Simplify creating identity tabels for other db and filling t

[kukeltje]

This is not so much a design issue for the core or webapp, but it is (imo) for the build system.

Working from CVS head (3.2) it is to cumbersome to get the identity tables created. Maybe I realy mis something, but the ddl scripts do not contain the jbpm_id_user etc tables. The load identities task works using some not (at least not easy) configurable hibernate file so it uses hsqldb.

This should be easy configurable and I can spend some time on it if someone tells me what the ideas are. I'd thought of creating a jira issue for this immediately, but wanted to discuss it first.

[tom.baeyens]

the thinking behind this was the following:

identity and jpdl are two separate independent sub projects.

it is the suite (previously called the starters-kit) that will combine the two components.

therefor i thought it would be best to remove the dependency that was introduced by adding the identity mappings to the default mappings as they are specified in the jpdl/jar project.

the mappings should be there in comments in the default hibernate configuration file. so that you can create an ant script with some replace tasks that uncomments the identity mappings.

does this make sense ?

[kukeltje]

Yes, I've seen that, but
- loadidentities it is currently hard connected to hsqldb since it uses a specific config file (not configurable)
- the identity component is not easily replaced (it is not pluggable as a service!!), people think it is (my experience from the forum)
- people want to demonstrate jBPM with *their* db and initally do not care about the identity being there, they even 'like' it

If I want to have it in mysql (the demo we have does NOT perform with hsqldb), I have to do 'a lot' of work manually.

If it would be possible to e.g. pass a parameter to some ant tasks that say 'use mysql' and configure e.g. the properties for the correct database. The correct scripts for just that DB, including the (optional) identity scripts (another ddl?) and create a loadidentities that uses this db config to populate the database would be nice.

So to summarize, the separation is nice and makes sence, but it should be easy to generate the identity ddl AND load some default (identities.xml) for ANY database.

I'm looking at CRUD for the ui, but if CRUD is implemented for hsqldb, I think the other databases should not be forgotten.