Hi, I'd like branch my 'Are all NodeBuilders required to inherit from the MemoryNodeBuilder?' thread to put the focus back on the root issue.
Some repository/workspace initializers unconditionally wrap a passed NodeState with a MemoryNodeStore, given that up till now all NodeState instances extend the MemoryNodeState. [1][2][3] However, once we get a NodeState that does not inherit from a MemoryNodeState it all breaks down - like in my multiplexing POC. I currently have a hack^H^H^H^H isolated way of exposing a MemoryNodeBuilder from a MultiplexingNodeBuilder, but I'd like a more elegant approach. I wonder if the issue is that the initializers don't get access to the 'original' NodeStore and we can extend the API to make it available? Or maybe there are other ways of addressing it that I don't see due to my limited exposes to Oak's core. Thanks, Robert [1]: https://github.com/apache/jackrabbit-oak/blob/1fdae3a77e4172cf5716 6ffe77eb35a4bd93c76b/oak- core/src/main/java/org/apache/jackrabbit/oak/plugins/nodetype/write/Ini tialContent.java#L118-L119 [2]: https://github.com/apache/jackrabbit-oak/blob/1fdae3a77e4172cf5716 6ffe77eb35a4bd93c76b/oak- core/src/main/java/org/apache/jackrabbit/oak/security/user/UserInitiali zer.java#L94-L95 [3]: https://github.com/apache/jackrabbit-oak/blob/1fdae3a77e4172cf5716 6ffe77eb35a4bd93c76b/oak- core/src/main/java/org/apache/jackrabbit/oak/security/privilege/Privile geInitializer.java#L58-L59
