Re: some validate_relation_kind() tidying

On 19.02.26 09:48, Ashutosh Bapat wrote:
> The || should be &&. The bug shows up as an initdb failure
> running bootstrap script ... 2026-02-19 14:06:43.411 IST [197482]
> FATAL: cannot open relation "pg_type"
> 2026-02-19 14:06:43.411 IST [197482] DETAIL: This operation is not
> supported for tables.
>
> I think this is more future-proof. If a relkind gets added and needs
> to be in this list, we will notice it from the error. I think we
> should avoid mentioning specific relkinds in the comment as well since
> that list will need to be updated as the set of relkinds changes. Just
> mentioning the criteria should be enough. I have slightly improved the
> comment in the attached 0003.
>
>> Secondly, the sequence.c one was probably copied from the table.c one,
>> but I think we can make the error message a bit more direct by just
>> saying "... is not a sequence" instead of "cannot open relation".
>>
>
> +1.
>
>> These are the two attached patches. This is just something I found
>> while working on something else nearby.
>
> Attached are your two patches + bug fix in 0002 + my suggestions in 0003.

Thanks, committed with your fixes. (I fine-tuned the comment in patch
0003 a bit further, since arguably you can "reference" a composite type
in a query.)