Drip is a launcher for the Java Virtual Machine that provides much faster startup times than the java command. The drip script is intended to be a drop-in replacement for the java command, only faster.

Drip is a single bash script and a little bit of C and Java code. It is intended to work with any JVM-based language and anywhere bash is available.

Unlike other tools intended to solve the JVM startup problem (e.g. Nailgun, Cake), Drip does not use a persistent JVM. There are many pitfalls to using a persistent JVM, which we discovered while working on the Cake build tool for Clojure. The main problem is that the state of the persistent JVM gets dirty over time, producing strange errors and requiring liberal use of cake kill whenever any error is encountered, just in case dirty state is the cause.

Instead of going down this road, Drip uses a different strategy. It keeps a fresh JVM spun up in reserve with the correct classpath and other JVM options so you can quickly connect and use it when needed, then throw it away. Drip hashes the JVM options and stores information about how to connect to the JVM in a directory with the hash value as its name.

The following instructions assume that ~/bin is on your $PATH. If that is not the case, you can substitute your favorite location.

Standalone — We recommend this to get started quickly.

curl -L http://drip.flatland.org > ~/bin/drip
chmod 755 ~/bin/drip

Checkout — If you want to hack on Drip or follow the latest development, this is the way to go.

git clone https://github.com/flatland/drip.git
cd drip && make prefix=~/bin install

Homebrew — This is a convenient way to brew drip on OS X.

brew install drip

Note: Installing brew requires gcc. Here are instructions for how to install it on OS X Mountain Lion.

You can call drip with the same arguments as java. Try it. The first time you execute drip with new arguments, it will take as long as a plain java command, because it has to spin up a JVM from scratch, but after that it will be fast.

For example, to start a Clojure repl with drip:

drip -cp clojure.jar clojure.main

The Drip JVM will eventually shut itself down if you never connect to it. The time limit defaults to four hours, but you can change this by setting the DRIP_SHUTDOWN environment variable before calling drip to set a timeout, in minutes:

DRIP_SHUTDOWN=30 drip -cp clojure.jar clojure.main

This creates a Clojure repl as usual, either by starting up a new one or connecting to a waiting JVM. But the JVM that is spun up to serve future requests with the same classpath will have a 30-minute timeout to deactivation.

For more information about how to integrate Drip with your favorite JVM language, check out the wiki.

Drip supports the following advanced settings.

By default, Drip only loads your main class at startup, but you can tell Drip to run additional code at startup. This can be used to load classes or execute any initialization code you like. For a language like Clojure, which compiles code on-the-fly, this can be used to precompile commonly used code by requiring it.

To tell Drip how to initialize a new JVM, use the DRIP_INIT and DRIP_INIT_CLASS environment variables. DRIP_INIT should be a newline-separated list of args to be passed to the main() function of DRIP_INIT_CLASS. DRIP_INIT_CLASS defaults to the main class the JVM was started with.

Sometimes, you need to set Java system properties, but you don't want them to be included in the JVM options used for hashing. In this case, use two dashes instead of one, and the options won't be passed to the JVM at startup, instead they will be passed at runtime. Keep in mind that any system properties passed this way will not be set during initialization.

Drip passes all environment variables exported at runtime to the JVM and merges them into the map returned by System.getenv. Keep in mind that the environment isn't modified until we connect to the JVM; during initialization, the environment will be derived from the previous process that launched the spare JVM.