Hi guys, just to inform you that I'm done with the cleanup of the lookup operation.
The last commit [1] exposes the change smade. othing really serious, except a change in teh Store API : the previously named lookup( Sring[, Dn]) method is now named fetch(), to avoid any confusion with the exposed lookup method. This method is only used to fetch an entry when we have its UUID. This make some sense when the cursor store this entry in a varibale whch name is 'prefetched'. Everything else is about improvements and cleanups. The lookup performance is 5% faster than the number I posted this morning. [1] http://svn.apache.org/viewvc?rev=1421601&view=rev Log: o Renamed the internal Store.lookup(id) to fetch(id), to avoid any confusion with a standard lookup o Replaced the filterAttributeType() by a call to ServerEntryUtils.filterContent() in the TopFilter (SchemaManager) o Removed the lookup method from the ExceptionInterceptor : the subschemaSubentry processing is now done in the OperationalAttributeInterceptor o Moved the code that was grabbing the subentrySchema to the OperationalAttributeInterceptor lookup method o Added a dependency on core-shared in the apacheds-interceptors-operational module to b able to access the SchemaService class o Removed a lot of useless code in the SchemaService class : ,o need to filter the attributes there. o Removed the useless DefaultPartitionNexus.initializeSystemPartition() method o Removed the useless public Entry lookup( Dn dn ) throws LdapException method from CoreSession and the implementations o Ignored the testSearchFetchNonExistingAttributeOption test in core-integ SearchIT, we don't support options atm. o The BaseEntryFilteringCursor.next() method not anymore filter the attributes : it's already done in the SchemaInterceptor o Speedup the ServerEntryUtils.filterContents() by adding acontinue to avoid useless computation to take place -- Regards, Cordialement, Emmanuel Lécharny www.iktek.com
