I have a (non-osgi) application that uses commons-digester to
construct/inject runtime objects.
I want to adopt OSGI and restructure this application into a set of bundles.

It is unclear to me how I can use commons-digester if I break this
application up into bundles with only the interfaces exposed as API.
Specifically: If digester is being used by one bundle to
construct/inject it's runtime objects, I don't see how digester would
have access to the implementation classes hidden in other dependent
bundles.

The only way that I can see this working is if all of the dependent
bundles export all of their packages.  This way digester should have
the access it needs to the target classes.
I would prefer to hide all implementation and just export the API
packages in my bundles to reduce coupling.

Questions:
1. Am I understanding this correctly?
2. How can I balance desire to hide implementation with desire to
continue using commons-digester as injection engine?
2. What other options/approaches should be considered?

Thanks for any advice.
T
_______________________________________________
OSGi Developer Mail List
[email protected]
https://mail.osgi.org/mailman/listinfo/osgi-dev

Reply via email to