On 27 Nov 2015, at 6:30pm, R Smith <rsmith at rsweb.co.za> wrote: > Let me explain better, let's assume the query contains MAX(x)... The result > will be from whichever row contains the max, and if the x was not in the > aggregate function, it will be from some row (which may or may not be the one > with the max value). > In both cases, you have NO chance of knowing which row that would be until > the query finishes, so ALWAYS assume an arbitrary row (or > Always-non-deterministic, to use the OP's words).
You missed the general principle: there may be any number of rows with the max value. For example, 8,1,6,3,8,4,8,7,6,8 . So even if you're using MAX you still can't predict which row is supplying the other values. So as you (R Smith) says, always assume an arbitrary row. Simon.
