The POC for OAK-6575 aims at introducing the AdapterManager and AdapterFactory API in Oak. This is just a step away from introducing a full-fledged implementation of Sling-like adapters in Oak. In fact, the OakConversionService is just an Adapter for a single, very specific conversion.
We had a very similar conversation for OAK-1963 in a thread [1] in this mailing list. In that thread many people raised that concerns about adapters. Many of us, me included, see adapters as a glorified way of breaking data encapsulation and are not in favour of introducing them. After some quiet time in the thread and in OAK-1963, a new thread [2] was created, a new issue [3] was opened and a POC [4] was written. The content of the POC is little more than a permutation of what was already discussed in OAK-1963. Why a new thread and issue were needed? Why couldn't we discuss this implementation in the light of the OAK-1963? Why the old concerns are not taken into consideration once again? We already talked about all that. If the majority of the people here is alright in going forward with the aforementioned POC, so be it. But, at least, let's avoid closing threads and issues and create new ones that are just carbon copies. This might be misinterpreted as a way to clear concerns out of our way. [1]: http://markmail.org/thread/enmibgfwedypjnos [2]: https://lists.apache.org/thread.html/2efb1da82b9b93a3a394c3abccbac960175ffc7a13facaac79c9181a@%3Coak-dev.jackrabbit.apache.org%3E [3]: https://issues.apache.org/jira/browse/OAK-6575 [4]: https://github.com/apache/jackrabbit-oak/compare/trunk...ieb:OAK-6575?expand=1
