Two promising technologies in two days and both do not seems to be going the way i think this development process should move. This process comes closer to my idea of how ultimately the request processing is a workflow which uses component to get the work done. This idea comes quite well in this technology. My issue is with reinventing the wheel. I would have really like the product to use a workflow definition language for achieving the request flow and data flow. On an initial review it looks like it borrowed the idea from the BEA/Apache Beehive (did i get it right?). But I would really have loved the idea of extracting database schema or from entity relationship diagram and generate objects or vice-versa automatically and drawing the request and data flow using GUI instead of editing xml and in a workflow language instead of developing your own.
The meta-data about data constraints is fine but that can not be extended to Web interface. Now what needs to be displayed as editable or non-editable and sorting decisions are not a business logic decision (as it can be an authorization decision) and thus should not live with the bean definition. It is a interface decision and should be part of that!! This is where even I am stuck w.r.t. to how to tie the workflow to interface. What is the answer? But that is a separate topic...
The technology does look promising and can work as inspiration for other technologies...
No comments:
Post a Comment