Re: amcheck support for BRIN indexes

On Sun, Jul 6, 2025 at 10:49 PM Álvaro Herrera <alvherre(at)kurilemu(dot)de> wrote:
>
> On 2025-Jul-06, Arseniy Mukhin wrote:
>
> > Sorry, forget to run a full test run with the new patch version. Some
> > tests were unhappy with the new unknown support function. Here the new
> > version with the fix.
>
> Hello, I think this patch is probably a good idea. I don't think it
> makes sense to introduce a bunch of code in 0003 only to rewrite it
> completely in 0005. I would ask that you re-split your WITHIN_RANGE
> (0004) to appear before the amcheck code, and then write the amcheck
> code using that new functionality.

Hi, Álvaro!

Thank you for looking into this.

OK, we can easily revert to the version with consistent function if
needed, so let's get rid of it.

>
> > /*
> > * Return a tuple descriptor used for on-disk storage of BRIN tuples.
> > */
> > -static TupleDesc
> > +TupleDesc
> > brtuple_disk_tupdesc(BrinDesc *brdesc)
>
> I think we should give this function a better name if it's going to be
> exported. How about brin_tuple_tupdesc? (in brin_tuple.h we
> seem to distinguish "brin tuples" which are the stored ones, from "brin
> mem tuples" which are the ones to be used in memory.)
>

'brin_tuple_tupdesc' sounds good to me. Done.

So here is a new version. 0001, 0002 - index structure check. 0003,
0004 - all heap indexed using WITHIN_RANGE approach.

Thank you!

Best regards,
Arseniy Mukhin