G
George Potemkin
Guest
It would be interesting to know why there is a difference in transaction undoes triggered by a "stopper" and by a "terminator". Most likely the UNDO variables are treated as the fields of some pseudo tables (one pseudo table per procedure) in some pseudo database. But in my test program all variables are defined as NO-UNDO and 'strings lbi*' shown only the contents of the customer.name fields. I used the simple transactions in sports database: REPEAT: FOR EACH sports.customer EXCLUSIVE-LOCK: ASSIGN sports.customer.name = "x" + sports.customer.name. ASSIGN sports.customer.name = SUBSTRING(sports.customer.name, 2). END. IF NOW GE vEndTime THEN LEAVE. END.
Continue reading...
Continue reading...