The order of result changed in oracle 10g after migration from 9i.

Question

i am using the below query to fetch data from two views.
SQL:
select distinct PVCID pvcid, PATH_ID pathId, fnn fnn, g.XDM_SERVICEID xdmServiceid, TC_SERVICETYPE tcServicetype, TC_PORTTYPE tcPorttype, TERMINATION_TYPE terminationType,
TC_PRODUCTCODES tcProductcodes, TERM_ID termId, TERM_NAME termName, TERMINATION_TYPE_ID terminationTypeId, A_EQUIPMENT_ID equipmentId,
A_EQUIPMENT_NAME equipmentName, A_EQUIPMENT_TYPE equipmentType, A_EQUIP_TYPE equipTypeId, AEND_BLOCK_TEST_REQ aendBlockTestReq, Z_EQUIPMENT_ID ZEquipmentId, Z_EQUIPMENT_NAME ZEquipmentName,
Z_EQUIPMENT_TYPE ZEquipType, Z_BLOCK_TEST_REQ blockTestReq, RADIUS_ENABLED radiusEnabled, D_PBR DPbr, U_PBR UPbr, BRAS_TYPE brasType from V_GET_PVCID g,V_TERM v where FNN='A0721122234' and g.xdm_serviceid=v.xdm_serviceid order by TC_SERVICETYPE
this gives two rows as output, but in different order for 9i and 10g.
This is affecting the working of the application. can any one help me out here.
thanks in advance

Answer 1

Since you specify only 1 column in the ORDER BY, ordering is done only against that column. Rows returned may appear differently appeared with respect to other columns.
If you need a specific order, include all the required columns in the ORDER BY clause.
(10g introduced GroupByHashAggregation so ordering isn't guaranteed to be the same as in 9i unless you correctly specify an ORDER BY clause. Even in 9i, Oracle didn't gurantee that they would always appear in the same order --- the appearance was only an after-effect of the manner in which Oracle implemented DISTINCT and GROUP BY operations in 9i).
Hemant K Chitale