Recently we upgraded to TD15.10 from TD15.0 and facing a weird thing. We have a complex view which was giving a single record for a unique identifier column but now we are getting duplicate records for the same Unique identifier after upgrade. When we just run the select part of the view it is not giving any duplicates. This is reported to Teradata Support team ,they suggested to use "DIAGNOSTIC NOVIEWFOLD ON FOR SESSION" before querying the view. Executing this statement helped in giving correct results(no dupes). My question is why do we use this statement, how did it help here?
What you descibed sounds like a bug und you got an workaround.
To my knowledge this statements forces the optimizer into some different direction for complex views.
I guess you got an incident number and will it will be fixed at some point of time...