Sunday, April 25, 2010

Technical Research Topics

As you can probably guess from reading this blog, I'm the sort of person that's constantly researching new technologies, often to the point where I move on from a given topic before I've given it the time and attention it deserves.  The point of this post is to both mention what I'm doing to make my attention less transient and to discuss some of the technologies I've decided to focus on.

How am I trying to keep my research interests under control?  Composing a list.  I created a Google spreadsheet and started entering my professional interests.  That probably sounds rather unexciting, but for the first time in a while, I had a concrete list I could look over and evaluate.  The first thing I noticed was that it was long, far too long.  There was no practical way to work on all of the items in the short term.  The natural next step was to sort the list into a rough priority order.  It was hard to push items down ("Ooh, I bet that would be fun!"), but there are only so many things a person can do.

Here's a subset of the list, not necessarily in priority order (I'm still trying to figure out how much I care about the technologies being relevant to my day-to-day work):

  • Git is the distributed/decentralized source control system that Linus Torvalds put together in 2005 to manage the source for the Linux kernel.  It took me a while to grok the notion of distributed source control (I'm not alone), but once I did, I became convinced that it could lead to great productivity gains.  Rather than trying to convince you of this directly, I refer you to the site Why Git is Better Than X, the Getting Started with Git DZone RefCard and Git 101 screencast by Matthew McCullough (registration required), and GitHub.  If you need any extra encouragement, you should know that Git is being used by Grails (and SpringSource generally), Eclipse, and many others that I'm forgetting at the moment.  Learning Git is definitely my top research priority.
  • Static Groovy (most call it Groovy++, but I can't stand that name!) is a project based on the not-yet-released 1.8 version of Groovy that provides annotations allowing developers to designate sections of code (methods or classes) as statically typed.  This brings large performance improvements for many operations, allows some errors to be caught at compile-time rather than run-time, and allows type inferencing similar to what's done in Scala.  I dug into this a bit and ended up changing my mind.  Along with making Groovy run faster, it makes significant semantic changes.  It's not just an addition to the language, it's a whole separate dialect.
  • GPars (formerly GParallelizer) is a project that aims to make it easier to write parallel programs in Groovy.  I originally learned about its smooth support for Fork/Join, but it also enables several other concurrent programming models such as actors, agents, and dataflow concurrency.
  • Gradle is a new build system/language written in Groovy that learns from and improves upon Maven.  Like Maven and Grails, it emphasizes convention over configuration, but unlike Maven, it doesn't become painfully awkward to deal with when you diverge from its conventions.  It uses Apache Ivy to implement its dependency management features.  Speaking of which…
  • Ivy is a sub-project of the long-lived Ant build system and separates management of dependencies from the rest of the build process.  It can be used with Ant or other build systems (see previous).
  • CouchDB is a relatively mature open source document-oriented database that is frequently grouped with other "NoSQL" databases.  It's written in Erlang and stores JSON documents.  It allows you to build views based on Map-Reduce functions written in JavaScript.  I recently signed up for a beta account with Cloudant, a company that provides a hosted version of CouchDB.  As a related alternative, Riak is also definitely worth considering.