-
Bug
-
Resolution: Fixed
-
Major
-
4.2.0
-
None
-
4.2 M9
Few issues here :
- in AclCacheManager#_computeBitSetAndPutInCache, we do values.put(__CACHE_BISET_KEY, null); after writeLock.tryLock(), and if a second user asked its acl, it removes all BitSet entries before putting its own, that is to say not more than one BitSet is present in the cache.
- but this was hidden by filterCache and queryResultCache and in practice, the BitSet cache was never used => we should find a way to not use them when there is an AclQuery