Resolving a bundle in start() breaks ServiceMix

[thomas.diesler]

Ales, could you please document comprehensively in javadoc how this ServiceMix stuff is supposed to work.

This code required by the TCK breaks the ServiceMix tests

      // If this bundle's state is not RESOLVED, an attempt is made to resolve this bundle. If the Framework cannot resolve this bundle,
      // a BundleException is thrown.
      if (getState() != Bundle.RESOLVED)
      {
         try
         {
            getBundleManager().resolveBundle(this, true);
         }
         catch (RuntimeException ex)
         {
            throw new BundleException("Cannot resolve bundle: " + this, ex);
         }
      }

I put the test in the surefire exclude list.

[alesj]

I put the test in the surefire exclude list.

I hope this doesn't become a habit, as that's not why the tests are for.

I'll have a look on why it fails and add a few useful lines to the javadocs.

[thomas.diesler]

The test is re-enabled again. Good javadoc that comprehensively explains how stuff is supposed to work is still needed in this area.

[thomas.diesler]

WRT excluding test:

The TCK Hudson run as well as the offline osgitck-setup work of the latest jboss-osgi-framework-1.0.0-SNAPSHOT-all.jar that has been published to the local maven repository. Hence, all framework tests must pass to see changes reflected in the TCK run.