JRubyHub is a Rails 3 application that showcases JRuby and Java technologies mixed with Rails.
We're planning to run a live instance of JRubyHub on the http://jruby.org site eventually. Currently we're in heavy development mode.
- We're using a custom fork of Spork for fast re-running of Cucumber
features and RSpec examples in a single JRuby process. We had to
tweak some things to get this working clean:
- Nick's Spork fork features a single process looper (originally by Roger Pack) that also has to reset the Cucumber and RSpec shared state each time so it doesn't report previous runs' results.
- Set
config.cache_classes = falseinconfig/environments/test.rbso that code gets reloaded in the test environment. Cucumber warns about this breaking its use_transactional_fixtures method, but we're not using transactional fixtures here. - Run
::ActiveSupport::Dependencies.clearin the Spork.each_run block to reload code.
- To run the Spork servers, type
jruby script/autodev. This starts up two Spork servers, one each for Cucumber and RSpec, as well as a Watchr script (script/project.watchr) that pings Cucumber and/or RSpec every time the appropriate files are touched.
JRubyHub uses Neo4j for data storage. To specify the disk path where
data should be kept, set the Java system property jrubyhub.home. The
path defaults to the system-wide temporary directory found in
'java.io.tmpdir' + '/jrubyhub' if not specified.
To use Warbler to create a .war file of JRubyHub, you'll need to perform the following steps in order to work an issue with Warbler and Bundler interaction:
$ bundle install --without test
$ warble
If you want to go back to develop-and-test mode, you'll need to
bundle install again.