True, Regarding working from your source code - I am currently looking into writing a pluggable module for XDoclet that interprets the JavaDoc tags I genereate from Middlegen. This should allow one to to modify the mapping in the Java source to generate a more accurate repository.xml. Generating the ddl is a nice idea. One should be able to generate Torque templates from the source code as well and then one can create the tables in the platform of your choice. Schema migration does become problematic though - what happens to the existing data?
Maybe a mapping workbench should simply modify the tags in existing Javacode. This would help the novice to do the mapping through a gui tool, but also allow the the expert to code the mapping mannually. But I don't think Middlegen is quite there, although maybe they might be open to the idea... Maybe we should ask them? -----Original Message----- From: Michael [mailto:[EMAIL PROTECTED]] Sent: 07 September 2002 09:34 To: OJB Users List Subject: RE: A mapping workbench >From what I saw of MiddleGen it started with the database. I'm working on a new project and right now I'm designing our object model without even thinking about the database. What I'd really like is a tool that could generate the repository.xml file from my java objects. If it could also create the database schema that'd be great (but if not I could do it by hand I guess). This is one part of EJB Entities I actually liked. I defined my entities and didn't even think about the database schema. Michael > -----Original Message----- > From: Ampie Barnard [mailto:[EMAIL PROTECTED]] > Sent: Saturday, September 07, 2002 9:35 PM > To: OJB Users List > Subject: A mapping workbench > > > Just an idea... > > I recently started playing with Middlegen (available on SourceForge). Its > latest cvs version has a very well designed plugin architecture. It also > comes with a GUI for visual mapping of your plugin constructs (like JDO > classes) to tables and relationships. It was originally written > to generate > the tons of code required to make entity beans work, but it took > me exactly > 2 days to write a plugin that generates most of the elements of the > repository.xml. Currently Middlegen's mechanism is one-way: it just > generates java code/xml. > > Now here is the idea... Why not co-oporate with the Middlegen guys to > implement the future version of the OJB mapping tool? I know the OJB guys > have probably already written the code for ReverseDB, so maybe its a bad > idea. (I have not been able to assess ReverseDB, since I can't seem to get > it working on MySQL.) But I also do believe that this could be a mutually > beneficial relationship. > > Would love to hear your ideas on this. > > Ampie > > > > -- > To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> > For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> > -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>
