Minor Description
Default data cache operation when it is full is to randomly toss an entity. For queries that return >1 entity, this means the most recent query is not guaranteed to be cached in its entirety.
Jeremy provided this code to me which certainly works, and should be integrated into the base data cache support...
1) Create a new plugin class: (bundle it with your test app)
package test;
import org.apache.openjpa.datacache.AbstractDataCache;
import org.apache.openjpa.datacache.ConcurrentDataCache;
import org.apache.openjpa.util.CacheMap;
public class TomsCache extends ConcurrentDataCache {
protected CacheMap newCacheMap() {
return new CacheMap(true, 1000) {
protected void entryRemoved(Object key, Object value,
boolean expired)
};
}
}
2) Enable your plugin using a persistence property.
<property name="openjpa.DataCache" value="test.TomsCache(CacheSize=5, SoftReferenceSize=0)"/>
The oldest entry should get evicted first.
