I think you meant 'EJB Container should be the central piece of a J2EE impl, but I'm not sure...'
At any rate, I'm very happy to throw my hat in the ring on this one. This is a very positive move for Apache, and one that, contrary to some popular belief, is impeccably timed. The spec has hardened enough such that now is an excellent time to build this thing from scratch using industry lessons learned, and leveraging some of the best, most evolved features of the current generation specification. To be honest, I think that the biggest risk to the success of the implementation is with the quality of the middle tier caching and transactional control layers. 'If this ain't right, ain't none of it gonna be right'. I would humbly suggest that the EJB implementation design begin post-haste before anything else. That being said, the design should definitely accommodate the existing Tomcat engine (assuming the community deems that codebase worthy of integration). I would also humbly ask that much of the code be constructed from scratch, or that third-party libraries, if used, be pruned down to the extent that the product does not become overly burdensome and bloated. This project has the potential to catch-on like wildfire (Arizonans and Coloradans forgive the analogy) if a positive user experience as derived from a 'lightweight' (I realize the relativism, but you ol' battle axes out there I'm sure would be with me here), stable implementation of the specification that is portable (one of my biggest gripes with SunOne is that I have to download and install a different version of the server for a given platform - say it ain't so, Sun :o( ) and intuitive. I have to admit that I've only worked on commercial software, and that this is my first open-source implementation experience. That being said, I hope to have many thousands of useful ideas and lines of code code to contribute througout the lifecycle of the project. Best, John C. Dale Professional Services Compuware On Thu, 7 Aug 2003 16:22:30 -0400, "Nanduri, Amarnath" wrote: > > I agree.... > > Cheers, > Amar.. > > Amarnath Nanduri > AgilQuest Corp. > (ph) 804 745 0467 xt 127 > > > -----Original Message----- > From: Alex Blewitt [mailto:[EMAIL PROTECTED] > Sent: Thursday, August 07, 2003 7:36 AM > To: Christian Trutz > Cc: [EMAIL PROTECTED] > Subject: Re: Central piece: to JMX or not to JMX > > > I think a central piece of an > > J2EE implementation should be the > > EJB container and that should be the > > first think to concentrate. > > I don't believe that a J2EE server should be the /central/ piece, > though it should obviously be a main piece. > > One approach would be to configure various services with JMX, which > leads to the question: should components be built on top of JMX, or > should an interface be provided to control via JMX? My preference would > be (initially) for the latter; although projects like JBoss are built > on top of JMX I'm not sure it's necessary (or even desirable) to depend > on JMX directly. (ApacheJ2EE hasA JMX interface, not ApacheJ2EE isA JMX > interface). > > I think a micro-kernel architecture for setting up the server and > configuring the various pieces is probably a good idea though, and that > should be the central piece. The other components can then sit on top > of that: > > + GeronimoKernel > |-- GeronimoJNDIService > |-- GeronimoEJBContainer > |-- GeronimoWebContainer (maybe an interface to plug in Tomcat etc) > |-- GeronimoSecurityService > > and so on. > > Alex. > > al
