On Fri, Apr 24, 2009 at 10:11 PM, Luciano Resende <[email protected]> wrote: > On Tue, Apr 21, 2009 at 11:39 PM, Ramkumar R <[email protected]> wrote: >>> >>> This would be ok for now, but the proper fix would be to have pojo >>> related interfaces available in a separated module that could be >>> exported via OSGi and used by all pojo related extensions (e.g java, >>> ejb, spring). I have started doing that in implementation.pojo, but >>> that still need some more work, and then migrate the related >>> extensions to use it. >> >> Hi Luciano, >> >> That seems to be a good idea, as you said that would also benefit the jee >> modules which also needs the java introspecting techniques. >> >> If you wish, I can go ahead and create a seperate module as you have >> suggested. >> As I am currently working in 2.x, there should not be any issue in taking >> this up. >> > > I finally found the work I have initiated for implementation-pojo at > [1]. I'd probably recommend renaming the module to something like > core-pojo to avoid confusion with implementation-pojo sample that we > have in 1.x. Note that the current changes I had to the pojo based > extensions to use implementation-pojo is available as a patch in > TUSCANY-2988 and you might need to review and complete it before > applying these changes to avoid any breakage. >
I agree implementation-pojo doesn't seem like the perfect module name - its not an implementation and they're not POJOs. Coming up with a good name might be easier once we can see all the code that goes in it, so not worry too much about it and use any old name for now. From what i can glean from whats been said about this is that its just a way to share some common SCA introspection code so i don't think it needs to be engineered too much with interfaces and pluggable implementations and factories etc, and as its not going to have any new external dependencies it might even just fit as a new package in the existing spi module. ...ant
