Matthias, Sunday, July 20, 2003, 6:35:49 PM, you wrote: MB> MDA model transformations (PIM-to-PIM or PIM-to-PSM) are not yet MB> supported by AndroMDA. However, the OMG already proposes them and now MB> you have shown two very nice and simple examples for them, it's easy to MB> see that transformations have an advantage over "simple brute template MB> force". :-)
I had the concept of transformations for some time in my head. Some time ago when I was talking with Anthony Mowers I sugest him a idea to implement a model compilance test using something similar to a transformation. The conformance test will be a transformation receiving the model and looking on it. MB> I think, this would be an excellent feature for AndroMDA 3.x, the new MB> version that will be developed after 2.x has gone final (see my email MB> "Short term plans and long term plans", Sunday July 13, 2003). I already read it. MB> If you MB> want, you should join forces with the AndroMDA developer team and bring MB> this feature into AndroMDA. I see a good chance that it fits into the MB> architecture quite easily, some minor improvements to the existing MB> cartridge system would have to be made. MB> You talked about an MDA tool that you're writing. Is it open source or MB> closed source? If it's open, we should definitely join forces and MB> centralize the features in one tool (AndroMDA) instead of writing MB> several (possibly incompatible) tools. If it's close source, we do not MB> have any chance. It is used only by me. It is integrated with Eclipse (in a very simple way). It do not run stand alone (as a Ant task). If you use Eclipse I can send it to you. I do not have any problems on make it open source or join AndroMDA. I have 2 things I do not like on AndroMDA : - the need of write a lot of templates. I hate templates. They can look as a good idea in the begin but then they become difficult to write and debug. In fact to make it usefull you must have some processing on the model to create something easy to print on the source code. A example of this is the "$transform.getOperationSignature($op)" operation. But with the use of model transformation this is not a problem. Because I will not need to change the templates everytime. In fact if we can ever make a single template to java classes (<<JavaClass>> stereotype) who will create all java code for us. It can be xdoclet, hibernate or POJO classes. - what I really hate on AndroMDA if the use of Netbeans MDR. I really think is best to have a simpler model. On a TheServerSide post you talk about AndroMDA only care about the model and ignore diagrams. The Netbeans MDR is full of diagrams details. We need only the model. I really miss the SimpleOO of Uml2Ejb (but I do not miss JAXB problems, of course :) ). To the transformations be easy of write (and debug) we need they to operate on a simple, clean model. On my tool I use Castor to make the xml binding. This way I can save the model after apply a transformation to it, so it is really easy to see how it is working. And MDR make impossible to work with UML tools who do not follow is standart. Using something like simpleOO I can make a simple translator. In fact I do not see any advantage on use MDR. So if you accept the idea of transformations (what, I think, you already did) and the replacement of Netbeans MDR by something simpler (something like SimpleOO using Castor) I am sure I will join the AndroMDA 3.0 team. Or convince me of Netbeans MDR be a good idea. -- Best regards, Danilo mailto:[EMAIL PROTECTED] ------------------------------------------------------- This SF.net email is sponsored by: VM Ware With VMware you can run multiple operating systems on a single machine. WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines at the same time. Free trial click here: http://www.vmware.com/wl/offer/345/0 _______________________________________________ Andromda-user mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/andromda-user
