On 2010-12-08 11.03, Niclas Hedhman wrote:
That doesn't solve the "Re-index" issue, does it?
It solves the problem that was described, as far as I can tell, that is, how a bootstrap service can determine whether to create the initial data or not.
Another thing we could consider is supporting "isAvailable()" for the EntityFinder, which goes false during Re-indexing, and have that checked either directly by client code, and toss an Exception at query.iterator()...
That is one option. Another that I have been playing with for my event stores is that a backup restore operation actually creates a temporary datastore first and fills that, and when it's done it switches to it in one short operation. This ensures that if a restore fails in the middle it won't corrupt the existing data. The same logic could apply to reindexing, and might actually allow you to do reindexing more regularly as part of normal operations.
/Rickard _______________________________________________ qi4j-dev mailing list [email protected] http://lists.ops4j.org/mailman/listinfo/qi4j-dev
