On 2002.01.09 23:38:57 -0500 Dmitri Colebatch wrote: > > On 2002.01.09 21:45:44 -0500 Paul McLachlan wrote: > > > How do I do inheritance with cmp2? Maybe that is why Castor will > survive. > > > > Maybe. I don't know how to do polymorphism w/cmp2. But how can you > build > > a big system w/o decarative tx and security ;-)? > > > > From the little I've looked into the jdo spec, it looks to me as if the > > persistence mechanism is pretty similar to what you need for ejb2. So > > perhaps soon we will see something with both the metaprogramming of > ejbs > > and the oo-ness of jdo. > > can I pop my 2c in here.... take it as you will. > > firstly, David, without seeming rude.... you do know that castor doesn't > aim > to be compliant with the JDO spec, but rather aims to solve the same > problem > in a similar fashion (my understanding)
No problem. I still think the jca spi interfaces are the ideal way for anything with "connections" to interface with an app server. I was kind of disappointed they didn't respond with any interest to my offer to potentially implement it. > > we recently switched back to entity beans, after having a dabble with > castor, attracted by the rich feature set, mainly the polymorphism, > shared > locking etc.. however, the further we got into it, the more problems we > had, specifically with the polymorphism, which was the main advantage it > has > over entity beans. we switched back because it was decided we couldn't > spend any more time looking into the aforementioned issues. > > having said that, castor can run inside an ejb container, and be used in > conjunction with session and message driven beans, and still execute in > the > same tx scope. yes you lose the declaritive tx scoping, and security, > but > I'd argue the pros and bigger than the cons..... except of course that > you > are using something that is completely non-spec (but very portable). Whether or not you implement the rest of the jdo spec, using their ideas on jca does or would give you declarative tx scoping and security at the ejb-castor boundary. Ejb2 extends this declarative model down to every object, while reducing the possibility of support for polymorphism. > > I'm not really arguing one way or the other, just sharing our experience > with it... and perhaps my thoughts on what is and isn't important in the > decision... I'd be happy to hear of others' thoughts on both those > issues. I certainly appreciate hearing your experience! Thanks david jencks > > cheesr > dim > > > _______________________________________________ > JBoss-user mailing list > [EMAIL PROTECTED] > https://lists.sourceforge.net/lists/listinfo/jboss-user > > _______________________________________________ JBoss-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jboss-user
