Re: pg_column_toast_chunk_id: a function to get a chunk ID of a TOASTed value

On Tue, 30 Jan 2024 14:57:20 +0800
jian he <jian(dot)universality(at)gmail(dot)com> wrote:

> On Tue, Jan 30, 2024 at 1:56 PM Yugo NAGATA <nagata(at)sraoss(dot)co(dot)jp> wrote:
> >
> > I attached the correct one, v4.
> >
>
> +-- Test pg_column_toast_chunk_id:
> +-- Check if the returned chunk_id exists in the TOAST table
> +CREATE TABLE test_chunk_id (v1 text, v2 text);
> +INSERT INTO test_chunk_id VALUES (
> + repeat('0123456789', 10), -- v1: small enough not to be TOASTed
> + repeat('0123456789', 100000)); -- v2: large enough to be TOASTed
>
> select pg_size_pretty(100000::bigint);
> return 98kb.
>
> I think this is just too much, maybe I didn't consider the
> implications of compression.
> Anyway, I refactored the tests, making the toast value size be small.

Actually the data is compressed and the size is much smaller,
but I agree with you it is better not to generate large data unnecessarily.
I rewrote the test to disallow compression in the toast data using
"ALTER TABLE ... SET STORAGE EXTERNAL". In this case, any text larger
than 2k will be TOASTed on disk without compression, and it makes the
test simple, not required to use string_agg.
>
> I aslo refactor the doc.
> pg_column_toast_chunk_id entry will be right after pg_column_compression entry.
> You can check the screenshot.

I found the document order was not changed between my patch and yours.
In both, pg_column_toast_chunk_id entry is right after
pg_column_compression.

Here is a updated patch, v6.

Regards,
Yugo Nagata

--
Yugo NAGATA <nagata(at)sraoss(dot)co(dot)jp>