From: Andy D. <an...@ma...> - 2004-05-24 17:17:46
|
As mentioned previously, I'm currently developing a rich client. One of the things this rich client must do is present to the user a unified UI that is composed of information from various services. For example, an employee's salary might come from the payroll service, while the employee's security roles will come from the security service. To the end user, it is all one unified front, while in reality there are many back end services that need to be called into and managed. On top of that, the UI must be dynamic... not all users will be allowed to see an employee's salary or their security information, in which case I don't need to waste the bandwidth gathering this information (anyway, I would get a security exception if I tried). I'm guessing this could be a fairly common issue in rich clients: building a UI that is a composite of many POJO DTOs from many different services. What I would like is the ability to dynamically compose "logical objects" at runtime that are a composite of actual POJOs from various services. For example, a logical "employee" object might contain a salary field from the payroll service, a securityRoles field from the security service, a name field from the HR service, and so on. After digging around a bit, I discovered the Service Data Objects specification from IBM+BEA (which has now been made into a JSR). This specification seems to address this very issue. I'm wondering if it would be worth the effort to offer any kind of integration with this spec in the spring-rcp project? Of course the spec is very new and still evolving, but when has that ever stopped open source? :-) More details can be found here: Original Spec: http://dev2dev.bea.com/technologies/commonj/sdo/index.jsp JSR 235: http://www.jcp.org/en/jsr/detail?id=235 theserverside.com discussion: http://theserverside.com/news/thread.tss?thread_id=22646 |