Yep, now that you mention it, tiny bundles were what Toni Menzel of the PAX team suggested using when we tripped over a similar issue with the SMX integration tests during the Karaf switch-over.
Cheers, Eoghan 2010/1/22 <dav...@apache.org> > Thanks for the heads up Eoghan. > > I understood that the recent tiny bundles integration could take care > of this. Have a look at the bottom example in > http://wiki.ops4j.org/display/paxexam/ExamAndTinybundles. Do you think > that will do it? > > Cheers, > > David > > 2010/1/22 Eoghan Glynn <eogl...@gmail.com>: > > > > David, > > > > One thing to note about pax-exam is that its doesn't AFAIK have a feature > > analogous to the spring-osgi-test support for accessing and adding to the > > manifest for the on-the-fly bundle. > > > > Now I don't know whether we could possibly live without this > > manifest-mangling as currently done by the dOSGi systests, specifically > > setting the DynamicImport-Package header to "*". > > > > If we can live without this setting, no worries. If not, its something to > > consider about adopting pax-exam. > > > > Cheers, > > Eoghan > > > > > > 2010/1/22 <dav...@apache.org> > >> > >> Hi all, > >> > >> One of the things to-do for the DOSGi refactoring work that's > >> currently happening on trunk is to re-enable the system tests. > >> Now both Sergey and I have been fighting with the spring-osgi based > >> testing system that's there before and I can tell you it's generally > >> no fun at all. The biggest problem that I've encountered with it is > >> the interference of the test framework with the CXF-DOSGi code as they > >> both use Spring-DM but sometimes have conflicting requirements. > >> > >> So I'd like to spend a little bit of time refactoring the system tests > >> to use Pax-Exam. I haven't used it in anger yet but I've heard good > >> things about it and it should not suffer from the interference > >> problem. > >> > >> Thoughts anyone? > >> > >> David > > > > >
