On 21.03.23 00:51, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
>> On 2023-03-20 10:37:36 -0400, Tom Lane wrote:
>>> I agree that attinhcount could be narrowed, but I have some concern
>>> about attstattarget. IIRC, the limit on attstattarget was once 1000
>>> and then we raised it to 10000. Is it inconceivable that we might
>>> want to raise it to 100000 someday?
>
>> Hard to believe that'd happen in a minor version - and I don't think there'd
>> an issue with widening it again in a major version?
>
> True. However, I think Tomas' idea of making these columns nullable
> is even better than narrowing them.
The context of my message was to do the proposed change for PG16 to buy
back a few bytes that are being added by another feature, and then
consider doing a larger detangling of pg_attribute and tuple descriptors
in PG17, which might well involve taking the attstattarget out of the
hot path. Making attstattarget nullable (i.e., not part of the fixed
part of pg_attribute) would require fairly significant surgery, so I
think it would be better done as part of a more comprehensive change
that would allow the same treatment for other columns as well.