4 comments

[ 2.8 ms ] story [ 26.1 ms ] thread
I wonder why they decided they really wanted to use the JVM (not saying it doesn't make sense, its just its not mentioned in that presentation).
By leveraging the JVM, they were able to re-use existing compiled Java classes alongside the Scala classes.

Slide 25 hints at this, but I seem to recall Alex saying they used the Java-based Apache HttpClient in their test cases.

I'm learning Scala as a first JVM language, and it is definitely great to be able to access so many Java libraries so seamlessly (Mina, Apache commons, Hadoop, ...), all the Scala syntax/type inference goodness just works and in some cases make the libraries significantly nicer to use in Scala than in Java (IMO).
JVM is pretty fast, and pretty stable, and pretty easy to monitor, compared to other newer language runtimes.

For example, a crashing/security bug like the one that bit Ruby in June is in much less likely for the JVM. And, there are lots of ways to profile the memory usage and code/threading performance of Java apps.