Andres Freund <andres@anarazel.de> writes:
> On 2023-07-06 11:16:26 -0400, Tom Lane wrote:
>> It does seem like we could do
>> uint64 startelem = SH_MAX_SIZE;
>> ...
>> Assert(startelem < SH_MAX_SIZE);
>> which'd make it a little clearer that the expectation is for
>> startelem to have changed value.
> I guess? I find it easier to understand all-bits-set in a coredump as
> too-large than SH_MAX_SIZE, but ...
What'd help even more is a comment:
/* We should have found an empty element */
Assert(startelem < SH_MAX_SIZE);
regards, tom lane