Jean-Sebastien Delfino wrote: - A simpler metadata model (the recursive model is much simpler than the 0.9 spec, this is a great opportunity for us to simplify our implementation)
I was chatting with Oisin at ApacheConEU about the model that they were using in the STP project. They currently have a model of the 0.9 spec and he was wondering if they could leverage the changes that we have made to support the 1.0 model. Having a simple POJO-based model was appealing to both of us due to flexibility that it provided and the accessibility that it provided to developers. You didn't need special tooling or codegen to work with it (although tooling could be used if you had it) which made it open to the community. We recognized that the two projects had some different requirements. In a runtime situation you don't want to be digging around a model on every request (as the model changes very infrequently compared to the number of requests you process). However, a tooling environment is all about manipulating the model so you want UI components to be notified for every change that is made. Our thought was that if we captured metadata about the relationships in the model, the tooling runtime would be able to monitor them and notify components as needed. There was a lot of work done in JSR-220 for this and we were wondering about reusing their relationship annotations. This would give us a couple of interesting benefits. Firstly, the model would be directly persistable to a database using any JPA implementation. Secondly, the same model (POJOs) could be used in the tool environment by generating an ECore representation from the embedded metadata (I'm not sure quite what that means, I'm channeling Oisin here). Thirdly, the same model could be used to create SDO type definitions from Java classes allowing us to jump-start that work. -- Jeremy --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
