3 ms·
Yes, yes, yes! Having the "late" commit (session 1) fail in this case is exactly what I would expect as "correct" behavior. However, I (now) see what MJB mean
by Roboprog 13y ago
Yes, yes, yes! Having the "late" commit (session 1) fail in this case is exactly what I would expect as "correct" behavior.
However, I (now) see what MJB meant about needing an explicit (table/row) lock so that if you can start such a sequence, you should be able to complete it.