Playing around with the Ruote workflow engine. The example in this workflow does not really do anything. I just created something to model a similar workflow that I am using on a work related project.

There is currently no GUI accept for RuoteKit running under http://localhost:3000/_ruote

Application depends on mysql. sqlite will not work

bundle install

rake db:migrate

Run the specs: rake spec

Start Workers: rake ruote:run_worker

In another terminal start up the rails console and create a RequestItem. You should see the some activity in the worker terminal as it starts the workflow. After a few seconds, reload the RequestItem that was created and check the state. If everything went ok it should be "completed"

The workflow is defined in RequestItem::PDEF_PROCESS. A RequestItem does not represent anything in particular. It is just something that has to be run through this workflow in order to be processed.

In order for a RequestItem to be processed it needs to go through the following workflow:

1. Call Service1. This is a participant that represents a web service call to "service1"
2. Call Service2. This participant represents a web service call to "service2"
3. Go into a monitoring phase. This will monitoring something until the particpant decides that it is done
4. Do some post processing after monitoring.

At certain points in the workflow, the RequestItem's state will be updated. A RequestItem should end up in the "completed" state when the workflow is done.

At any point that an error happens, the workflow will put the RequestItem into an error state. The error will need to be reviewed by a RequestItemAdmin. The admin can choose to reprocess the request item or cancel the process. Reprocessing the item will start the workflow up again.

The participant implementations are just placeholders and do not contain any real logic besides responding to the engine.

All particpants that represent service calls will be continually retried until they respond or an error is thrown. I modelled it this way to have automatic retries if the service is not up.