John Gray <jgray@azuli.co.uk> writes:
> On Mon, 2002-05-27 at 16:30, Gregory Seidman wrote:
>> Which, while accurate, fails to answer the answer the question. A column is
>> not being added. A table is not being created. Rows are being updated and
>> the system is complaining that there are too many columns. Why?
> Don't know what the answer is, but it would be interesting to confirm
> whether PG thinks there are only 1600 columns.
Oh, wait, of course. The table has 1600 columns. But an UPDATE
operation needs to form an intermediate tuple containing all 1600
data columns plus the CTID column.
We should allow that, IMHO, which is another reason why heap_formtuple
is not the place to be applying the restriction. Looks like there is
some work to do here.
regards, tom lane