Peter Eisentraut <peter_e@gmx.net> writes:
> Tom Lane wrote:
>> I just noticed that when the BY option was added to plpgsql FOR
>> loops, no real error checking was done. If you specify a zero step
>> value, you'll have an infinite loop. If you specify a negative
>> value, the loop variable will increment in the "wrong direction"
>> until integer overflow occurs. Neither of these behaviors seem
>> desirable in the least.
> That seems to be fairly normal proramming language behavior.
Well, it's about what I'd expect from C or something at a similar level
of (non) abstraction. But I dislike the idea that plpgsql should have
behavior as machine-dependent as that the number of iterations will
depend on the value of INT_MIN. Also, at the SQL level our usual policy
is to throw errors for obvious programmer mistakes, and it's hard to
argue that a zero or negative step isn't a programmer mistake. Had we
defined the stepping behavior differently (ie, make "BY -1" work like
REVERSE) then there would be some sanity in allowing negative steps,
but I don't see the sanity in it given the implemented behavior.
regards, tom lane
