On Mon, 24 Apr 2006, Martijn van Oosterhout wrote:
> Perhaps someone could check if changing the test explicitly check
> against NULL:
>
>> ((attnum) > (int) (tup)->t_data->t_natts) ? \
>> ( \
>> (((isnull) != NULL)? (*(isnull) = true) : (dummyret)NULL), \
>> (Datum)NULL \
>
> removes the warning. It seems silly for the GCC people to add warnings
> for this kind of stuff without a simple way to bypass it...
Yes, this coding removes the warning.
Kris Jurka