Lars Torunski - [20/Jul/05 12:31 AM ] Please provide the patch when you are finished.

Christoph Kutzinski - [20/Jul/05 05:05 AM ] I haven't found any tests testing groups, so I added I method to TestAbstractCache.java Doing this I stumbled upon this: When I have overflow persistence enabled and put an entry with a group (without exceeding the maxEntries limit) the entry is correctly not persisted, but the group is. Is this expected behaviour or a bug?

Christoph Kutzinski - [03/Aug/05 03:39 AM ] These are the changed classes AbstractConcurrentReadCache and TestAbstractCache. I've already checked them in into my local CVS. I don't how how to create a patch in this case, so I send the full files, sorry. I've made some more changes to AbstractConcurrentReadCache: - a new removeForce-Method that will remove the entry form the persistent store in any case (even if you have unlimitedDisk or overflowtoDisk). This is important if you need a removeEntry analog to flushEntry as I do. I call this method in my own Cache subclass - some changes to serialization (de/serialization wouldn't work anyway if one would use it as it is currently implemented):   - listeners should not be serialized   - readObject() must be implemented in subclasses, as AbstractConcurrentReadCache's readObject call the subclasses via put(key, value) and e.g. the list in LRUCache would be null in that moment

Christoph Kutzinski - [03/Aug/05 03:53 AM ] Working with the new remove method I realized that removing a bigger number of entries is incredible slow now if one is using a DiskPersistenceListener (ca. 55 sec for 760 entries). This is apperently due to the fact that for each entry you don't have to touch just one file but 1 + x files. Where x is the number of groups the entry is in. While I didn't fix the source of the problem (yet), I've made some changes to disk persistence (escpecially for groups) that will speed up things generally a little bit. I have now ca. 36 sec for the 760 entries. Should I add this as an attachment for this case, too, or should I open a new feature request?

Lars Torunski - [11/Sep/05 04:58 AM ] Christoph, please open an improvement issue for speed up the disk persistence and attach the source code.

Christoph Kutzinski - [26/Sep/05 06:53 AM ] I opened the improvement issue under : http://jira.opensymphony.com/browse/CACHE-197

Lars Torunski - [24/Jun/06 02:21 AM ] AbstractConcurrentReadCache fixed in CVS HEAD. All unit tests and the new unit test group are OK.