On Wed, Apr 22, 2009 at 7:57 AM, Luciano Resende <[email protected]>wrote:
> On Tue, Apr 21, 2009 at 11:52 PM, ant elder <[email protected]> wrote: > > > > > > On Wed, Apr 22, 2009 at 7:50 AM, Luciano Resende <[email protected]> > > wrote: > >> > >> On Tue, Apr 21, 2009 at 11:37 PM, ant elder <[email protected]> > wrote: > >> > > >> > I don't actually see any serious issues with fixing this by just > >> > changing > >> > the export. It doesn't make much difference if the spring extension > >> > depends > >> > on implementation-java or implementation-pojo does it? > >> > > >> > >> By doing this, you are allowing modules to have direct dependency on > >> your implementation artifacts breaking OSGi modularity. As I > >> mentioned, this would be ok during bring up, but I expect the proper > >> fix to be available in the future. > >> > > > > Its not "breaking" anything, its going to be depending on the same code > > whether its in one module or somewhere else. > > > > It' s not breaking because he will be hacking the OSGi manifests, > which is OK during the bring-up phase, but not as a final solution. > > Lets not argue if this is going to be changed anyway, but just labeling things "breaking" or "hacking" doesn't make it something true. A valid export is what *we* decide is what we want the module to export. Any new module is going to use exactly the same export. ...ant
