Minutes of the JEE 5 Working Group meeting Jan 11, 2007
Teleconference on JEE 5 Support in WTP Jan 11, 2007
- Naci Dai - Eteration
- Rob Stryker - JBoss
- Rob Frost -BEA
- Kaloyan Raev - SAP
- Hristo Sabev - SAP
- Chuck Bridgham - IBM
- Neil Hauge - Oracle
- John Goodman - IBM
- Paul Fullbright - Oracle
- Meeting frequency
- Review M5 plan items for Facets and Projects
- Comments and feedback on JEE 5 Model Design Doc
- Other items
- ND: The first thing on our agenda is the frequency of these meetings. I recommend that we reduce the meeting frequency to once every two weeks or once a month if our agenda is not full.
- CB: Agreed. We still have a few things to discuss next week. But we can call extra meetings if there is a need.
- ND: OK then.
- ND: Second on the agenda is a review of the M5 plan items for Facets and Projects and Models. Chuck, can you provide a summary?
- CB: EJB3 and EAR5 facets are in the build and Units tests are working. Default facet is still ejb 2.1,ear 1.4. That makes sure that all existing code works. We should also have at least a model registration mechanism in place by M5 and capability for extending these models. I will concentrate on registration of new models. I need adopters to try it and provide feedback. There are few more things that can be done with new project types. We will also add some more new facets like servlet 2.5.
- ND: Can SAP try these works?
- KR: If we get a link to the builds we can try
- CB: There is not much now. Currently we can add a facet, there is no modelling support.
- KR: So, if I have a runtime that has ejb3 facet, would the project pick that runtime?
- CB: I am not sure. But it should.
- ND: To summarize, we will have facets, projects and JEE 5 models and extensibility in M5.
- CB: I'd like people to try these, and I'd like to hear from people what is impeding their progress. What do they like to see? We would really appreciate comments. We would like to know that we are doing things that they need. For example, are these useful to the Dali project?
- NH: I'll look at what is there, and provide a definitive list.
- CB: We will also fix the translator framework. Jesper updated his patches. These were also needed by Dali.
- PF: We need some documentation for translators. There is the bug that claims to work but does not. Also, are going to have support for plain java projects? i.e. Convert a plain project to a Utility? We also would like to have capability add libraries to a runtime.
- RS: We have an issue with EJB3 conflicts. For example you cannot add ejb facet to a dynamic web project. This is OK for EJB 2.1 but we need it for EJB3. JST facets conflict with group module (jst.web, and jst.ejb), for EJB3 it should work.
- CB: Yes, it should probably be different than 2.1.
- ND: This maybe ok for JPA but EJB3 should be still packaged as an EJB-JAR. and deployed to the server separate from a WAR. So it would conflict. We do not support multiple JavaEE modules per project.
- CB: Maybe we will need a new fucn. to package the JAR. I don't know. We should open a defect.
- ND: Next item on our agenda is the comments on SAP design doc. See bug (167807).
- ND: This proposal has many pluses, it provides capability to have different models and modeling technologies to co-exist. WTP currently provides EMF based models only. It provides a capability to extend these models. My concerns are mostly for scheduling this work, chuck already mentioned h will be able to provide something in M5 to meet some of these requirements. We should Review available DOM-like modeling technologies which can offer similar capabilities before we start building our own. The proposal is extensive, WTP v2.0 time frame is too short to do it properly. however, we can review what can be done for v2.0, so that existing JEE5 models may be wrapped as a "model element". At least we should not prevent an adopter to go ahead and implement this concept.
- CB: I divide the issues into two - 1) What we need for JEE5 and 2) Model extension ( a pluggable model registration)- Common interface of them. Models can have their own persistence mech (xml, annotations, etc.) and they can be extended and traversed. It looked liked these were the requirements. We had some of this implemented in EMF, but without having a generic registration. We should focus on defining specific interfaces for JEE5, and then have means for registering your own. If we can agree on interfaces we can go from there. They can have their own serialization and notification, and multiple resource management. EMF has some of this in it already but hard to extend it in this release. To wrap it up, the modeling infrastructure is more appropriate for another project, we should focus on JEE5 and satisfy the requirements for extensibility.
- HS: From our pt of view this is OK. We can extend it. You allow a Domain interface to be defined and not tied to EMF. We can Find nodes, and provide extensions. The only thing is what you propose is extend the model, what we suggested was to associate. If we can provide additional semantics that would work.
- CB: That is what I meant also. Not really inherit from them. EMF provides things like that.
- HS: That is good news. I will need more info on EditModel.
- CB: I can send you few more details.
- ND: Any comments from BEA or Oracle?
- RF: I read your comments, We agree with them and do not have much to add it for now.
- NH: Chuck you mentioned that there was WTP driver that has the EJB3 facet, which one is it?
- CB: The latest from today's build.
- ND:- Thank you for attending this call. We will meet again next week
Please add your comment here: