

For example a query which takes less than 1 second on average using an inner join will take consistently around 150 seconds when using a left join. We also have an issue with slow left joins. > getting the author to reverse the tables in ajoin(for h2's sake) might be > then run the problematic query manually (that is, not using > to re-write the query, and if you are using a tool such as Hibernate > believe, and currently it's not a very high priority for me. > I thought I would ask since it's been a while would you be able to give a > I thank you for your time, and a providing a truly *outstanding* piece > one, and helping to keep it from getting lost in the shuffle. > I took the liberty of creating a issue for this since I didn't notice I will investigate some of the workarounds you suggest. PUBLIC.OM_ATTRVAL_ID: OMATTRIBID = 25 AND OBJVALUE >= Sorry I did not provide more info sooner, please let me know if more

The hardware is a 1Ghz CPU 1GB RAM (50% free). (and other variations on Xmx and objectCacheSize) Jar h2-1.2.143.jar org.h2.tools.Server -webAllowOthers -tcpAllowOthers Java -Xmx384M -Dh2.returnLobObjects=false -Dh2.objectCacheSize=50000. PUBLIC.OM_ATTRVAL_ID: OMATTRIBID = 83 AND OBJVALUE = */ /* scanCount: 3130722 */ LEFT OUTER JOIN PUBLIC.OM OWN /* PUBLIC.OM B /* PUBLIC.OM_ATTRVAL_ID: OMATTRIBID = 25 AND OBJVALUE >= 'SELECT A.OBJID FROM PUBLIC.OM A /* PUBLIC.OM_ATTRVAL_ID: OMATTRIBID =Ģ8 */ /* WHERE A.OMATTRIBID = 28 */ /* scanCount: 49695 */ INNER JOIN Omid bigint DEFAULT nextval('om_omid_seq') NOT NULLĬREATE INDEX om_attrval_id ON om (omattribid, objvalue) ĬREATE INDEX om_objid_omattribid ON om (objid, omattribid) ĬREATE INDEX om_objvalue_idx ON om (objvalue) ĬREATE INDEX om_omschemaid_idx ON om (omattribid)
