ENTRY POINT CONTAINER DEFINITION

We are running Java and PL/1 in an application context.

As the developers have started to integrate Java in our environment we have the need of an communication between two applications.

This all works fine, if you are not shipping data from one language to another one.

We now have discussed how our interface should look like. We think the interface must be described in and .xsd file, because then you can generate PL/1 structures and JAVA Classes. The data in the container must be shipped in XML Form, because this is best for both sides and the interface. Also it would be possible to run automated tests against the interface with an xml. With structured data this is difficult.

We now have two ways:

a) JAVA to PL/1

1) Put data to java objects and convert the object with JAXB to a xml string.
2) Put the xml to an container and call the PL/1 Program
3) call an XMLTransform in PL/1 to get the structured data out of the container.

b) PL/1 to Java (the same way in the other direction)

This all is working, but in fact you have to defne a lot of XMLTransforms and I thiink at the moment the xmlTransform is not versionable (not sure about this).

What we want is a versionable way to do this transformation at the interface point itself, so that the programer has not to code anything in the program.

We want to define an entry point and add a container with an xsd/binding to the entrypoint.

If you now call the entry point with an invoke application or an cics link and the interface realises that a container with this name exists, then it should automatic transform the data. If the program flow goes back, the data should be automaticly converted back in the same way.

On the java side (liberty) you could provide an @inject mechanism to automaticly generate objects out of the container.

As this would be included in an applicaiton, you would get an well defined interface, which is easy to handle for the developpers.

This also would have a big value for the 5.3 comunication thing....