IronJacamcar 2 / Deployers

Activation metadata

The ironjacamar.xml and -ra.xml files will be converted into an activation metadata type, which is used in the deployer chain.

The ironjacamar.xml will be converted into 1 activation metadata entry, where -ra.xml will be converted into X activation metadata entries based on the number of resource-adapter elements there are.

Any services which deploys at run-time will use 1 activation metadata entry to describe the activation.

Deployer chain

The deployer chain will contain a state machine which will "progress" the deployment along while filling out the deployment structure for the deployment repository.

For standalone and embedded the state machine and deployment structure will be kept in the

com.github.fungal.spi.deployers.Context

instance attached to the deployer chain.

The deployer chain must be as fine grained as possible, with clear functionality separation, like

Create java.io.File descriptor (extract archive)
Parse META-INF/ra.xml
Parse META-INF/ironjacamar.xml
Parse -ra.xml
Create ResourceAdapter instance
Create ConnectionFactory instances
Create AdminObject instances

and so on. The priority of the deployer must be defined in the central location.

All the deployer chain modules must delegate as much as possible to an abstract class or static library methods in order to re-use as much functionality as possible across deployment models.

Data from each step should be recorded in order to maximize feedback to the end-user once the deployment has been processed successfully or rejected. Validation feedback of the created or used objects must be included in these reports.

The deployer chain must be documented in the developer guide.