Hi, Welcome to the 7th edition of the informal news on what's going on in Oak.
* There is an initial effort implementing JCR versioning. See the discussion on different approaches and content structures at http://markmail.org/message/k7bbvwvf25irn6jt. This discussion also revived an older discussion on how we should handle JCR work spaces in Oak. Expect more at http://markmail.org/message/bqwi2xiu3zg73kdg.
* The MongoDB based Microkernel implementation is now better integrated with the normal build. The related components are included in the build by default. Tests are run if a local MongoDB instance is available. Otherwise they are simply skipped: http://markmail.org/message/bczjttkj6edinwk3
* As a new, optional feature the Microkernel no supports to retrieve items by identifier/hash. This has been introduced in order to avoid extra traversals when items are needed whose parent has been retrieved earlier already. See http://markmail.org/message/nnsyvzfhy2k26xvb for the discussion and https://issues.apache.org/jira/browse/OAK-468 for the related issue.
* We started to run the Jackrabbit query tests on Oak. This gives us much better test coverage and will help us identifying backward compatibility issues. See https://issues.apache.org/jira/browse/OAK-237.
* There was a discussion on whether Oak should support large multi valued properties: http://markmail.org/message/ak7g3mq37kj6hxyw. While there is nothing in Oak itself which would prevent us from doing so, JCR does not have a way to access individual elements of a multi valued property. From this most concluded that it is not necessary to support such a feature for the time being.
* The discussion about larger multi valued property also sparked a discussion on whether we could support hash maps as properties. This would simplify matters for scripting languages. However, the most common view on this was that JCR already supports this by means of the properties of a node.
* While prototyping a way to retry a failed commit, we ran into the deeper issue of move operations not being correctly detected by the way node states are currently being diffed. While this has been discussed at length earlier, the discussion gained new momentum. See http://markmail.org/message/lbc3rx2p3sssvqj5 for the discussion and https://issues.apache.org/jira/browse/OAK-464 for the main issue and a summary of possible ways to tackle this.
* The MemoryNodeBuilder, which is at the core of tracking state changes to the content tree, has been refactored to make the code easier to follow. See https://issues.apache.org/jira/browse/OAK-452.
* There is an ongoing effort to get repository initialisation right. The main problem here is its interaction with the commit hooks. While during initialisation some hooks might already be needed, others might be in the way. See https://issues.apache.org/jira/browse/OAK-410.
Michael
