I use a loop to determine complex calculatoins for a matching record. inside the loop the record (matched by primary key) is updated. Unfortunately a complex update has been realized as too complex (using the loop is up to 10 times quicker).
Now here is the big question : What is preferred to be used for locating the record? Shall I use the primary key or the table's ROWID?
Afaik ROWID is the quicker solution, but is safe to use ROWID inside the loop?
Is there any dependency to the isolation level? Or does the rowid not change when I loop/update it ?