Ask HN: Java vs. Scala for JVM based computational intelligence library

1 point

10 years ago

I am creating a JVM based library for computational intelligence algorithms, similar to https://github.com/deap/deap (python). One of my primary goals with the library is commercial usability and therefore performance, as such I will be relying on Skymind's ND4J/S linear algebra library quite heavily (if you have any experience with alternatives, I am keen on hearing about them).

My question is, should I create the library in Java (8) or Scala?

- Using Java allows compatibility with most other JVM languages and frameworks (including Scala) if limited to Java 7. - I prefer Scala over Java, as Scala is a more powerful language that will allow me to do more (especially functionally), but primarily I am worried that it will exclude users and/or contributors. - I plan on integrating with Skymind's DL4J neural networks, though I think both languages will allow that. - Apache Spark integration is also a goal. - I am somewhat worried about the longevity of Scala. - I am proficient in both languages, in both OO and functional styles.

Any feedback is much appreciated!

1 comment