You’re absolutely right, the GTT is in fact completely unnecessary and you could reference the itinerary_ext table directly in the creation statement for exp_itinerary. Looking back, the table names I used could’ve been less confusing as well.
Of course, the GTT does make the code a little more readable ( well, I think so, anyway). There’s also my own personal dislike of using too much dynamic SQL, especially DDL statements which are always so much fun to debug.
When all’s said and done however, I just have to hold my hands up and say, erm, I missed that one. Well spotted and thanks for pointing it out.

Mike

When you do the transform and output the results via datapump, why do you use a GTT, instead of basing the table directly on the query? If you skipped the GTT the data would no longer need to be duplicated in the PGA, the query results could go straight out to disk.