has revolutionized web programming with Ruby on Rails... five years ago. There's a wide array of tools written in Ruby, and used in Linux and BSD servers -- FreeBSD's package tools have been written in Ruby for over a decade, for instance.
In fact, Python is not particularly older or more popular than Ruby -- it just attracts different crowds. So to call Ruby a "new" language and Python a "tried-and-tested" is non-sense.
So, my advise is that you actually learn a bit more about what's actually out there, before you even consider learning a new language.
As for learning a new language, it depends a lot on what you wish to accomplish and what your usual expertize lays.
If you are a .Net developer, you should definitely learn F#. Chances are high that interfacing with F# code will become common.
As a Java developer, the best bet is Scala and Groovy. Groovy is close enough to Java that it will hardly feel like learning a new language anyway, but while it is useful in some places, it can never replace Java.
If you are a Python developer, Ruby fits a similar niche, but is unlikely to be attractive if you think the Python way is the Right way.
And if you just want to stretch your horizons, learn Haskell.
This is great news
I converted to Scala a year ago and haven't looked back (after being a Java dev for 10 years). The HUGE benefit is that you don't need to convert all at once since you can call all your Java code from scala (and vice versa).
More Typesafety...no GC required
The IT world had a serious regression since the days of ALGOL. C, Unix, Windows and Java confused the minds of many people in the "computer trade" (for a lack of better words).
I blame it on "but I can make more money with Java|C#|C++" Way Of Thinking.
A Proper Type System (which also covers the aspect of multithreading) combined with Stack Allocation, Destructors, Value Arrays, References and so on can get very close to C/C++ runtime/memory efficiency without all the security risks of C/C++. JVM-based languages certainly will never reach the C/C++ efficiency.