Re: BUG #19769: int8range() raises an error for an inclusive upper bound at BIGINT_MAX

PG Bug reporting form <noreply(at)postgresql(dot)org> writes:
> The following call fails, even though both bounds are valid bigint values
> and the requested range represents the singleton set containing the maximum
> bigint value (9223372036854775807).

The same thing happens with int4range:

regression=# select '[2147483647,2147483647]'::int4range;
ERROR: integer out of range

and with other range types.

> The PostgreSQL range documentation describes int8range as a discrete range
> type whose canonical form is lower-inclusive and upper-exclusive ([)). Based
> on this behavior, I expected the singleton range to be accepted and
> represented equivalently rather than raising an out-of-range error.

I think your expectation is mistaken. The documentation is quite
clear that we normalize [x,y] to [x,y+1).

regards, tom lane