I don't know if it might help but there is already a ReferenceTracker class used in WorkspaceImporter. Maybe you could extend it and reuse it.
why should that help improving the changelog in respect to references? this tracker does something completely different. it records altered uuids of imported nodes in order to be able to adjust the reference properties. regards, toby
BR, Nicolas my blog! http://www.deviant-abstraction.net !! On 11/21/06, Thomas Mueller <[EMAIL PROTECTED]> wrote: > > Hi, > > In Jackrabbit, there is currently a limitation on how many references > (properties referencing to this node) a node can have. The reason is > that node references are kept in an in-memory list when getting the > references, and also when adding/deleting a reference. > > I would like to work on this problem. There is already a Jira issue > about it, but it is closed as 'Won't Fix': JCR-74 "NodeReferences > handling". I could create a new issue, or this issue can be re-opened > (it looks like I can not do that however). > > This is somewhat related to JCR-642 "Support flat content > hierarchies", but can be implemented independently. > > My current idea is: instead of reading/updating/storing the list of > references whenever there is a change, use a new class > NodeReferencesChanges to keep track of changes where possible. Then > change the implementation of NodeReferences (that is still used, but > only when calling getReferences) to be file-buffered when it is too > large. > > Of course it can be implemented in another way as well, for example > delegating the reference management to a new component, but the impact > in the current code would be bigger I think (not sure if the amount of > work would be bigger however). > > What do you think? > > Thomas >
-- -----------------------------------------< [EMAIL PROTECTED] >--- Tobias Bocanegra, Day Management AG, Barfuesserplatz 6, CH - 4001 Basel T +41 61 226 98 98, F +41 61 226 98 97 -----------------------------------------------< http://www.day.com >---
