my project has a significant and isolated part that was written in ruby(jruby compatible). It is a commandline application that we run it in the terminal and provide it with various option flags.
My client wants to use this tool but only willing to use it if it is wrapped in a java class. I went through a lot of trouble to convert the ruby code to java by using jrubyc --javac A.rb. Inspecting the converted .java file, it is calling a Ruby Runtime to execute the ruby script. Like this:
org.jruby.Ruby.getGlobalRuntime().executeScript(stringBuiltFromARubyFile, 'path')
My question is performance wise, is this better than just wrap the create a runnable jar, wrap it with a java class that takes certain parameters, and execute the jar via Runtime.getRuntime().exec("java -jar A.jar args") ?
The application A.rb uses multi-threading. I bring in ruby dependencies (gems) using jruby-gradle plugin.
What are some other options I can explore?
Thanks in advance.
The approach you are using likely won't work. You are launching a JVM within a ruby context; and what you've been asked for is launching some of your ruby code within a JVM context.
I would look at C to Ruby bindings, and then use a Java to C (JNI) interface to launch the required Ruby code from the C layer. If such a thing is not practical, as your Ruby is more of a standing service than a CLI process, I would then consider making a set of Java libraries to call the process through a network call.