This tool allows you to connect a Scala REPL console to running Oracle (Sun) JVM processes without any prior setup at the target process.
Download and extract scalive-1.5.zip, you will see:
scalive-1.5/
scalive
scalive.cmd
scalive-1.5.jar
scala-library-2.10.4.jar
scala-compiler-2.10.4.jar
scala-reflect-2.10.4.jar
scala-library-2.11.4.jar
scala-compiler-2.11.4.jar
scala-reflect-2.11.4.jar
scala-library, scala-compiler, and scala-reflect of the appropriate version will be loaded to your running JVM process, if they have not been loaded. The REPL console needs these libraries.
For example, your process has already loaded scala-library 2.10.4 by itself, but scala-compiler and scala-reflect haven't been loaded, Scalive will automatically load their version 2.10.4.
If none of them has been loaded, Scalive will load version 2.11.4.
For convenience, Scala 2.10.4 and 2.11.4 JARs are preincluded. If your process is using a different Scala version, you need to manually download the corresponding JARs from the Internet and save them in the same directory as above.
Run the shell script scalive (*nix) or scalive.cmd (Windows).
To see a list of running JVM processes and their process IDs:
scalive
To connect a Scala REPL console to a process:
scalive <pid>
Scalive only automatically adds scala-library.jar, scala-compiler.jar,
scala-reflect.jar, and scalive.jar to the system classpath. If you want to
load additional classes in other JARs, first run these in the REPL console to
add the JAR to the system class loader:
val cl = ClassLoader.getSystemClassLoader.asInstanceOf[java.net.URLClassLoader]
val searchDirs = Array("/dir/containing/the/jar")
val jarbase = "mylib" // Will match "mylibxxx.jar", convenient when there's version number in the file name
scalive.Classpath.findAndAddJar(cl, searchDirs, jarbase)
Now the trick is just quit the REPL console and connect it to the target process again. You will be able to use your classes in the JAR as normal:
import mylib.foo.Bar
...
Scalive uses the Attach API to tell the target process to load an agent.
The agent then creates a TCP server to let the Scalive process interact with the target process. The Scalive process acts as a client.
See also liverepl, a similar REPL console for Clojure.
For simplicity and to avoid memory leak when you attach/detach many times, Scalive only supports processes with only the default system class loader, without additional class loaders (Ex: normal standalone JVM processes, like Play or Xitrum in production mode).
Processes with multiple class loaders like Tomcat or SBT are not supported (with SBT, you already has the SBT console, so it's not a big deal).
These features will be added in the future: