Hi, On Jan 11, 2008 9:22 PM, Padraic Hannon <[EMAIL PROTECTED]> wrote: > From Jukka's email it sounds like everyone is pushing NGP out for > a long ways.
I wasn't speaking for anyone else than myself. My personal impression (as the main author/proponent of the NGP proposal) is that NGP (or anything like that) won't be ready for prime time in quite a while yet. > I wonder if perhaps before adding too much more functionality > (ie JCR 2.0) it may be better to start exploring implementing NGP sooner? Not really. We have a lot of things to do before JCR 2.0 and to remain the reference implementation we need to have at least somewhat functional implementations of all the new features in JCR 2.0 by the time the spec is finalized, so I definitely don't want to start messing with Jackrabbit internals too much at this stage, especially as there's no functional deficiency in the current internals that would prevent us from implementing all of JCR 2.0. There are also a number of open questions regarding NGP, like how to best handle indexing and queries; will it perform well enough; what are the storage requirements; etc. Currently I'm slowly prototyping the proposal in the jackrabbit-ngp sandbox project, and before we have at least some results from the prototyping phase I don't want to start pushing for jackrabbit-core to adopt the NGP architecture. If you're interested in helping out, feel free to review the NGP proposal at [1] and the current rough prototype at [2]. Any comments, patches, ideas, alternatives, etc. are welcome. However, it's not like the NGP work will be totally fruitless before 2010, for example the DataStore feature in Jackrabbit 1.4 is directly inspired by NGP concepts and I expect the prototyping phase to result in other such ideas that we can use already with the current Jackrabbit architecture. [1] http://jackrabbit.apache.org/dev/ngp.html [2] https://svn.apache.org/repos/asf/jackrabbit/sandbox/jackrabbit-ngp/ BR, Jukka Zitting
