September 27, 2010, 11:02 AM — Rubinius, an implementation of the Ruby programming language featuring enhancements for JIT (Just In Time) compilation and garbage collection, is being upgraded Friday, gaining improvements in debugging, memory usage, and performance.
[ See InfoWorld's coverage of this week's Java EE vs. Spring debate held at the JavaOne conference. | Keep up with app dev issues and trends with InfoWorld's Fatal Exception blog and Developer World newsletter. ]
"We're really fast under unique workloads" such as numerical performance applications, said Evan Phoenix, lead Rubinius developer at Rubinius at Engine Yard, which also employs lead developers for JRuby, a version of Ruby for the Java Virtual Machine.
"Rubinius is much easier to use in terms of its startup [than JRuby]," Phoenix said in an interview on Friday. "It's much more lightweight than JRuby can be. We've got better memory utilization overall."
Written in C++, the Rubinius bytecode virtual machine incorporates LLVM (Low Level Virtual Machine) bytecode to machine code. The bytecode compiler and most core classes are written in pure Ruby.
Features of 1.1 include a debugger API and reference CLI (Command Line Interface). Although a debugger is provided, developers also can write their own debugger with version 1.1. "When we first did the first release, we didn't have a good debugger story for people to use," Phoenix said. This situation is corrected in Rubinius 1.1.
Developers can integrate their debuggers with IDEs. Aslo, a heap dump memory debugging tool is featured for writing out contents of a program, enabling analysis for bugs.
Performance improvements are featured, such as boosts involving for the Array#pack and String#unpack methods for converting data. "We spent a lot of time on this release improving those," said Phoenix.
Memory usage is improved via instance variable packing. Developers to do not have to declare how classes are going to work for memory purposes.