rtfm is not all info there is. There are also a lot of unittests in the source that can shed some light on these issues.
The question I have is: "what do you mean by launch it in the server?"
Connecting to the server via webservices is not possible as you probably have (not) read in the fine manual (since it is not in there) and you probably have seen the jira issues on implementing those and read the forum posts about them.
What I mean by launch in the server is:
1) I use the designer to create a process and unit tests that work as I intend
2) I deploy the process to the server, it works as I intend, I can interact with it from the console
3) Now, I want to write a class that starts a new instance of the process in the server. That's all. That class will be imported into a drool so that, under the right conditions, we will launch a new instance of the process.
In other words, now that I have successfully created, debugged, and deployed a good process, how do I launch a new instance of it programmitically? All I want to do is exactly what the jbpm console does: possibly get a list of deployed processes, and start up a new instance of one.
This exact use has to be the raison d'etre of jbpm, and the web console does it, so I figure I must be missing something really basic and obvious to not grasp how I correctly do jbpmContext.getProcessInstance(processId). The only way this works when I try it is if I first parse the process, create the schema, store the process, and then get it back, which means the instance of the Jbpm server I am talking to is my own, not the one used by the console, not the one started up by start.bat
3) then just write a class that does that... the most basic examples do exactly this.
- open context
- search process instance
- do work
- close context
(do not have an explicit example at hand)